modelId
stringlengths 5
139
| author
stringlengths 2
42
| last_modified
timestamp[us, tz=UTC]date 2020-02-15 11:33:14
2025-09-12 12:31:00
| downloads
int64 0
223M
| likes
int64 0
11.7k
| library_name
stringclasses 555
values | tags
listlengths 1
4.05k
| pipeline_tag
stringclasses 55
values | createdAt
timestamp[us, tz=UTC]date 2022-03-02 23:29:04
2025-09-12 12:28:53
| card
stringlengths 11
1.01M
|
|---|---|---|---|---|---|---|---|---|---|
raghav-gaggar/stable-diffusion-thumbs-up
|
raghav-gaggar
| 2023-07-06T23:57:55Z | 6 | 0 |
diffusers
|
[
"diffusers",
"safetensors",
"text-to-image",
"stable-diffusion",
"autotrain_compatible",
"endpoints_compatible",
"diffusers:StableDiffusionPipeline",
"region:us"
] |
text-to-image
| 2023-07-06T02:25:26Z |
---
tags:
- text-to-image
- stable-diffusion
---
Stable Diffusion model, fine-tuned for generating images of people with their thumbs up.
How to use it:
```py
from diffusers import StableDiffusionPipeline
import torch
from torchmetrics.functional.multimodal import clip_score
from functools import partial
model_ckpt = "raghav-gaggar/stable-diffusion-thumbs-up"
sd_pipeline = StableDiffusionPipeline.from_pretrained(model_ckpt, torch_dtype=torch.float16).to("cuda")
prompts = [
"thumbs up",
"thumbs up",
"thumbs up",
"thumbs up",
"thumbs up",
"thumbs up",
"thumbs up",
"thumbs up",
"thumbs up",
"thumbs up",
]
images = sd_pipeline(prompts, num_images_per_prompt=1, output_type="numpy").images
print(images.shape)
clip_score_fn = partial(clip_score, model_name_or_path="openai/clip-vit-base-patch16")
def calculate_clip_score(images, prompts):
images_int = (images * 255).astype("uint8")
clip_score = clip_score_fn(torch.from_numpy(images_int).permute(0, 3, 1, 2), prompts).detach()
return round(float(clip_score), 4)
sd_clip_score = calculate_clip_score(images, prompts)
print(f"CLIP score: {sd_clip_score}")
```
Sample pictures of this concept:
|
Raizel123/Feliclora
|
Raizel123
| 2023-07-06T23:40:44Z | 0 | 0 | null |
[
"license:creativeml-openrail-m",
"region:us"
] | null | 2023-07-06T23:32:38Z |
---
license: creativeml-openrail-m
---
|
AbduBot/dqn-SpaceInvadersNoFrameskip-v4
|
AbduBot
| 2023-07-06T23:27:22Z | 2 | 0 |
stable-baselines3
|
[
"stable-baselines3",
"SpaceInvadersNoFrameskip-v4",
"deep-reinforcement-learning",
"reinforcement-learning",
"model-index",
"region:us"
] |
reinforcement-learning
| 2023-07-06T23:26:45Z |
---
library_name: stable-baselines3
tags:
- SpaceInvadersNoFrameskip-v4
- deep-reinforcement-learning
- reinforcement-learning
- stable-baselines3
model-index:
- name: DQN
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: SpaceInvadersNoFrameskip-v4
type: SpaceInvadersNoFrameskip-v4
metrics:
- type: mean_reward
value: 653.50 +/- 202.04
name: mean_reward
verified: false
---
# **DQN** Agent playing **SpaceInvadersNoFrameskip-v4**
This is a trained model of a **DQN** agent playing **SpaceInvadersNoFrameskip-v4**
using the [stable-baselines3 library](https://github.com/DLR-RM/stable-baselines3)
and the [RL Zoo](https://github.com/DLR-RM/rl-baselines3-zoo).
The RL Zoo is a training framework for Stable Baselines3
reinforcement learning agents,
with hyperparameter optimization and pre-trained agents included.
## Usage (with SB3 RL Zoo)
RL Zoo: https://github.com/DLR-RM/rl-baselines3-zoo<br/>
SB3: https://github.com/DLR-RM/stable-baselines3<br/>
SB3 Contrib: https://github.com/Stable-Baselines-Team/stable-baselines3-contrib
Install the RL Zoo (with SB3 and SB3-Contrib):
```bash
pip install rl_zoo3
```
```
# Download model and save it into the logs/ folder
python -m rl_zoo3.load_from_hub --algo dqn --env SpaceInvadersNoFrameskip-v4 -orga AbduBot -f logs/
python -m rl_zoo3.enjoy --algo dqn --env SpaceInvadersNoFrameskip-v4 -f logs/
```
If you installed the RL Zoo3 via pip (`pip install rl_zoo3`), from anywhere you can do:
```
python -m rl_zoo3.load_from_hub --algo dqn --env SpaceInvadersNoFrameskip-v4 -orga AbduBot -f logs/
python -m rl_zoo3.enjoy --algo dqn --env SpaceInvadersNoFrameskip-v4 -f logs/
```
## Training (with the RL Zoo)
```
python -m rl_zoo3.train --algo dqn --env SpaceInvadersNoFrameskip-v4 -f logs/
# Upload the model and generate video (when possible)
python -m rl_zoo3.push_to_hub --algo dqn --env SpaceInvadersNoFrameskip-v4 -f logs/ -orga AbduBot
```
## Hyperparameters
```python
OrderedDict([('batch_size', 32),
('buffer_size', 100000),
('env_wrapper',
['stable_baselines3.common.atari_wrappers.AtariWrapper']),
('exploration_final_eps', 0.01),
('exploration_fraction', 0.1),
('frame_stack', 4),
('gradient_steps', 1),
('learning_rate', 0.0001),
('learning_starts', 100000),
('n_timesteps', 1000000.0),
('optimize_memory_usage', False),
('policy', 'CnnPolicy'),
('target_update_interval', 1000),
('train_freq', 4),
('normalize', False)])
```
# Environment Arguments
```python
{'render_mode': 'rgb_array'}
```
|
YakovElm/Hyperledger_15_BERT_More_Properties
|
YakovElm
| 2023-07-06T23:25:07Z | 61 | 0 |
transformers
|
[
"transformers",
"tf",
"bert",
"text-classification",
"generated_from_keras_callback",
"license:apache-2.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2023-07-06T23:24:32Z |
---
license: apache-2.0
tags:
- generated_from_keras_callback
model-index:
- name: Hyperledger_15_BERT_More_Properties
results: []
---
<!-- This model card has been generated automatically according to the information Keras had access to. You should
probably proofread and complete it, then remove this comment. -->
# Hyperledger_15_BERT_More_Properties
This model is a fine-tuned version of [bert-base-uncased](https://huggingface.co/bert-base-uncased) on an unknown dataset.
It achieves the following results on the evaluation set:
- Train Loss: 0.3137
- Train Accuracy: 0.9035
- Validation Loss: 0.3679
- Validation Accuracy: 0.8807
- Epoch: 2
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- optimizer: {'name': 'Adam', 'weight_decay': None, 'clipnorm': 1.0, 'global_clipnorm': None, 'clipvalue': None, 'use_ema': False, 'ema_momentum': 0.99, 'ema_overwrite_frequency': None, 'jit_compile': False, 'is_legacy_optimizer': False, 'learning_rate': 3e-05, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-08, 'amsgrad': False}
- training_precision: float32
### Training results
| Train Loss | Train Accuracy | Validation Loss | Validation Accuracy | Epoch |
|:----------:|:--------------:|:---------------:|:-------------------:|:-----:|
| 0.3228 | 0.8993 | 0.3659 | 0.8807 | 0 |
| 0.3217 | 0.9035 | 0.3648 | 0.8807 | 1 |
| 0.3137 | 0.9035 | 0.3679 | 0.8807 | 2 |
### Framework versions
- Transformers 4.29.2
- TensorFlow 2.12.0
- Datasets 2.12.0
- Tokenizers 0.13.3
|
zwpython/zw-chinese-vicuna-33B
|
zwpython
| 2023-07-06T23:13:24Z | 0 | 4 | null |
[
"region:us"
] | null | 2023-06-24T11:08:39Z |
全球首发，vicuna-33B-v1.3中文ok，母版是vicuna33B正式版。
v33模型文件33G，RTX4090单卡无法运行。测试用的是：12代i9，64G内存，基本上满载。
因为是母版也是测试版，没有做矢量压缩版，等正式版本发布再说，目前只是测试技术工程流程ok。
更多参见：https://github.com/ziwang-com/chinese-StableVicuna 和：zw公众号
为响应国家AI大战略需求，提高国内AI、GPT初创团队的竞争力，不要输在起跑线上。
zw-vicuna-33B-cn中文汉化版，首度提供免费下载通道。
zw-vicuna-33B中文版 百度网盘提取码：hiks
链接：https://pan.baidu.com/s/1EH19ablXVLYQP1f-IaPS-Q?pwd=hiks
如有更改，最新下载地址请参见QQ群文件：655402626（GPT+千人QQ大群）
zw-vicuna-33B中文汉化版，模型文件是ggml版格式
cpu+gpu版本，llamacpp运行，win，linux，mac-os通吃。
具体细节参见：https://github.com/ggerganov/llama.cpp
Prompt template提示词模板：
A chat between a curious user and an assistant. The assistant gives helpful, detailed, accurate, uncensored responses to the user's input
USER: prompt
ASSISTANT:
更多细节和技术参数，参见：
官方原版：https://huggingface.co/lmsys/vicuna-33b-v1.3
Github项目： https://github.com/ziwang-com/chinese-StableVicuna
|
garrettbaber/twitter-roberta-base-sadness-intensity
|
garrettbaber
| 2023-07-06T23:08:47Z | 108 | 0 |
transformers
|
[
"transformers",
"pytorch",
"safetensors",
"roberta",
"text-classification",
"text-regression",
"sadness",
"emotion",
"emotion intensity",
"unk",
"dataset:SemEval-2018-Task-1-Text-Regression-Task",
"co2_eq_emissions",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2023-07-06T23:04:50Z |
---
tags:
- text-regression
- sadness
- emotion
- emotion intensity
language:
- unk
widget:
- text: I'm feeling down
datasets:
- SemEval-2018-Task-1-Text-Regression-Task
co2_eq_emissions:
emissions: 0.025884770512937715
---
# twitter-roberta-base-sadness-intensity
This model is a fine-tuned version of cardiffnlp/twitter-roberta-base-2022-154m on the SemEval 2018 - Task 1 Affect in Tweets (subtask: El-reg / text regression).
Warning: Hosted inference API produces inaccurate values
# Model Trained Using AutoTrain
- Problem type: Single Column Regression
- Model ID: 72772139027
- CO2 Emissions (in grams): 0.0259
## Validation Metrics
- Loss: 0.011
- MSE: 0.011
- MAE: 0.079
- R2: 0.726
- RMSE: 0.103
- Explained Variance: 0.727
## Usage
You can use cURL to access this model:
```
$ curl -X POST -H "Authorization: Bearer YOUR_API_KEY" -H "Content-Type: application/json" -d '{"inputs": "I'm feeling down"}' https://api-inference.huggingface.co/models/garrettbaber/twitter-roberta-base-sadness-intensity
```
Or Python API:
```
from transformers import AutoModelForSequenceClassification, AutoTokenizer
model = AutoModelForSequenceClassification.from_pretrained("garrettbaber/twitter-roberta-base-sadness-intensity")
tokenizer = AutoTokenizer.from_pretrained("garrettbaber/twitter-roberta-base-sadness-intensity")
inputs = tokenizer("I'm feeling down", return_tensors="pt")
outputs = model(**inputs)
```
|
aroot/eng-mya-simcse_central_usrb
|
aroot
| 2023-07-06T23:03:39Z | 103 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"mbart",
"text2text-generation",
"translation",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
translation
| 2023-07-06T22:46:54Z |
---
tags:
- translation
- generated_from_trainer
metrics:
- bleu
model-index:
- name: eng-mya-simcse_central_usrb
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# eng-mya-simcse_central_usrb
This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 1.8932
- Bleu: 4.0471
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3
|
yongsun-yoon/minilmv2-bertscore-distilled
|
yongsun-yoon
| 2023-07-06T23:00:56Z | 140 | 0 |
transformers
|
[
"transformers",
"pytorch",
"roberta",
"feature-extraction",
"text-embeddings-inference",
"endpoints_compatible",
"region:us"
] |
feature-extraction
| 2023-06-30T22:04:37Z |
This is a distilled BERTScore model.
Please read [this post](https://medium.com/@yongsun.yoon/bertscore-knowledge-distillation-42721b3508e2) for details.
```python
from bert_score import BERTScorer
texts1 = ['This is a text.']
texts2 = ['This is another text.']
scorer = BERTScorer(model_type='yongsun-yoon/minilmv2-bertscore-distilled', num_layers=6)
P, R, F = scorer.score(texts1, texts2)
```
|
garrettbaber/twitter-roberta-base-joy-intensity
|
garrettbaber
| 2023-07-06T22:58:25Z | 107 | 0 |
transformers
|
[
"transformers",
"pytorch",
"safetensors",
"deberta",
"text-classification",
"text-regression",
"joy",
"emotion",
"emotion intensity",
"en",
"dataset:SemEval-2018-Task-1-Text-Regression-Task",
"co2_eq_emissions",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2023-07-06T22:47:30Z |
---
tags:
- text-regression
- joy
- emotion
- emotion intensity
language:
- en
widget:
- text: I am elated!
datasets:
- SemEval-2018-Task-1-Text-Regression-Task
co2_eq_emissions:
emissions: 0.03988347977318191
---
# twitter-roberta-base-joy-intensity
This model is a fine-tuned version of cardiffnlp/twitter-roberta-base-2022-154m on the SemEval 2018 - Task 1 Affect in Tweets (subtask: El-reg / text regression).
Warning: Hosted inference API produces inaccurate values
# Model Trained Using AutoTrain
- Problem type: Single Column Regression
- Model ID: 72771139026
- CO2 Emissions (in grams): 0.0399
## Validation Metrics
- Loss: 0.013
- MSE: 0.013
- MAE: 0.088
- R2: 0.707
- RMSE: 0.116
- Explained Variance: 0.709
## Usage
You can use cURL to access this model:
```
$ curl -X POST -H "Authorization: Bearer YOUR_API_KEY" -H "Content-Type: application/json" -d '{"inputs": "I am elated!"}' https://api-inference.huggingface.co/models/garrettbaber/twitter-roberta-base-joy-intensity
```
Or Python API:
```
from transformers import AutoModelForSequenceClassification, AutoTokenizer
model = AutoModelForSequenceClassification.from_pretrained("garrettbaber/twitter-roberta-base-joy-intensity")
tokenizer = AutoTokenizer.from_pretrained("garrettbaber/twitter-roberta-base-joy-intensity")
inputs = tokenizer("I am elated!", return_tensors="pt")
outputs = model(**inputs)
```
|
osunlp/PubMedBERT_Replica
|
osunlp
| 2023-07-06T22:54:34Z | 125 | 1 |
transformers
|
[
"transformers",
"pytorch",
"bert",
"fill-mask",
"arxiv:2306.17649",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
fill-mask
| 2023-07-06T17:39:50Z |
We replicate the PubMedBERT model using the same data, hardware and code as our new [BioVocabBERT](https://huggingface.co/osunlp/BioVocabBERT) model to ensure their comparion is fair.
Details about our pre-training procedure and downstream results can be found in our [BioNLP @ ACL 2023 paper](https://arxiv.org/abs/2306.17649).
---
license: apache-2.0
---
|
YakovElm/Hyperledger_10_BERT_More_Properties
|
YakovElm
| 2023-07-06T22:43:18Z | 61 | 0 |
transformers
|
[
"transformers",
"tf",
"bert",
"text-classification",
"generated_from_keras_callback",
"license:apache-2.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2023-07-06T22:42:43Z |
---
license: apache-2.0
tags:
- generated_from_keras_callback
model-index:
- name: Hyperledger_10_BERT_More_Properties
results: []
---
<!-- This model card has been generated automatically according to the information Keras had access to. You should
probably proofread and complete it, then remove this comment. -->
# Hyperledger_10_BERT_More_Properties
This model is a fine-tuned version of [bert-base-uncased](https://huggingface.co/bert-base-uncased) on an unknown dataset.
It achieves the following results on the evaluation set:
- Train Loss: 0.3569
- Train Accuracy: 0.8838
- Validation Loss: 0.3936
- Validation Accuracy: 0.8600
- Epoch: 2
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- optimizer: {'name': 'Adam', 'weight_decay': None, 'clipnorm': 1.0, 'global_clipnorm': None, 'clipvalue': None, 'use_ema': False, 'ema_momentum': 0.99, 'ema_overwrite_frequency': None, 'jit_compile': False, 'is_legacy_optimizer': False, 'learning_rate': 3e-05, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-08, 'amsgrad': False}
- training_precision: float32
### Training results
| Train Loss | Train Accuracy | Validation Loss | Validation Accuracy | Epoch |
|:----------:|:--------------:|:---------------:|:-------------------:|:-----:|
| 0.3701 | 0.8762 | 0.4061 | 0.8600 | 0 |
| 0.3614 | 0.8838 | 0.4039 | 0.8600 | 1 |
| 0.3569 | 0.8838 | 0.3936 | 0.8600 | 2 |
### Framework versions
- Transformers 4.29.2
- TensorFlow 2.12.0
- Datasets 2.12.0
- Tokenizers 0.13.3
|
aroot/eng-guj-simcse_central_usrb
|
aroot
| 2023-07-06T22:26:28Z | 102 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"mbart",
"text2text-generation",
"translation",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
translation
| 2023-07-06T22:03:36Z |
---
tags:
- translation
- generated_from_trainer
metrics:
- bleu
model-index:
- name: eng-guj-simcse_central_usrb
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# eng-guj-simcse_central_usrb
This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 3.2880
- Bleu: 2.6776
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3
|
S1X3L4/ppo-Huggy
|
S1X3L4
| 2023-07-06T22:24:21Z | 12 | 0 |
ml-agents
|
[
"ml-agents",
"tensorboard",
"onnx",
"Huggy",
"deep-reinforcement-learning",
"reinforcement-learning",
"ML-Agents-Huggy",
"region:us"
] |
reinforcement-learning
| 2023-07-06T22:24:16Z |
---
library_name: ml-agents
tags:
- Huggy
- deep-reinforcement-learning
- reinforcement-learning
- ML-Agents-Huggy
---
# **ppo** Agent playing **Huggy**
This is a trained model of a **ppo** agent playing **Huggy**
using the [Unity ML-Agents Library](https://github.com/Unity-Technologies/ml-agents).
## Usage (with ML-Agents)
The Documentation: https://unity-technologies.github.io/ml-agents/ML-Agents-Toolkit-Documentation/
We wrote a complete tutorial to learn to train your first agent using ML-Agents and publish it to the Hub:
- A *short tutorial* where you teach Huggy the Dog 🐶 to fetch the stick and then play with him directly in your
browser: https://huggingface.co/learn/deep-rl-course/unitbonus1/introduction
- A *longer tutorial* to understand how works ML-Agents:
https://huggingface.co/learn/deep-rl-course/unit5/introduction
### Resume the training
```bash
mlagents-learn <your_configuration_file_path.yaml> --run-id=<run_id> --resume
```
### Watch your Agent play
You can watch your agent **playing directly in your browser**
1. If the environment is part of ML-Agents official environments, go to https://huggingface.co/unity
2. Step 1: Find your model_id: S1X3L4/ppo-Huggy
3. Step 2: Select your *.nn /*.onnx file
4. Click on Watch the agent play 👀
|
cactusfriend/nightmare-promptgen-XL
|
cactusfriend
| 2023-07-06T22:16:34Z | 127 | 5 |
transformers
|
[
"transformers",
"pytorch",
"safetensors",
"gpt_neo",
"text-generation",
"license:openrail",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-generation
| 2023-06-26T14:07:40Z |
---
license: openrail
pipeline_tag: text-generation
library_name: transformers
widget:
- text: "a photograph of"
example_title: "photo"
- text: "a bizarre cg render"
example_title: "render"
- text: "the spaghetti"
example_title: "meal?"
- text: "a (detailed+ intricate)+ picture"
example_title: "weights"
- text: "photograph of various"
example_title: "variety"
inference:
parameters:
temperature: 2.6
max_new_tokens: 250
---
Experimental 'XL' version of [Nightmare InvokeAI Prompts](https://huggingface.co/cactusfriend/nightmare-invokeai-prompts). Very early version and may be deleted.
|
YakovElm/Hyperledger_5_BERT_More_Properties
|
YakovElm
| 2023-07-06T22:00:56Z | 61 | 0 |
transformers
|
[
"transformers",
"tf",
"bert",
"text-classification",
"generated_from_keras_callback",
"license:apache-2.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2023-07-06T22:00:21Z |
---
license: apache-2.0
tags:
- generated_from_keras_callback
model-index:
- name: Hyperledger_5_BERT_More_Properties
results: []
---
<!-- This model card has been generated automatically according to the information Keras had access to. You should
probably proofread and complete it, then remove this comment. -->
# Hyperledger_5_BERT_More_Properties
This model is a fine-tuned version of [bert-base-uncased](https://huggingface.co/bert-base-uncased) on an unknown dataset.
It achieves the following results on the evaluation set:
- Train Loss: 0.4169
- Train Accuracy: 0.8547
- Validation Loss: 0.4397
- Validation Accuracy: 0.8361
- Epoch: 2
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- optimizer: {'name': 'Adam', 'weight_decay': None, 'clipnorm': 1.0, 'global_clipnorm': None, 'clipvalue': None, 'use_ema': False, 'ema_momentum': 0.99, 'ema_overwrite_frequency': None, 'jit_compile': False, 'is_legacy_optimizer': False, 'learning_rate': 3e-05, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-08, 'amsgrad': False}
- training_precision: float32
### Training results
| Train Loss | Train Accuracy | Validation Loss | Validation Accuracy | Epoch |
|:----------:|:--------------:|:---------------:|:-------------------:|:-----:|
| 0.4271 | 0.8495 | 0.4528 | 0.8361 | 0 |
| 0.4171 | 0.8547 | 0.4638 | 0.8361 | 1 |
| 0.4169 | 0.8547 | 0.4397 | 0.8361 | 2 |
### Framework versions
- Transformers 4.29.2
- TensorFlow 2.12.0
- Datasets 2.12.0
- Tokenizers 0.13.3
|
juancopi81/lmd-8bars-2048-epochs20_v3
|
juancopi81
| 2023-07-06T21:58:13Z | 136 | 0 |
transformers
|
[
"transformers",
"pytorch",
"safetensors",
"gpt2",
"text-generation",
"generated_from_trainer",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] |
text-generation
| 2023-07-05T23:21:51Z |
---
tags:
- generated_from_trainer
model-index:
- name: lmd-8bars-2048-epochs20_v3
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# lmd-8bars-2048-epochs20_v3
This model is a fine-tuned version of [juancopi81/lmd-8bars-2048-epochs20_v2](https://huggingface.co/juancopi81/lmd-8bars-2048-epochs20_v2) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.9563
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0005
- train_batch_size: 8
- eval_batch_size: 4
- seed: 1
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- num_epochs: 1
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 1.0054 | 0.5 | 4994 | 0.9774 |
| 0.9784 | 1.0 | 9988 | 0.9563 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu118
- Datasets 2.13.1
- Tokenizers 0.13.3
|
aroot/eng-fra-simcse_central_usrb
|
aroot
| 2023-07-06T21:50:10Z | 103 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"mbart",
"text2text-generation",
"translation",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
translation
| 2023-07-06T21:31:40Z |
---
tags:
- translation
- generated_from_trainer
metrics:
- bleu
model-index:
- name: eng-fra-simcse_central_usrb
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# eng-fra-simcse_central_usrb
This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 1.1496
- Bleu: 31.8498
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3
|
AustinCarthy/Benign10MGPT2_suffix_100KP_BFall_fromB_90K_topP_0.75_ratio2.63
|
AustinCarthy
| 2023-07-06T21:45:14Z | 0 | 0 | null |
[
"tensorboard",
"generated_from_trainer",
"license:apache-2.0",
"region:us"
] | null | 2023-07-06T19:32:46Z |
---
license: apache-2.0
tags:
- generated_from_trainer
metrics:
- accuracy
- f1
- precision
- recall
model-index:
- name: Benign10MGPT2_suffix_100KP_BFall_fromB_90K_topP_0.75_ratio2.63
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# Benign10MGPT2_suffix_100KP_BFall_fromB_90K_topP_0.75_ratio2.63
This model is a fine-tuned version of [bert-base-uncased](https://huggingface.co/bert-base-uncased) on the Train benign: Fall,Test Benign: Fall, Train phish: Fall, Test phish: Fall, generated url dataset: generated_phish_Benign10MGPT2_using_benign_95K_top_p_0.75suffix dataset.
It achieves the following results on the evaluation set:
- Loss: 0.1028
- Accuracy: 0.9801
- F1: 0.8205
- Precision: 0.7196
- Recall: 0.9542
- Roc Auc Score: 0.9678
- Tpr At Fpr 0.01: 0.787
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5.0
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 | Precision | Recall | Roc Auc Score | Tpr At Fpr 0.01 |
|:-------------:|:-----:|:------:|:---------------:|:--------:|:------:|:---------:|:------:|:-------------:|:---------------:|
| 0.1253 | 1.0 | 21554 | 0.0991 | 0.9691 | 0.7343 | 0.6212 | 0.8978 | 0.9352 | 0.7494 |
| 0.0986 | 2.0 | 43108 | 0.0631 | 0.9831 | 0.8441 | 0.7519 | 0.962 | 0.9731 | 0.7858 |
| 0.0817 | 3.0 | 64662 | 0.0598 | 0.9838 | 0.8468 | 0.7704 | 0.94 | 0.9630 | 0.7012 |
| 0.06 | 4.0 | 86216 | 0.0799 | 0.9822 | 0.8367 | 0.7414 | 0.9602 | 0.9717 | 0.8236 |
| 0.0405 | 5.0 | 107770 | 0.1028 | 0.9801 | 0.8205 | 0.7196 | 0.9542 | 0.9678 | 0.787 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.0+cu118
- Datasets 2.13.1
- Tokenizers 0.13.3
|
TheBloke/Selfee-7B-SuperHOT-8K-GGML
|
TheBloke
| 2023-07-06T21:43:47Z | 0 | 1 | null |
[
"license:other",
"region:us"
] | null | 2023-07-06T17:38:13Z |
---
inference: false
license: other
---
<!-- header start -->
<div style="width: 100%;">
<img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
</div>
<div style="display: flex; justify-content: space-between; width: 100%;">
<div style="display: flex; flex-direction: column; align-items: flex-start;">
<p><a href="https://discord.gg/theblokeai">Chat & support: my new Discord server</a></p>
</div>
<div style="display: flex; flex-direction: column; align-items: flex-end;">
<p><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p>
</div>
</div>
<!-- header end -->
# Kaist AI's Selfee 7B GGML
These files are GGML format model files for [Kaist AI's Selfee 7B](https://huggingface.co/TheBloke/selfee-7B-fp16).
These are SuperHOT GGMLs with an increased context length. SuperHOT is a new system that employs RoPE to expand context beyond what was originally possible for a model. It was discovered and developed by [kaiokendev](https://huggingface.co/kaiokendev).
In order to use the increased context length, you can presently use:
* [KoboldCpp](https://github.com/LostRuins/koboldcpp) - [release 1.33](https://github.com/LostRuins/koboldcpp/releases/tag/v1.33) or later.
Support is also expected to come to llama.cpp, however work is still being done to find the optimal implementation.
To use the increased context with KoboldCpp, simply use `--contextsize` to set the desired context, eg `--contextsize 4096` or `--contextsize 8192`.
**NOTE**: Increased context length is an area seeing rapid developments and improvements. It is quite possible that these models may be superseded by new developments in the coming days. If that's the case, I will remove them, or update this README as appropriate.
## Repositories available
* [4-bit GPTQ models for GPU inference](https://huggingface.co/TheBloke/Selfee-7B-SuperHOT-8K-GPTQ)
* [2, 3, 4, 5, 6 and 8-bit GGML models for CPU inference](https://huggingface.co/TheBloke/Selfee-7B-SuperHOT-8K-GGML)
* [Unquantised SuperHOT fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/TheBloke/Selfee-7B-SuperHOT-8K-fp16)
* [Unquantised base fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/kaist-ai/selfee-7b-delta)
<!-- compatibility_ggml start -->
## Compatibility
These GGMLs will work with any llama.cpp-compatible GGML client that supports k-quants.
However the increased context length won't work without specific support. See the note in the introduction for details on using increased context.
## Explanation of the new k-quant methods
The new methods available are:
* GGML_TYPE_Q2_K - "type-1" 2-bit quantization in super-blocks containing 16 blocks, each block having 16 weight. Block scales and mins are quantized with 4 bits. This ends up effectively using 2.5625 bits per weight (bpw)
* GGML_TYPE_Q3_K - "type-0" 3-bit quantization in super-blocks containing 16 blocks, each block having 16 weights. Scales are quantized with 6 bits. This end up using 3.4375 bpw.
* GGML_TYPE_Q4_K - "type-1" 4-bit quantization in super-blocks containing 8 blocks, each block having 32 weights. Scales and mins are quantized with 6 bits. This ends up using 4.5 bpw.
* GGML_TYPE_Q5_K - "type-1" 5-bit quantization. Same super-block structure as GGML_TYPE_Q4_K resulting in 5.5 bpw
* GGML_TYPE_Q6_K - "type-0" 6-bit quantization. Super-blocks with 16 blocks, each block having 16 weights. Scales are quantized with 8 bits. This ends up using 6.5625 bpw
* GGML_TYPE_Q8_K - "type-0" 8-bit quantization. Only used for quantizing intermediate results. The difference to the existing Q8_0 is that the block size is 256. All 2-6 bit dot products are implemented for this quantization type.
Refer to the Provided Files table below to see what files use which methods, and how.
<!-- compatibility_ggml end -->
## Provided files
| Name | Quant method | Bits | Size | Max RAM required | Use case |
| ---- | ---- | ---- | ---- | ---- | ----- |
| selfee-7b-superhot-8k.ggmlv3.q4_0.bin | q4_0 | 4 | 3.79 GB | 6.29 GB | Original llama.cpp quant method, 4-bit. |
| selfee-7b-superhot-8k.ggmlv3.q4_1.bin | q4_1 | 4 | 4.21 GB | 6.71 GB | Original llama.cpp quant method, 4-bit. Higher accuracy than q4_0 but not as high as q5_0. However has quicker inference than q5 models. |
| selfee-7b-superhot-8k.ggmlv3.q5_0.bin | q5_0 | 5 | 4.63 GB | 7.13 GB | Original llama.cpp quant method, 5-bit. Higher accuracy, higher resource usage and slower inference. |
| selfee-7b-superhot-8k.ggmlv3.q5_1.bin | q5_1 | 5 | 5.06 GB | 7.56 GB | Original llama.cpp quant method, 5-bit. Even higher accuracy, resource usage and slower inference. |
| selfee-7b-superhot-8k.ggmlv3.q8_0.bin | q8_0 | 8 | 7.16 GB | 9.66 GB | Original llama.cpp quant method, 8-bit. Almost indistinguishable from float16. High resource use and slow. Not recommended for most users. |
**Note**: the above RAM figures assume no GPU offloading. If layers are offloaded to the GPU, this will reduce RAM usage and use VRAM instead.
## How to run in `koboldcpp`
On Linux I use the following command line to launch the KoboldCpp UI with OpenCL aceleration and a context size of 4096:
```
python ./koboldcpp.py --stream --unbantokens --threads 8 --usecublas --gpulayers 100 selfee-7b-superhot-8k.ggmlv3.q4_K_M.bin
```
Change `--gpulayers 100` to the number of layers you want/are able to offload to the GPU. Remove it if you don't have GPU acceleration.
For OpenCL acceleration, change `--usecublas` to `--useclblast 0 0`. You may need to change the second `0` to `1` if you have both an iGPU and a discrete GPU.
<!-- footer start -->
## Discord
For further support, and discussions on these models and AI in general, join us at:
[TheBloke AI's Discord server](https://discord.gg/theblokeai)
## Thanks, and how to contribute.
Thanks to the [chirper.ai](https://chirper.ai) team!
I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.
If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.
Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
* Patreon: https://patreon.com/TheBlokeAI
* Ko-Fi: https://ko-fi.com/TheBlokeAI
**Special thanks to**: Luke from CarbonQuill, Aemon Algiz.
**Patreon special mentions**: RoA, Lone Striker, Gabriel Puliatti, Derek Yates, Randy H, Jonathan Leane, Eugene Pentland, Karl Bernard, Viktor Bowallius, senxiiz, Daniel P. Andersen, Pierre Kircher, Deep Realms, Cory Kujawski, Oscar Rangel, Fen Risland, Ajan Kanaga, LangChain4j, webtim, Nikolai Manek, Trenton Dambrowitz, Raven Klaugh, Kalila, Khalefa Al-Ahmad, Chris McCloskey, Luke @flexchar, Ai Maven, Dave, Asp the Wyvern, Sean Connelly, Imad Khwaja, Space Cruiser, Rainer Wilmers, subjectnull, Alps Aficionado, Willian Hasse, Fred von Graf, Artur Olbinski, Johann-Peter Hartmann, WelcomeToTheClub, Willem Michiel, Michael Levine, Iucharbius , Spiking Neurons AB, K, biorpg, John Villwock, Pyrater, Greatston Gnanesh, Mano Prime, Junyu Yang, Stephen Murray, John Detwiler, Luke Pendergrass, terasurfer , Pieter, zynix , Edmond Seymore, theTransient, Nathan LeClaire, vamX, Kevin Schuppel, Preetika Verma, ya boyyy, Alex , SuperWojo, Ghost , Joseph William Delisle, Matthew Berman, Talal Aujan, chris gileta, Illia Dulskyi.
Thank you to all my generous patrons and donaters!
<!-- footer end -->
# Original model card: Kaio Ken's SuperHOT 8K
### SuperHOT Prototype 2 w/ 8K Context
This is a second prototype of SuperHOT, a NSFW focused LoRA, this time 7B with 8K context and no RLHF, using the same technique described in [the github blog](https://kaiokendev.github.io/til#extending-context-to-8k).
#### Looking for Merged & Quantized Models?
Make some please :)
#### Using the monkey-patch?
You will **NEED** to **apply the monkeypatch** or, if you are already using the monkeypatch, **change the scaling factor to 0.25 and the maximum sequence length to 8192**
The monkeypatch is only necessary if you are using a front-end/back-end that does not already support scaling and said front-end/back-end is Python-based (i.e. Huggingface Transformers). To apply the patch, you will need to copy the `llama_rope_scaled_monkey_patch.py` into your working directory and call the exported function `replace_llama_rope_with_scaled_rope` at the very start of your Python program. It will modify the Transformers library's implementation of RoPE to properly apply the scaling factor.
#### Using Oobabooga with Exllama?
Switch your loader to `exllama` or `exllama_hf` Add the arguments `max_seq_len 8192` and `compress_pos_emb 4`. **While the model may work well with `compress_pos_emb 2`, it was trained on 4, so that is what I advocate for you to use**
Example in the command-line:
- `python server.py --max_seq_len 8192 --compress_pos_emb 4 --loader exllama_hf`
In the UI, you will see the loader option in the `Models` tab. Once you select either `exllama` or `exllama_hf`, the `max_seq_len` and `compress_pos_emb` settings will appear.
#### Training Details
I trained the LoRA with the following configuration:
- 1200 samples (~400 samples over 2048 sequence length)
- learning rate of 3e-4
- 3 epochs
- The exported modules are:
- q_proj
- k_proj
- v_proj
- o_proj
- no bias
- Rank = 4
- Alpha = 8
- no dropout
- weight decay of 0.1
- AdamW beta1 of 0.9 and beta2 0.99, epsilon of 1e-5
- Trained on 4-bit base model
- Cutoff length: 4096
# Original model card: Kaist AI's Selfee 7B
<!-- header start -->
<div style="width: 100%;">
<img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
</div>
<div style="display: flex; justify-content: space-between; width: 100%;">
<div style="display: flex; flex-direction: column; align-items: flex-start;">
<p><a href="https://discord.gg/Jq4vkcDakD">Chat & support: my new Discord server</a></p>
</div>
<div style="display: flex; flex-direction: column; align-items: flex-end;">
<p><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p>
</div>
</div>
<!-- header end -->
# Kaist AI's Selfee 7B fp16
These files are pytorch format fp16 model files for [Kaist AI's Selfee 7B](https://huggingface.co/kaist-ai/selfee-7b-delta).
It is the result of merging and/or converting the source repository to float16.
## Repositories available
* [4-bit GPTQ models for GPU inference](https://huggingface.co/TheBloke/selfee-7B-fp16)
* [2, 3, 4, 5, 6 and 8-bit GGML models for CPU+GPU inference](https://huggingface.co/TheBloke/selfee-7B-GGML)
* [Unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/TheBloke/selfee-7B-fp16)
<!-- footer start -->
## Discord
For further support, and discussions on these models and AI in general, join us at:
[TheBloke AI's Discord server](https://discord.gg/Jq4vkcDakD)
## Thanks, and how to contribute.
Thanks to the [chirper.ai](https://chirper.ai) team!
I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.
If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.
Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
* Patreon: https://patreon.com/TheBlokeAI
* Ko-Fi: https://ko-fi.com/TheBlokeAI
**Special thanks to**: Luke from CarbonQuill, Aemon Algiz, Dmitriy Samsonov.
**Patreon special mentions**: Ajan Kanaga, Kalila, Derek Yates, Sean Connelly, Luke, Nathan LeClaire, Trenton Dambrowitz, Mano Prime, David Flickinger, vamX, Nikolai Manek, senxiiz, Khalefa Al-Ahmad, Illia Dulskyi, trip7s trip, Jonathan Leane, Talal Aujan, Artur Olbinski, Cory Kujawski, Joseph William Delisle, Pyrater, Oscar Rangel, Lone Striker, Luke Pendergrass, Eugene Pentland, Johann-Peter Hartmann.
Thank you to all my generous patrons and donaters!
<!-- footer end -->
# Original model card: Kaist AI's Selfee 7B
|
TheBloke/Selfee-13B-SuperHOT-8K-GGML
|
TheBloke
| 2023-07-06T21:38:08Z | 0 | 1 | null |
[
"license:other",
"region:us"
] | null | 2023-07-06T17:16:57Z |
---
inference: false
license: other
---
<!-- header start -->
<div style="width: 100%;">
<img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
</div>
<div style="display: flex; justify-content: space-between; width: 100%;">
<div style="display: flex; flex-direction: column; align-items: flex-start;">
<p><a href="https://discord.gg/theblokeai">Chat & support: my new Discord server</a></p>
</div>
<div style="display: flex; flex-direction: column; align-items: flex-end;">
<p><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p>
</div>
</div>
<!-- header end -->
# Kaist AI's Selfee 13B GGML
These files are GGML format model files for [Kaist AI's Selfee 13B](https://huggingface.co/TheBloke/selfee-13b-fp16).
These are SuperHOT GGMLs with an increased context length. SuperHOT is a new system that employs RoPE to expand context beyond what was originally possible for a model. It was discovered and developed by [kaiokendev](https://huggingface.co/kaiokendev).
In order to use the increased context length, you can presently use:
* [KoboldCpp](https://github.com/LostRuins/koboldcpp) - [release 1.33](https://github.com/LostRuins/koboldcpp/releases/tag/v1.33) or later.
Support is also expected to come to llama.cpp, however work is still being done to find the optimal implementation.
To use the increased context with KoboldCpp, simply use `--contextsize` to set the desired context, eg `--contextsize 4096` or `--contextsize 8192`.
**NOTE**: Increased context length is an area seeing rapid developments and improvements. It is quite possible that these models may be superseded by new developments in the coming days. If that's the case, I will remove them, or update this README as appropriate.
## Repositories available
* [4-bit GPTQ models for GPU inference](https://huggingface.co/TheBloke/Selfee-13B-SuperHOT-8K-GPTQ)
* [2, 3, 4, 5, 6 and 8-bit GGML models for CPU inference](https://huggingface.co/TheBloke/Selfee-13B-SuperHOT-8K-GGML)
* [Unquantised SuperHOT fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/TheBloke/Selfee-13B-SuperHOT-8K-fp16)
* [Unquantised base fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/kaist-ai/selfee-13b-delta)
<!-- compatibility_ggml start -->
## Compatibility
These GGMLs will work with any llama.cpp-compatible GGML client that supports k-quants.
However the increased context length won't work without specific support. See the note in the introduction for details on using increased context.
## Explanation of the new k-quant methods
The new methods available are:
* GGML_TYPE_Q2_K - "type-1" 2-bit quantization in super-blocks containing 16 blocks, each block having 16 weight. Block scales and mins are quantized with 4 bits. This ends up effectively using 2.5625 bits per weight (bpw)
* GGML_TYPE_Q3_K - "type-0" 3-bit quantization in super-blocks containing 16 blocks, each block having 16 weights. Scales are quantized with 6 bits. This end up using 3.4375 bpw.
* GGML_TYPE_Q4_K - "type-1" 4-bit quantization in super-blocks containing 8 blocks, each block having 32 weights. Scales and mins are quantized with 6 bits. This ends up using 4.5 bpw.
* GGML_TYPE_Q5_K - "type-1" 5-bit quantization. Same super-block structure as GGML_TYPE_Q4_K resulting in 5.5 bpw
* GGML_TYPE_Q6_K - "type-0" 6-bit quantization. Super-blocks with 16 blocks, each block having 16 weights. Scales are quantized with 8 bits. This ends up using 6.5625 bpw
* GGML_TYPE_Q8_K - "type-0" 8-bit quantization. Only used for quantizing intermediate results. The difference to the existing Q8_0 is that the block size is 256. All 2-6 bit dot products are implemented for this quantization type.
Refer to the Provided Files table below to see what files use which methods, and how.
<!-- compatibility_ggml end -->
## Provided files
| Name | Quant method | Bits | Size | Max RAM required | Use case |
| ---- | ---- | ---- | ---- | ---- | ----- |
| selfee-13b-superhot-8k.ggmlv3.q4_0.bin | q4_0 | 4 | 7.32 GB | 9.82 GB | Original llama.cpp quant method, 4-bit. |
| selfee-13b-superhot-8k.ggmlv3.q4_1.bin | q4_1 | 4 | 8.14 GB | 10.64 GB | Original llama.cpp quant method, 4-bit. Higher accuracy than q4_0 but not as high as q5_0. However has quicker inference than q5 models. |
| selfee-13b-superhot-8k.ggmlv3.q5_0.bin | q5_0 | 5 | 8.95 GB | 11.45 GB | Original llama.cpp quant method, 5-bit. Higher accuracy, higher resource usage and slower inference. |
| selfee-13b-superhot-8k.ggmlv3.q5_1.bin | q5_1 | 5 | 9.76 GB | 12.26 GB | Original llama.cpp quant method, 5-bit. Even higher accuracy, resource usage and slower inference. |
| selfee-13b-superhot-8k.ggmlv3.q8_0.bin | q8_0 | 8 | 13.83 GB | 16.33 GB | Original llama.cpp quant method, 8-bit. Almost indistinguishable from float16. High resource use and slow. Not recommended for most users. |
**Note**: the above RAM figures assume no GPU offloading. If layers are offloaded to the GPU, this will reduce RAM usage and use VRAM instead.
## How to run in `koboldcpp`
On Linux I use the following command line to launch the KoboldCpp UI with OpenCL aceleration and a context size of 4096:
```
python ./koboldcpp.py --stream --unbantokens --threads 8 --usecublas --gpulayers 100 selfee-13b-superhot-8k.ggmlv3.q4_K_M.bin
```
Change `--gpulayers 100` to the number of layers you want/are able to offload to the GPU. Remove it if you don't have GPU acceleration.
For OpenCL acceleration, change `--usecublas` to `--useclblast 0 0`. You may need to change the second `0` to `1` if you have both an iGPU and a discrete GPU.
<!-- footer start -->
## Discord
For further support, and discussions on these models and AI in general, join us at:
[TheBloke AI's Discord server](https://discord.gg/theblokeai)
## Thanks, and how to contribute.
Thanks to the [chirper.ai](https://chirper.ai) team!
I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.
If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.
Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
* Patreon: https://patreon.com/TheBlokeAI
* Ko-Fi: https://ko-fi.com/TheBlokeAI
**Special thanks to**: Luke from CarbonQuill, Aemon Algiz.
**Patreon special mentions**: RoA, Lone Striker, Gabriel Puliatti, Derek Yates, Randy H, Jonathan Leane, Eugene Pentland, Karl Bernard, Viktor Bowallius, senxiiz, Daniel P. Andersen, Pierre Kircher, Deep Realms, Cory Kujawski, Oscar Rangel, Fen Risland, Ajan Kanaga, LangChain4j, webtim, Nikolai Manek, Trenton Dambrowitz, Raven Klaugh, Kalila, Khalefa Al-Ahmad, Chris McCloskey, Luke @flexchar, Ai Maven, Dave, Asp the Wyvern, Sean Connelly, Imad Khwaja, Space Cruiser, Rainer Wilmers, subjectnull, Alps Aficionado, Willian Hasse, Fred von Graf, Artur Olbinski, Johann-Peter Hartmann, WelcomeToTheClub, Willem Michiel, Michael Levine, Iucharbius , Spiking Neurons AB, K, biorpg, John Villwock, Pyrater, Greatston Gnanesh, Mano Prime, Junyu Yang, Stephen Murray, John Detwiler, Luke Pendergrass, terasurfer , Pieter, zynix , Edmond Seymore, theTransient, Nathan LeClaire, vamX, Kevin Schuppel, Preetika Verma, ya boyyy, Alex , SuperWojo, Ghost , Joseph William Delisle, Matthew Berman, Talal Aujan, chris gileta, Illia Dulskyi.
Thank you to all my generous patrons and donaters!
<!-- footer end -->
# Original model card: Kaio Ken's SuperHOT 8K
### SuperHOT Prototype 2 w/ 8K Context
This is a second prototype of SuperHOT, a NSFW focused LoRA, this time 7B with 8K context and no RLHF, using the same technique described in [the github blog](https://kaiokendev.github.io/til#extending-context-to-8k).
#### Looking for Merged & Quantized Models?
Make some please :)
#### Using the monkey-patch?
You will **NEED** to **apply the monkeypatch** or, if you are already using the monkeypatch, **change the scaling factor to 0.25 and the maximum sequence length to 8192**
The monkeypatch is only necessary if you are using a front-end/back-end that does not already support scaling and said front-end/back-end is Python-based (i.e. Huggingface Transformers). To apply the patch, you will need to copy the `llama_rope_scaled_monkey_patch.py` into your working directory and call the exported function `replace_llama_rope_with_scaled_rope` at the very start of your Python program. It will modify the Transformers library's implementation of RoPE to properly apply the scaling factor.
#### Using Oobabooga with Exllama?
Switch your loader to `exllama` or `exllama_hf` Add the arguments `max_seq_len 8192` and `compress_pos_emb 4`. **While the model may work well with `compress_pos_emb 2`, it was trained on 4, so that is what I advocate for you to use**
Example in the command-line:
- `python server.py --max_seq_len 8192 --compress_pos_emb 4 --loader exllama_hf`
In the UI, you will see the loader option in the `Models` tab. Once you select either `exllama` or `exllama_hf`, the `max_seq_len` and `compress_pos_emb` settings will appear.
#### Training Details
I trained the LoRA with the following configuration:
- 1200 samples (~400 samples over 2048 sequence length)
- learning rate of 3e-4
- 3 epochs
- The exported modules are:
- q_proj
- k_proj
- v_proj
- o_proj
- no bias
- Rank = 4
- Alpha = 8
- no dropout
- weight decay of 0.1
- AdamW beta1 of 0.9 and beta2 0.99, epsilon of 1e-5
- Trained on 4-bit base model
- Cutoff length: 4096
# Original model card: Kaist AI's Selfee 13B
<!-- header start -->
<div style="width: 100%;">
<img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
</div>
<div style="display: flex; justify-content: space-between; width: 100%;">
<div style="display: flex; flex-direction: column; align-items: flex-start;">
<p><a href="https://discord.gg/Jq4vkcDakD">Chat & support: my new Discord server</a></p>
</div>
<div style="display: flex; flex-direction: column; align-items: flex-end;">
<p><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p>
</div>
</div>
<!-- header end -->
# Kaist AI's Selfee 13B GGML
This repo contains fp16 pytorch format model files for [Kaist AI's Selfee 13B](https://huggingface.co/kaist-ai/selfee-13b-delta).
It is the result of merging the diff at the above repo with base Llama 13B, then converting fp32 to fp16.
## Repositories available
* [4-bit GPTQ models for GPU inference](https://huggingface.co/TheBloke/Selfee-13B-GPTQ)
* [4-bit, 5-bit, and 8-bit GGML models for CPU+GPU inference](https://huggingface.co/TheBloke/Selfee-13B-GGML)
* [Unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/TheBloke/Selfee-13B-fp16)
<!-- footer start -->
## Discord
For further support, and discussions on these models and AI in general, join us at:
[TheBloke AI's Discord server](https://discord.gg/Jq4vkcDakD)
## Thanks, and how to contribute.
Thanks to the [chirper.ai](https://chirper.ai) team!
I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.
If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.
Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
* Patreon: https://patreon.com/TheBlokeAI
* Ko-Fi: https://ko-fi.com/TheBlokeAI
**Special thanks to**: Luke from CarbonQuill, Aemon Algiz, Dmitriy Samsonov.
**Patreon special mentions**: Derek Yates, Sean Connelly, Luke, Nathan LeClaire, Trenton Dambrowitz, Mano Prime, David Flickinger, vamX, Nikolai Manek, senxiiz, Khalefa Al-Ahmad, Illia Dulskyi, trip7s trip, Jonathan Leane, Talal Aujan, Artur Olbinski, Cory Kujawski, Joseph William Delisle, Pyrater, Oscar Rangel, Lone Striker, Luke Pendergrass, Eugene Pentland, Johann-Peter Hartmann.
Thank you to all my generous patrons and donaters!
<!-- footer end -->
# Original model card: Kaist AI's Selfee 13B
<p align="center" width="100%">
<a href="https://kaistai.github.io/SelFee/demo" target="_blank"><img src="https://raw.githubusercontent.com/kaistAI/SelFee/main/assets/llama_selfie.png" alt="KAIST-Selfee" style="width: 30%; min-width: 200px; display: block; margin: auto;"></a>
</p>
# SelFee: Iterative Self-Revising LLM Empowered by <br/> Self-Feedback Generation
[](https://github.com/tatsu-lab/stanford_alpaca/blob/main/LICENSE)
[](https://github.com/tatsu-lab/stanford_alpaca/blob/main/DATA_LICENSE)
[](https://www.python.org/downloads/release/python-390/)
[](https://github.com/psf/black)
## News
[May 31, 2023] Initial release: We released the first version of SelFee! Check out the <a href="https://kaistai.github.io/SelFee/">blog post</a> for more details.
## Overview
This is the repository for the KAIST SelFee project, which aims to build and share an instruction-following LLaMA model. This repo mainly has five contents:
- The selection process of the 178K training data for SelFee ([detail](#data-release), [code](data_collection)).
- The generation process for the training data and its result. ([detail](#data-generation-process), [code](data_augmentation)).
- The training process for the model ([detail](#training), [code](train)).
- The inference process for the model ([detail](#inference), [code](inference)).
- The evaluation method and dataset ([detail](#evaluation), [code](evaluation)).
This repository is based on the [Stanford-Alpaca](https://github.com/tatsu-lab/stanford_alpaca/) and [Vicuna](https://github.com/lm-sys/FastChat/) repository. Thanks to all the contributors for these awesome repositories!! 🙌
**We highly recommend you read our [blog post](https://kaistai.github.io/SelFee/) for more details about the model.**
## Data Release
For data collection, we collected datasets from five different fields. These are the Stanford Alpaca dataset, math collection, code collection, Flan collection, and ShareGPT. We provide code that we used to make a dataset for training. We also provide code how we preprocessed ShareGPT. For ShareGPT, we only use the first (question, answer) pair from human and GPT, respectively. We only use instances which are classified as english,and filter instance which is not a form of question.
For other datsets, we do not need special data collection method.
## Data Generation Process
To train our model with high-quality instructions and answer pairs, we utilized data augmentation using OpenAI API calls. The process involved three steps. <br>
Firstly, we collected various instructions from multiple fields and fed them to ChatGPT to generate answers. <br>
Secondly, we gathered feedback on the generated answer by querying ChatGPT again and asked it to determine if the initial answer required any revision. <br>
Thirdly, if a revision was necessary, we passed the instruction, initial answer, and feedback pair to ChatGPT to generate a revised answer and its feedback pair.
We repeated the process until we received feedback that required no further revision or hit the maximum iteration. However, due to the token limitation of the ChatGPT API, we had to truncate some instances that needed more than 4096 tokens while augmenting.<br>
You can see the details with command [here](data_augmentation/README.md).<br>
*We provide the whole dataset after collection and augmentation using huggingface([code](data_collection/download_train.py)), so you can either use the code or follow our [data merging step](outputs/README.md) to replicate the training dataset. Feel free to use any of them!
## Training
We utilize <a href="https://github.com/lm-sys/FastChat">FastChat</a> to train the model. Given the instruction, we fine-tune the model to generate the answer and feedback chain (including the revisions).<br>
To reproduce the training procedure, here are the steps. <br>
```
pip install -r requirements.txt
```
```
torchrun --nproc_per_node=4 train/train_mem.py \
--model_name_or_path llama-7b \
--data_path outputs/feedback_gpt_3.5_turbo_merged_whole.json \
--bf16 True \
--output_dir ckpt/selfee-7b \
--num_train_epochs 3 \
--per_device_train_batch_size 16 \
--per_device_eval_batch_size 16 \
--gradient_accumulation_steps 2 \
--evaluation_strategy "no" \
--save_strategy "steps" \
--save_steps 5000 \
--save_total_limit 1 \
--learning_rate 2e-5 \
--weight_decay 0. \
--warmup_ratio 0.03 \
--lr_scheduler_type "cosine" \
--logging_steps 1 \
--fsdp "shard_grad_op auto_wrap" \
--fsdp_transformer_layer_cls_to_wrap 'LlamaDecoderLayer' \
--tf32 True \
--model_max_length 2048 \
--gradient_checkpointing True \
--lazy_preprocess True \
--training_objective full \
```
The hyperparameters are as follows, following Vicuna and Alpaca.
| Hyperparameter | Global Batch Size | Learning rate | Epochs | Max length | Weight decay |
| --- | ---: | ---: | ---: | ---: | ---: |
| SelFee (7B, 13B) | 128 | 2e-5 | 3 | 2048 | 0 |
## Inference
<b>Restoring checkpoint using diff</b><br>
We provide diff weight and code which can restore the same model with SelFee. To restore the original SelFee weight, you first need to convert the Meta's original LLAMA checkpoint into huggingface format into your local machine. Once you are done, you can restore the same checkpoint of our model by using the following command
```
python inference/apply_delta.py --path_raw {path_to_llama_7b} --path_tuned /ckpt/selfee-7b --path_diff kaist-ai/selfee-7b-delta
```
<b>Autonomous Inference Mode</b><br>
Because SelFee is trained to generate iterative feedback and revisions until the response is satisfying, it automatically generates iterative feedback and revisions on a single forward pass. The model autonomously decides when to stop generating revisions based on the feedback. If the feedback chain ends with sequences like `Revision is not needed.`, the model autonomously terminates generation. <br>
For autonomous inference mode,
```
python inference/inference.py --model-path "ckpt/selfee-7b" --model-id "selfee" --question-file "evaluation/template/question.jsonl" --answer-file "evaluation/answer/selfee_7b_autonomous.jsonl"
```
<b>Revision Enforce Inference Mode</b><br>
We observed that increasing the minimum number of required revisions corresponds to a corresponding increase in performance. To enforce revisions, we automatically replace sequences such as `Revision is not needed.` into `Revision is needed.` during self-feedback generation. Because SelFee is trained to generate `Revision {index}:` after the sequence of `Revision is needed.`, the model would continually revise the answer.
For revision enforce inference mode, use the `max-num-revision` argument.
```
python inference/inference.py --model-path "ckpt/selfee-7b" --model-id "selfee" --question-file "evaluation/template/question.jsonl" --answer-file "evaluation/answer/selfee_7b_enforce_3_revision.jsonl" --max-num-revision 3
```
## Evaluation
Following evaluation setting of Vicuna, we evaluate on 80 diverse queries and utilize GPT-4 language model as the evaluator, scoring a model's response relative to ChatGPT's response. One of the difference with Vicuna evaluation is that due to positional bias of GPT-4, we employ a bidirectional evaluation setting. This means that each evaluation instance is inferred twice, depending on its position.<br>
We release the inference result of SelFee in the folder of `evaluation/answer` and also the scores generated by GPT-4 in the folder of `evaluation/review`. <br>
### GPT-4 Automatic Evaluation
First, you need to get your API key to get access to the GPT-4 API.
```
export OPENAI_API_KEYS={personal_key}
```
To compare the performance of a generation result (for example, located on `evaluation/answer/file_A.jsonl`) with another generation result (located on `evaluation/anwer/file_B.jsonl`),
```
python evaluation/gpt4_automatic_evaluation.py -q evaluation/template/question.jsonl -a evaluation/answer/file_A.jsonl evaluation/answer/file_B.jsonl -p evaluation/template/prompt.jsonl -r evaluation/template/reviewer.jsonl -o evaluation/review/A_vs_B.jsonl
```
To mitigate the positional bias of GPT-4 model, we apply a bidirectional evaluation setting. Therefore, automatic evaluation with opposite position is also needed.
```
python evaluation/gpt4_automatic_evaluation.py -q evaluation/template/question.jsonl -a evaluation/answer/file_B.jsonl evaluation/answer/file_A.jsonl -p evaluation/template/prompt.jsonl -r evaluation/template/reviewer.jsonl -o evaluation/review/B_vs_A.jsonl
```
## Limitations
Similar to other LLaMA-finetuned models, SelFee also make some mistakes especially for math, reasoning, factuality, and coding tasks. Although our performance outperforms ChatGPT on Vicuna setting, the evaluation setting contains some limitations in terms of comprehension (limited to 80 queries), inconsistency, and unreliability. Therefore, further research for a better evaluation setting is needed. Please take these claims with a grain of salt.
## Online demo
Check out the <a href="https://kaistai.github.io/SelFee/demo">demo</a>!
#### How to launch the demo yourself
To serve the web demo yourself, run the following commands:
1. Run the controller
```
python3 -m serve.controller
```
2. Run the model worker
```
python3 -m serve.model_worker --model-path $MODEL_PATH --port 21002 --worker-address=http://localhost:21002 --model-name=SelFee-13b
```
3. Run the web server
```
python3 -m serve.gradio_web_server --share
```
You can find the serving code [here](serve).
### Team members
<a href="https://seonghyeonye.github.io/)">Seonghyeon Ye*</a>, <a href="https://github.com/dreamgonfly">Yongrae Jo*</a>, <a href="https://github.com/doeyoungkim">Doyoung Kim*</a>, <a href="https://scholar.google.com/citations?user=xKrSnDoAAAAJ&hl">Sungdong Kim</a>, <a href="https://github.com/hbin0701">Hyeonbin Hwang</a>, and <a href="https://seominjoon.github.io/">Minjoon Seo</a>. <br/>
(* denotes equal contribution)
### Release
We have released the SelFee-7B and SelFee-13B model diff weights, which can be found with instructions here. Moreover, the training instances used to train SelFee is released on huggingface.
### License
The research preview online demo is only for non-commercial use and is subject to various licenses and terms of use, including the LLaMA model <a href="https://github.com/facebookresearch/llama/blob/main/MODEL_CARD.md">License</a>, OpenAI's <a href="https://openai.com/policies/terms-of-use">Terms of Use</a> for the generated data, and ShareGPT's <a href="https://chrome.google.com/webstore/detail/sharegpt-share-your-chatg/daiacboceoaocpibfodeljbdfacokfjb">Privacy Practices</a>. If you suspect any violations, please reach out to us.
### Citation
Please cite if you use the data or code in this repo.
```
@misc{selfee2023,
author = {Ye, Seonghyeon and Jo, Yongrae and Kim, Doyoung and Kim, Sungdong and Hwang, Hyeonbin and Seo, Minjoon},
title = {SelFee: Iterative Self-Revising LLM Empowered by Self-Feedback Generation},
url = {https://kaistai.github.io/SelFee/},
month = {May},
year = {2023},
howpublished = {Blog post}
}
```
|
ahmedALM1221/convnextv2-large-1k-224-finetuned-eurosat-50
|
ahmedALM1221
| 2023-07-06T21:34:41Z | 8 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"convnextv2",
"image-classification",
"generated_from_trainer",
"dataset:imagefolder",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
image-classification
| 2023-07-04T14:11:37Z |
---
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- imagefolder
metrics:
- accuracy
model-index:
- name: convnextv2-large-1k-224-finetuned-eurosat-50
results:
- task:
name: Image Classification
type: image-classification
dataset:
name: imagefolder
type: imagefolder
config: Augmented-Final
split: train
args: Augmented-Final
metrics:
- name: Accuracy
type: accuracy
value: 0.959917780061665
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# convnextv2-large-1k-224-finetuned-eurosat-50
This model is a fine-tuned version of [facebook/convnextv2-large-1k-224](https://huggingface.co/facebook/convnextv2-large-1k-224) on the imagefolder dataset.
It achieves the following results on the evaluation set:
- Loss: 0.1615
- Accuracy: 0.9599
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.9
- num_epochs: 12
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| 1.9003 | 1.0 | 122 | 1.8892 | 0.3587 |
| 1.6055 | 2.0 | 244 | 1.5936 | 0.5766 |
| 1.2982 | 3.0 | 366 | 1.2568 | 0.6639 |
| 1.0273 | 4.0 | 488 | 0.9558 | 0.7431 |
| 0.7863 | 5.0 | 610 | 0.7440 | 0.7965 |
| 0.8142 | 6.0 | 732 | 0.6206 | 0.8304 |
| 0.6118 | 7.0 | 854 | 0.4672 | 0.8684 |
| 0.4776 | 8.0 | 976 | 0.3756 | 0.9075 |
| 0.4176 | 9.0 | 1098 | 0.3065 | 0.9301 |
| 0.282 | 10.0 | 1220 | 0.2586 | 0.9404 |
| 0.306 | 11.0 | 1342 | 0.2098 | 0.9476 |
| 0.2006 | 12.0 | 1464 | 0.1615 | 0.9599 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu118
- Datasets 2.13.1
- Tokenizers 0.13.3
|
am-infoweb/question-answering-roberta-anu-SQuAD-v2
|
am-infoweb
| 2023-07-06T21:32:52Z | 106 | 0 |
transformers
|
[
"transformers",
"pytorch",
"safetensors",
"roberta",
"question-answering",
"Question Answering",
"license:apache-2.0",
"endpoints_compatible",
"region:us"
] |
question-answering
| 2023-07-06T21:27:17Z |
---
license: apache-2.0
tags:
- Question Answering
metrics:
- squad
model-index:
- name: anuragsingh28/question-answering-roberta-anu-s-v2
results: []
---
# Question Answering
The model is intended to be used for Q&A task, given the question & context, the model would attempt to infer the answer text, answer span & confidence score.<br>
Model is encoder-only (deepset/roberta-base-squad2) with QuestionAnswering LM Head, fine-tuned on SQUADx dataset with **exact_match:** 84.83 & **f1:** 91.80 performance scores.
Please follow this link for [Encoder based Question Answering V1](https://huggingface.co/anuragsingh28/question-answering-roberta-anu-s-v2/)
Example code:
```
from transformers import pipeline
model_checkpoint = "anuragsingh28/question-answering-roberta-anu-s-v2"
context = """
🤗 Transformers is backed by the three most popular deep learning libraries — Jax, PyTorch and TensorFlow — with a seamless integration
between them. It's straightforward to train your models with one before loading them for inference with the other.
"""
question = "Which deep learning libraries back 🤗 Transformers?"
question_answerer = pipeline("question-answering", model=model_checkpoint)
question_answerer(question=question, context=context)
```
## Training and evaluation data
SQUAD Split
## Training procedure
Preprocessing:
1. SQUAD Data longer chunks were sub-chunked with input context max-length 384 tokens and stride as 128 tokens.
2. Target answers readjusted for sub-chunks, sub-chunks with no-answers or partial answers were set to target answer span as (0,0)
Metrics:
1. Adjusted accordingly to handle sub-chunking.
2. n best = 20
3. skip answers with length zero or higher than max answer length (30)
### Training hyperparameters
Custom Training Loop:
The following hyperparameters were used during training:
- learning_rate: 2e-5
- train_batch_size: 32
- eval_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 2
### Training results
{'exact_match': 84.83443708609272, 'f1': 91.79987545811638}
### Framework versions
- Transformers 4.23.0.dev0
- Pytorch 1.12.1+cu113
- Datasets 2.5.2
- Tokenizers 0.13.0
|
aroot/eng-guj-simcse_random_usrb
|
aroot
| 2023-07-06T21:19:25Z | 103 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"mbart",
"text2text-generation",
"translation",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
translation
| 2023-07-06T20:57:54Z |
---
tags:
- translation
- generated_from_trainer
metrics:
- bleu
model-index:
- name: eng-guj-simcse_random_usrb
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# eng-guj-simcse_random_usrb
This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 3.2995
- Bleu: 2.6979
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3
|
YakovElm/Apache_20_BERT_More_Properties
|
YakovElm
| 2023-07-06T21:18:12Z | 61 | 0 |
transformers
|
[
"transformers",
"tf",
"bert",
"text-classification",
"generated_from_keras_callback",
"license:apache-2.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2023-07-06T21:17:37Z |
---
license: apache-2.0
tags:
- generated_from_keras_callback
model-index:
- name: Apache_20_BERT_More_Properties
results: []
---
<!-- This model card has been generated automatically according to the information Keras had access to. You should
probably proofread and complete it, then remove this comment. -->
# Apache_20_BERT_More_Properties
This model is a fine-tuned version of [bert-base-uncased](https://huggingface.co/bert-base-uncased) on an unknown dataset.
It achieves the following results on the evaluation set:
- Train Loss: 0.1597
- Train Accuracy: 0.9624
- Validation Loss: 0.3500
- Validation Accuracy: 0.9055
- Epoch: 2
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- optimizer: {'name': 'Adam', 'weight_decay': None, 'clipnorm': 1.0, 'global_clipnorm': None, 'clipvalue': None, 'use_ema': False, 'ema_momentum': 0.99, 'ema_overwrite_frequency': None, 'jit_compile': False, 'is_legacy_optimizer': False, 'learning_rate': 3e-05, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-08, 'amsgrad': False}
- training_precision: float32
### Training results
| Train Loss | Train Accuracy | Validation Loss | Validation Accuracy | Epoch |
|:----------:|:--------------:|:---------------:|:-------------------:|:-----:|
| 0.1743 | 0.9589 | 0.3407 | 0.9055 | 0 |
| 0.1613 | 0.9624 | 0.3694 | 0.9055 | 1 |
| 0.1597 | 0.9624 | 0.3500 | 0.9055 | 2 |
### Framework versions
- Transformers 4.29.2
- TensorFlow 2.12.0
- Datasets 2.12.0
- Tokenizers 0.13.3
|
cjohlmacher/unit2-taxi-focused
|
cjohlmacher
| 2023-07-06T21:13:18Z | 0 | 0 | null |
[
"Taxi-v3",
"q-learning",
"reinforcement-learning",
"custom-implementation",
"model-index",
"region:us"
] |
reinforcement-learning
| 2023-07-06T21:13:16Z |
---
tags:
- Taxi-v3
- q-learning
- reinforcement-learning
- custom-implementation
model-index:
- name: unit2-taxi-focused
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: Taxi-v3
type: Taxi-v3
metrics:
- type: mean_reward
value: 7.24 +/- 2.80
name: mean_reward
verified: false
---
# **Q-Learning** Agent playing1 **Taxi-v3**
This is a trained model of a **Q-Learning** agent playing **Taxi-v3** .
## Usage
```python
model = load_from_hub(repo_id="cjohlmacher/unit2-taxi-focused", filename="q-learning.pkl")
# Don't forget to check if you need to add additional attributes (is_slippery=False etc)
env = gym.make(model["env_id"])
```
|
aroot/eng-mya-simcse_central_usblu
|
aroot
| 2023-07-06T21:12:43Z | 103 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"mbart",
"text2text-generation",
"translation",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
translation
| 2023-07-06T20:52:09Z |
---
tags:
- translation
- generated_from_trainer
metrics:
- bleu
model-index:
- name: eng-mya-simcse_central_usblu
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# eng-mya-simcse_central_usblu
This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 1.8855
- Bleu: 4.1385
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3
|
BubbleJoe/swin-tiny-patch4-window7-224-finetuned-eurosat
|
BubbleJoe
| 2023-07-06T20:47:28Z | 213 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"swin",
"image-classification",
"generated_from_trainer",
"dataset:imagefolder",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
image-classification
| 2023-07-06T20:30:39Z |
---
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- imagefolder
metrics:
- accuracy
model-index:
- name: swin-tiny-patch4-window7-224-finetuned-eurosat
results:
- task:
name: Image Classification
type: image-classification
dataset:
name: imagefolder
type: imagefolder
config: default
split: train
args: default
metrics:
- name: Accuracy
type: accuracy
value: 0.9748148148148148
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# swin-tiny-patch4-window7-224-finetuned-eurosat
This model is a fine-tuned version of [microsoft/swin-tiny-patch4-window7-224](https://huggingface.co/microsoft/swin-tiny-patch4-window7-224) on the imagefolder dataset.
It achieves the following results on the evaluation set:
- Loss: 0.0742
- Accuracy: 0.9748
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 128
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 3
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| 0.2967 | 1.0 | 190 | 0.1191 | 0.9622 |
| 0.1776 | 2.0 | 380 | 0.0897 | 0.9719 |
| 0.1334 | 3.0 | 570 | 0.0742 | 0.9748 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu118
- Datasets 2.13.1
- Tokenizers 0.13.3
|
pineiden/fake-news-bert-base-spanish-wwm-cased
|
pineiden
| 2023-07-06T20:42:06Z | 111 | 0 |
transformers
|
[
"transformers",
"tensorboard",
"safetensors",
"bert",
"text-classification",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2023-07-06T20:31:38Z |
---
tags:
- generated_from_trainer
metrics:
- f1
model-index:
- name: fake-news-bert-base-spanish-wwm-cased
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# fake-news-bert-base-spanish-wwm-cased
This model is a fine-tuned version of [dccuchile/bert-base-spanish-wwm-cased](https://huggingface.co/dccuchile/bert-base-spanish-wwm-cased) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.5154
- F1: 0.8957
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 13
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5
### Training results
| Training Loss | Epoch | Step | Validation Loss | F1 |
|:-------------:|:-----:|:----:|:---------------:|:------:|
| 0.5098 | 1.0 | 140 | 0.4978 | 0.7340 |
| 0.2473 | 2.0 | 280 | 0.3888 | 0.8829 |
| 0.0908 | 3.0 | 420 | 0.4420 | 0.8969 |
| 0.0332 | 4.0 | 560 | 0.5604 | 0.8796 |
| 0.0052 | 5.0 | 700 | 0.5154 | 0.8957 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu118
- Datasets 2.13.1
- Tokenizers 0.13.3
|
espnet/simpleoier_librispeech_hubert_iter0_train_ssl_torchaudiohubert_base_960h_pretrain_it0_raw
|
espnet
| 2023-07-06T20:38:53Z | 1 | 0 |
espnet
|
[
"espnet",
"audio",
"self-supervised-learning",
"en",
"dataset:librispeech",
"arxiv:1804.00015",
"license:cc-by-4.0",
"region:us"
] | null | 2022-12-31T03:54:13Z |
---
tags:
- espnet
- audio
- self-supervised-learning
language: en
datasets:
- librispeech
license: cc-by-4.0
---
## ESPnet2 SSL model
### `simpleoier/simpleoier_librispeech_hubert_iter0_train_ssl_torchaudiohubert_base_960h_pretrain_it0_raw`
This model was trained by simpleoier using librispeech recipe in [espnet](https://github.com/espnet/espnet/).
### Demo: How to use in ESPnet2
Follow the [ESPnet installation instructions](https://espnet.github.io/espnet/installation.html)
if you haven't done that already.
```bash
cd espnet
git checkout 753f40d61813436d4e76660904d02eaed7a6649e
pip install -e .
cd egs2/librispeech/ssl1
./run.sh --skip_data_prep false --skip_train true --download_model simpleoier/simpleoier_librispeech_hubert_iter0_train_ssl_torchaudiohubert_base_960h_pretrain_it0_raw
```
## SSL config
<details><summary>expand</summary>
```
config: conf/tuning/train_ssl_torchaudiohubert_base_960h_pretrain_it0.yaml
print_config: false
log_level: INFO
dry_run: false
iterator_type: sequence
output_dir: exp/hubert_iter0_train_ssl_torchaudiohubert_base_960h_pretrain_it0_raw
ngpu: 1
seed: 0
num_workers: 64
num_att_plot: 3
dist_backend: nccl
dist_init_method: env://
dist_world_size: 8
dist_rank: 0
local_rank: 0
dist_master_addr: localhost
dist_master_port: 45091
dist_launcher: null
multiprocessing_distributed: true
unused_parameters: true
sharded_ddp: false
cudnn_enabled: true
cudnn_benchmark: false
cudnn_deterministic: true
collect_stats: false
write_collected_feats: false
max_epoch: 250
patience: null
val_scheduler_criterion:
- valid
- loss
early_stopping_criterion:
- valid
- loss
- min
best_model_criterion:
- - valid
- loss
- min
keep_nbest_models: 10
nbest_averaging_interval: 0
grad_clip: 5.0
grad_clip_type: 2.0
grad_noise: false
accum_grad: 2
no_forward_run: false
resume: true
train_dtype: float32
use_amp: true
log_interval: null
use_matplotlib: true
use_tensorboard: true
create_graph_in_tensorboard: false
use_wandb: false
wandb_project: null
wandb_id: null
wandb_entity: null
wandb_name: null
wandb_model_log_interval: -1
detect_anomaly: false
pretrain_path: null
init_param: []
ignore_init_mismatch: false
freeze_param: []
num_iters_per_epoch: null
batch_size: 20
valid_batch_size: null
batch_bins: 48000000
valid_batch_bins: null
train_shape_file:
- exp/hubert_iter0_stats_raw/train/speech_shape
- exp/hubert_iter0_stats_raw/train/text_shape.word
valid_shape_file:
- exp/hubert_iter0_stats_raw/valid/speech_shape
- exp/hubert_iter0_stats_raw/valid/text_shape.word
batch_type: numel
valid_batch_type: null
fold_length:
- 80000
- 400
sort_in_batch: descending
sort_batch: descending
multiple_iterator: false
chunk_length: 500
chunk_shift_ratio: 0.5
num_cache_chunks: 1024
train_data_path_and_name_and_type:
- - dump/raw/train_960/wav.scp
- speech
- sound
- - dump/raw/train_960/text.km.kmeans_iter0_mfcc_train_960_portion0.1
- text
- text
valid_data_path_and_name_and_type:
- - dump/raw/dev/wav.scp
- speech
- sound
- - dump/raw/dev/text.km.kmeans_iter0_mfcc_train_960_portion0.1
- text
- text
allow_variable_data_keys: false
max_cache_size: 0.0
max_cache_fd: 32
valid_max_cache_size: null
optim: adam
optim_conf:
lr: 0.0005
scheduler: warmuplr
scheduler_conf:
warmup_steps: 32000
token_list:
- '81'
- '5'
- '79'
- '84'
- '27'
- '35'
- '67'
- '56'
- '10'
- '99'
- '24'
- '3'
- '48'
- '8'
- '42'
- '16'
- '32'
- '31'
- '47'
- '43'
- '20'
- '73'
- '49'
- '86'
- '18'
- '64'
- '34'
- '59'
- '95'
- '0'
- '52'
- '44'
- '61'
- '57'
- '30'
- '1'
- '93'
- '6'
- '69'
- '19'
- '7'
- '65'
- '28'
- '89'
- '2'
- '96'
- '91'
- '72'
- '38'
- '78'
- '26'
- '13'
- '39'
- '94'
- '4'
- '88'
- '85'
- '51'
- '82'
- '41'
- '50'
- '21'
- '80'
- '97'
- '87'
- '25'
- '54'
- '12'
- '40'
- '60'
- '29'
- '11'
- '53'
- '71'
- '83'
- '74'
- '68'
- '55'
- '62'
- '76'
- '45'
- '75'
- '92'
- '46'
- '36'
- '66'
- '22'
- '77'
- '23'
- '63'
- '37'
- '58'
- '33'
- '15'
- '17'
- '90'
- '98'
- '14'
- '70'
- '9'
- <unk>
- <sos/eos>
init: null
collate_fn_conf:
label_downsampling: 2
pad: false
rand_crop: true
input_size: 1
num_classes: 100
use_preprocessor: true
token_type: word
bpemodel: null
non_linguistic_symbols: null
cleaner: null
g2p: null
speech_volume_normalize: null
rir_scp: null
rir_apply_prob: 1.0
noise_scp: null
noise_apply_prob: 1.0
noise_db_range: '13_15'
pred_masked_weight: 1.0
pred_nomask_weight: 0.0
loss_weights: 0.0
frontend: null
frontend_conf: {}
specaug: null
specaug_conf: {}
normalize: null
normalize_conf: {}
preencoder: null
preencoder_conf: {}
encoder: torchaudio_hubert
encoder_conf:
encoder_projection_dropout: 0.1
encoder_attention_dropout: 0.1
encoder_ff_interm_dropout: 0.0
encoder_dropout: 0.1
encoder_layer_drop: 0.05
model: torchaudio
model_conf: {}
required:
- output_dir
- token_list
version: '202209'
distributed: true
```
</details>
### Citing ESPnet
```BibTex
@inproceedings{watanabe2018espnet,
author={Shinji Watanabe and Takaaki Hori and Shigeki Karita and Tomoki Hayashi and Jiro Nishitoba and Yuya Unno and Nelson Yalta and Jahn Heymann and Matthew Wiesner and Nanxin Chen and Adithya Renduchintala and Tsubasa Ochiai},
title={{ESPnet}: End-to-End Speech Processing Toolkit},
year={2018},
booktitle={Proceedings of Interspeech},
pages={2207--2211},
doi={10.21437/Interspeech.2018-1456},
url={http://dx.doi.org/10.21437/Interspeech.2018-1456}
}
```
or arXiv:
```bibtex
@misc{watanabe2018espnet,
title={ESPnet: End-to-End Speech Processing Toolkit},
author={Shinji Watanabe and Takaaki Hori and Shigeki Karita and Tomoki Hayashi and Jiro Nishitoba and Yuya Unno and Nelson Yalta and Jahn Heymann and Matthew Wiesner and Nanxin Chen and Adithya Renduchintala and Tsubasa Ochiai},
year={2018},
eprint={1804.00015},
archivePrefix={arXiv},
primaryClass={cs.CL}
}
```
|
WALIDALI/oumadvenly
|
WALIDALI
| 2023-07-06T20:38:46Z | 0 | 0 |
diffusers
|
[
"diffusers",
"safetensors",
"text-to-image",
"stable-diffusion",
"license:creativeml-openrail-m",
"autotrain_compatible",
"endpoints_compatible",
"diffusers:StableDiffusionPipeline",
"region:us"
] |
text-to-image
| 2023-07-06T20:33:28Z |
---
license: creativeml-openrail-m
tags:
- text-to-image
- stable-diffusion
---
### oumadvenly Dreambooth model trained by WALIDALI with [TheLastBen's fast-DreamBooth](https://colab.research.google.com/github/TheLastBen/fast-stable-diffusion/blob/main/fast-DreamBooth.ipynb) notebook
Test the concept via A1111 Colab [fast-Colab-A1111](https://colab.research.google.com/github/TheLastBen/fast-stable-diffusion/blob/main/fast_stable_diffusion_AUTOMATIC1111.ipynb)
Sample pictures of this concept:
|
TheBloke/WizardLM-7B-V1-0-Uncensored-SuperHOT-8K-GGML
|
TheBloke
| 2023-07-06T20:38:02Z | 0 | 4 | null |
[
"license:other",
"region:us"
] | null | 2023-07-06T18:47:36Z |
---
inference: false
license: other
---
<!-- header start -->
<div style="width: 100%;">
<img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
</div>
<div style="display: flex; justify-content: space-between; width: 100%;">
<div style="display: flex; flex-direction: column; align-items: flex-start;">
<p><a href="https://discord.gg/theblokeai">Chat & support: my new Discord server</a></p>
</div>
<div style="display: flex; flex-direction: column; align-items: flex-end;">
<p><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p>
</div>
</div>
<!-- header end -->
# Eric Hartford's WizardLM-7B-V1.0-Uncensored GGML
These files are GGML format model files for [Eric Hartford's WizardLM-7B-V1.0-Uncensored](https://huggingface.co/ehartford/WizardLM-7B-V1.0-Uncensored).
These are SuperHOT GGMLs with an increased context length. SuperHOT is a new system that employs RoPE to expand context beyond what was originally possible for a model. It was discovered and developed by [kaiokendev](https://huggingface.co/kaiokendev).
In order to use the increased context length, you can presently use:
* [KoboldCpp](https://github.com/LostRuins/koboldcpp) - [release 1.33](https://github.com/LostRuins/koboldcpp/releases/tag/v1.33) or later.
Support is also expected to come to llama.cpp, however work is still being done to find the optimal implementation.
To use the increased context with KoboldCpp, simply use `--contextsize` to set the desired context, eg `--contextsize 4096` or `--contextsize 8192`.
**NOTE**: Increased context length is an area seeing rapid developments and improvements. It is quite possible that these models may be superseded by new developments in the coming days. If that's the case, I will remove them, or update this README as appropriate.
## Repositories available
* [4-bit GPTQ models for GPU inference](https://huggingface.co/TheBloke/WizardLM-7B-V1-0-Uncensored-SuperHOT-8K-GPTQ)
* [2, 3, 4, 5, 6 and 8-bit GGML models for CPU inference](https://huggingface.co/TheBloke/WizardLM-7B-V1-0-Uncensored-SuperHOT-8K-GGML)
* [Unquantised SuperHOT fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/TheBloke/WizardLM-7B-V1-0-Uncensored-SuperHOT-8K-fp16)
* [Unquantised base fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/ehartford/WizardLM-7B-V1.0-Uncensored)
<!-- compatibility_ggml start -->
## Compatibility
These GGMLs will work with any llama.cpp-compatible GGML client that supports k-quants.
However the increased context length won't work without specific support. See the note in the introduction for details on using increased context.
## Explanation of the new k-quant methods
The new methods available are:
* GGML_TYPE_Q2_K - "type-1" 2-bit quantization in super-blocks containing 16 blocks, each block having 16 weight. Block scales and mins are quantized with 4 bits. This ends up effectively using 2.5625 bits per weight (bpw)
* GGML_TYPE_Q3_K - "type-0" 3-bit quantization in super-blocks containing 16 blocks, each block having 16 weights. Scales are quantized with 6 bits. This end up using 3.4375 bpw.
* GGML_TYPE_Q4_K - "type-1" 4-bit quantization in super-blocks containing 8 blocks, each block having 32 weights. Scales and mins are quantized with 6 bits. This ends up using 4.5 bpw.
* GGML_TYPE_Q5_K - "type-1" 5-bit quantization. Same super-block structure as GGML_TYPE_Q4_K resulting in 5.5 bpw
* GGML_TYPE_Q6_K - "type-0" 6-bit quantization. Super-blocks with 16 blocks, each block having 16 weights. Scales are quantized with 8 bits. This ends up using 6.5625 bpw
* GGML_TYPE_Q8_K - "type-0" 8-bit quantization. Only used for quantizing intermediate results. The difference to the existing Q8_0 is that the block size is 256. All 2-6 bit dot products are implemented for this quantization type.
Refer to the Provided Files table below to see what files use which methods, and how.
<!-- compatibility_ggml end -->
## Provided files
| Name | Quant method | Bits | Size | Max RAM required | Use case |
| ---- | ---- | ---- | ---- | ---- | ----- |
| wizardlm-7b-v1.0-superhot-8k.ggmlv3.q2_K.bin | q2_K | 2 | 2.87 GB | 5.37 GB | New k-quant method. Uses GGML_TYPE_Q4_K for the attention.vw and feed_forward.w2 tensors, GGML_TYPE_Q2_K for the other tensors. |
| wizardlm-7b-v1.0-superhot-8k.ggmlv3.q3_K_L.bin | q3_K_L | 3 | 3.60 GB | 6.10 GB | New k-quant method. Uses GGML_TYPE_Q5_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else GGML_TYPE_Q3_K |
| wizardlm-7b-v1.0-superhot-8k.ggmlv3.q3_K_M.bin | q3_K_M | 3 | 3.28 GB | 5.78 GB | New k-quant method. Uses GGML_TYPE_Q4_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else GGML_TYPE_Q3_K |
| wizardlm-7b-v1.0-superhot-8k.ggmlv3.q3_K_S.bin | q3_K_S | 3 | 2.95 GB | 5.45 GB | New k-quant method. Uses GGML_TYPE_Q3_K for all tensors |
| wizardlm-7b-v1.0-superhot-8k.ggmlv3.q4_K_M.bin | q4_K_M | 4 | 4.08 GB | 6.58 GB | New k-quant method. Uses GGML_TYPE_Q6_K for half of the attention.wv and feed_forward.w2 tensors, else GGML_TYPE_Q4_K |
| wizardlm-7b-v1.0-superhot-8k.ggmlv3.q4_K_S.bin | q4_K_S | 4 | 3.83 GB | 6.33 GB | New k-quant method. Uses GGML_TYPE_Q4_K for all tensors |
| wizardlm-7b-v1.0-superhot-8k.ggmlv3.q5_K_M.bin | q5_K_M | 5 | 4.78 GB | 7.28 GB | New k-quant method. Uses GGML_TYPE_Q6_K for half of the attention.wv and feed_forward.w2 tensors, else GGML_TYPE_Q5_K |
| wizardlm-7b-v1.0-superhot-8k.ggmlv3.q5_K_S.bin | q5_K_S | 5 | 4.65 GB | 7.15 GB | New k-quant method. Uses GGML_TYPE_Q5_K for all tensors |
| wizardlm-7b-v1.0-superhot-8k.ggmlv3.q6_K.bin | q6_K | 6 | 5.53 GB | 8.03 GB | New k-quant method. Uses GGML_TYPE_Q8_K - 6-bit quantization - for all tensors |
**Note**: the above RAM figures assume no GPU offloading. If layers are offloaded to the GPU, this will reduce RAM usage and use VRAM instead.
## How to run in `koboldcpp`
On Linux I use the following command line to launch the KoboldCpp UI with OpenCL aceleration and a context size of 4096:
```
python ./koboldcpp.py --stream --unbantokens --threads 8 --usecublas --gpulayers 100 wizardlm-7b-v1.0-superhot-8k.ggmlv3.q4_K_M.bin
```
Change `--gpulayers 100` to the number of layers you want/are able to offload to the GPU. Remove it if you don't have GPU acceleration.
For OpenCL acceleration, change `--usecublas` to `--useclblast 0 0`. You may need to change the second `0` to `1` if you have both an iGPU and a discrete GPU.
<!-- footer start -->
## Discord
For further support, and discussions on these models and AI in general, join us at:
[TheBloke AI's Discord server](https://discord.gg/theblokeai)
## Thanks, and how to contribute.
Thanks to the [chirper.ai](https://chirper.ai) team!
I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.
If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.
Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
* Patreon: https://patreon.com/TheBlokeAI
* Ko-Fi: https://ko-fi.com/TheBlokeAI
**Special thanks to**: Luke from CarbonQuill, Aemon Algiz.
**Patreon special mentions**: RoA, Lone Striker, Gabriel Puliatti, Derek Yates, Randy H, Jonathan Leane, Eugene Pentland, Karl Bernard, Viktor Bowallius, senxiiz, Daniel P. Andersen, Pierre Kircher, Deep Realms, Cory Kujawski, Oscar Rangel, Fen Risland, Ajan Kanaga, LangChain4j, webtim, Nikolai Manek, Trenton Dambrowitz, Raven Klaugh, Kalila, Khalefa Al-Ahmad, Chris McCloskey, Luke @flexchar, Ai Maven, Dave, Asp the Wyvern, Sean Connelly, Imad Khwaja, Space Cruiser, Rainer Wilmers, subjectnull, Alps Aficionado, Willian Hasse, Fred von Graf, Artur Olbinski, Johann-Peter Hartmann, WelcomeToTheClub, Willem Michiel, Michael Levine, Iucharbius , Spiking Neurons AB, K, biorpg, John Villwock, Pyrater, Greatston Gnanesh, Mano Prime, Junyu Yang, Stephen Murray, John Detwiler, Luke Pendergrass, terasurfer , Pieter, zynix , Edmond Seymore, theTransient, Nathan LeClaire, vamX, Kevin Schuppel, Preetika Verma, ya boyyy, Alex , SuperWojo, Ghost , Joseph William Delisle, Matthew Berman, Talal Aujan, chris gileta, Illia Dulskyi.
Thank you to all my generous patrons and donaters!
<!-- footer end -->
# Original model card: Kaio Ken's SuperHOT 8K
### SuperHOT Prototype 2 w/ 8K Context
This is a second prototype of SuperHOT, a NSFW focused LoRA, this time 7B with 8K context and no RLHF, using the same technique described in [the github blog](https://kaiokendev.github.io/til#extending-context-to-8k).
#### Looking for Merged & Quantized Models?
Make some please :)
#### Using the monkey-patch?
You will **NEED** to **apply the monkeypatch** or, if you are already using the monkeypatch, **change the scaling factor to 0.25 and the maximum sequence length to 8192**
The monkeypatch is only necessary if you are using a front-end/back-end that does not already support scaling and said front-end/back-end is Python-based (i.e. Huggingface Transformers). To apply the patch, you will need to copy the `llama_rope_scaled_monkey_patch.py` into your working directory and call the exported function `replace_llama_rope_with_scaled_rope` at the very start of your Python program. It will modify the Transformers library's implementation of RoPE to properly apply the scaling factor.
#### Using Oobabooga with Exllama?
Switch your loader to `exllama` or `exllama_hf` Add the arguments `max_seq_len 8192` and `compress_pos_emb 4`. **While the model may work well with `compress_pos_emb 2`, it was trained on 4, so that is what I advocate for you to use**
Example in the command-line:
- `python server.py --max_seq_len 8192 --compress_pos_emb 4 --loader exllama_hf`
In the UI, you will see the loader option in the `Models` tab. Once you select either `exllama` or `exllama_hf`, the `max_seq_len` and `compress_pos_emb` settings will appear.
#### Training Details
I trained the LoRA with the following configuration:
- 1200 samples (~400 samples over 2048 sequence length)
- learning rate of 3e-4
- 3 epochs
- The exported modules are:
- q_proj
- k_proj
- v_proj
- o_proj
- no bias
- Rank = 4
- Alpha = 8
- no dropout
- weight decay of 0.1
- AdamW beta1 of 0.9 and beta2 0.99, epsilon of 1e-5
- Trained on 4-bit base model
- Cutoff length: 4096
# Original model card: Eric Hartford's WizardLM-7B-V1.0-Uncensored
This is a retraining of https://huggingface.co/WizardLM/WizardLM-7B-V1.0 with a filtered dataset, intended to reduce refusals, avoidance, and bias.
Note that LLaMA itself has inherent ethical beliefs, so there's no such thing as a "truly uncensored" model. But this model will be more compliant than WizardLM/WizardLM-7B-V1.0.
Shout out to the open source AI/ML community, and everyone who helped me out.
Note: An uncensored model has no guardrails. You are responsible for anything you do with the model, just as you are responsible for anything you do with any dangerous object such as a knife, gun, lighter, or car. Publishing anything this model generates is the same as publishing it yourself. You are responsible for the content you publish, and you cannot blame the model any more than you can blame the knife, gun, lighter, or car for what you do with it.
Unlike WizardLM/WizardLM-7B-V1.0, but like WizardLM/WizardLM-13B-V1.0 and WizardLM/WizardLM-33B-V1.0, this model is trained with Vicuna-1.1 style prompts.
```
You are a helpful AI assistant.
USER: <prompt>
ASSISTANT:
```
Thank you [chirper.ai](https://chirper.ai) for sponsoring some of my compute!
|
aroot/eng-guj-simcse_random_ssrb
|
aroot
| 2023-07-06T20:32:37Z | 3 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"mbart",
"text2text-generation",
"translation",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
translation
| 2023-07-06T20:10:52Z |
---
tags:
- translation
- generated_from_trainer
metrics:
- bleu
model-index:
- name: eng-guj-simcse_random_ssrb
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# eng-guj-simcse_random_ssrb
This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 3.2802
- Bleu: 2.8939
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3
|
sinny/ppo-LunarLander-v2
|
sinny
| 2023-07-06T20:30:19Z | 0 | 0 |
stable-baselines3
|
[
"stable-baselines3",
"LunarLander-v2",
"deep-reinforcement-learning",
"reinforcement-learning",
"model-index",
"region:us"
] |
reinforcement-learning
| 2023-07-03T10:35:57Z |
---
library_name: stable-baselines3
tags:
- LunarLander-v2
- deep-reinforcement-learning
- reinforcement-learning
- stable-baselines3
model-index:
- name: PPO
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: LunarLander-v2
type: LunarLander-v2
metrics:
- type: mean_reward
value: 256.00 +/- 13.38
name: mean_reward
verified: false
---
# **PPO** Agent playing **LunarLander-v2**
This is a trained model of a **PPO** agent playing **LunarLander-v2**
using the [stable-baselines3 library](https://github.com/DLR-RM/stable-baselines3).
## Usage (with Stable-baselines3)
TODO: Add your code
```python
from stable_baselines3 import ...
from huggingface_sb3 import load_from_hub
...
```
|
TheBloke/LongChat-7B-GGML
|
TheBloke
| 2023-07-06T20:29:59Z | 0 | 12 | null |
[
"license:other",
"region:us"
] | null | 2023-06-30T09:46:05Z |
---
inference: false
license: other
---
<!-- header start -->
<div style="width: 100%;">
<img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
</div>
<div style="display: flex; justify-content: space-between; width: 100%;">
<div style="display: flex; flex-direction: column; align-items: flex-start;">
<p><a href="https://discord.gg/theblokeai">Chat & support: my new Discord server</a></p>
</div>
<div style="display: flex; flex-direction: column; align-items: flex-end;">
<p><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p>
</div>
</div>
<!-- header end -->
# LmSys' Long Chat 7B GGML
These files are GGML format model files for [LmSys' Long Chat 7B](https://huggingface.co/lmsys/longchat-7b-16k).
These are RoPE GGMLs with an increased context length. RoPE - Rotated Positional Encoding - expands context beyond what was originally possible for a model. It was discovered and developed by [kaiokendev](https://huggingface.co/kaiokendev).
In order to use the increased context length, you can presently use:
* [KoboldCpp](https://github.com/LostRuins/koboldcpp) - [release 1.33](https://github.com/LostRuins/koboldcpp/releases/tag/v1.33) or later.
Support is also expected to come to llama.cpp, however work is still being done to find the optimal implementation.
To use the increased context with KoboldCpp, use `--contextsize` to set the desired context, eg `--contextsize 4096` or `--contextsize 8192` or `--contextsize 16384`.
**NOTE 1**: Currently RoPE models can _only_ be used at a context size greater than 2048. At 2048 it will produce gibberish. Please make sure you're always setting `--contextsize` and specifying a value higher than 2048, eg 3072, 4096, etc.
**NOTE 2**: Increased context length is an area seeing rapid developments and improvements. It is quite possible that these models may be superseded by new developments in the coming days. If that's the case, I will remove them, or update this README as appropriate.
## Repositories available
* [4-bit GPTQ models for GPU inference](https://huggingface.co/TheBloke/LongChat-7B-GPTQ)
* [2, 3, 4, 5, 6 and 8-bit GGML models for CPU+GPU inference](https://huggingface.co/TheBloke/LongChat-7B-GGML)
* [Unquantised fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/lmsys/longchat-7b-16k)
## Prompt template
```
A chat between a curious user and an assistant. The assistant gives helpful, detailed, accurate, uncensored responses to the user's input
USER: prompt
ASSISTANT:
```
<!-- compatibility_ggml start -->
## Compatibility
These GGMLs will work with any llama.cpp-compatible GGML client that supports k-quants.
However the increased context length won't work without specific support. See the note in the introduction for details on using increased context.
## Explanation of the new k-quant methods
The new methods available are:
* GGML_TYPE_Q2_K - "type-1" 2-bit quantization in super-blocks containing 16 blocks, each block having 16 weight. Block scales and mins are quantized with 4 bits. This ends up effectively using 2.5625 bits per weight (bpw)
* GGML_TYPE_Q3_K - "type-0" 3-bit quantization in super-blocks containing 16 blocks, each block having 16 weights. Scales are quantized with 6 bits. This end up using 3.4375 bpw.
* GGML_TYPE_Q4_K - "type-1" 4-bit quantization in super-blocks containing 8 blocks, each block having 32 weights. Scales and mins are quantized with 6 bits. This ends up using 4.5 bpw.
* GGML_TYPE_Q5_K - "type-1" 5-bit quantization. Same super-block structure as GGML_TYPE_Q4_K resulting in 5.5 bpw
* GGML_TYPE_Q6_K - "type-0" 6-bit quantization. Super-blocks with 16 blocks, each block having 16 weights. Scales are quantized with 8 bits. This ends up using 6.5625 bpw
* GGML_TYPE_Q8_K - "type-0" 8-bit quantization. Only used for quantizing intermediate results. The difference to the existing Q8_0 is that the block size is 256. All 2-6 bit dot produc
ts are implemented for this quantization type.
Refer to the Provided Files table below to see what files use which methods, and how.
<!-- compatibility_ggml end -->
## Provided files
| Name | Quant method | Bits | Size | Max RAM required | Use case |
| ---- | ---- | ---- | ---- | ---- | ----- |
| longchat-7b-16k.ggmlv3.q2_K.bin | q2_K | 2 | 2.87 GB | 5.37 GB | New k-quant method. Uses GGML_TYPE_Q4_K for the attention.vw and feed_forward.w2 tensors, GGML_TYPE_Q2_K for the other tensors. |
| longchat-7b-16k.ggmlv3.q3_K_L.bin | q3_K_L | 3 | 3.60 GB | 6.10 GB | New k-quant method. Uses GGML_TYPE_Q5_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else GGML_TYPE_Q3_K |
| longchat-7b-16k.ggmlv3.q3_K_M.bin | q3_K_M | 3 | 3.28 GB | 5.78 GB | New k-quant method. Uses GGML_TYPE_Q4_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else GGML_TYPE_Q3_K |
| longchat-7b-16k.ggmlv3.q3_K_S.bin | q3_K_S | 3 | 2.95 GB | 5.45 GB | New k-quant method. Uses GGML_TYPE_Q3_K for all tensors |
| longchat-7b-16k.ggmlv3.q4_K_M.bin | q4_K_M | 4 | 4.08 GB | 6.58 GB | New k-quant method. Uses GGML_TYPE_Q6_K for half of the attention.wv and feed_forward.w2 tensors, else GGML_TYPE_Q4_K |
| longchat-7b-16k.ggmlv3.q4_K_S.bin | q4_K_S | 4 | 3.83 GB | 6.33 GB | New k-quant method. Uses GGML_TYPE_Q4_K for all tensors |
| longchat-7b-16k.ggmlv3.q5_K_M.bin | q5_K_M | 5 | 4.78 GB | 7.28 GB | New k-quant method. Uses GGML_TYPE_Q6_K for half of the attention.wv and feed_forward.w2 tensors, else GGML_TYPE_Q5_K |
| longchat-7b-16k.ggmlv3.q5_K_S.bin | q5_K_S | 5 | 4.65 GB | 7.15 GB | New k-quant method. Uses GGML_TYPE_Q5_K for all tensors |
| longchat-7b-16k.ggmlv3.q6_K.bin | q6_K | 6 | 5.53 GB | 8.03 GB | New k-quant method. Uses GGML_TYPE_Q8_K - 6-bit quantization - for all tensors |
**Note**: the above RAM figures assume no GPU offloading. If layers are offloaded to the GPU, this will reduce RAM usage and use VRAM instead.
## How to run in `koboldcpp`
On Linux I use the following command line to launch the KoboldCpp UI with OpenCL aceleration and a context size of 4096:
```
python ./koboldcpp.py --contextsize 4096 --stream --unbantokens --threads 8 --usecublas --gpulayers 100 longchat-7b-16k.ggmlv3.q4_K_M.bin
```
Change `--contextsize` to the context size you want - **it must be higher than 2048 else the model will produce gibberish**
Change `--gpulayers 100` to the number of layers you want/are able to offload to the GPU. Remove it if you don't have GPU acceleration.
For OpenCL acceleration, change `--usecublas` to `--useclblast 0 0`. You may need to change the second `0` to `1` if you have both an iGPU and a discrete GPU.
<!-- footer start -->
## Discord
For further support, and discussions on these models and AI in general, join us at:
[TheBloke AI's Discord server](https://discord.gg/theblokeai)
## Thanks, and how to contribute.
Thanks to the [chirper.ai](https://chirper.ai) team!
I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.
If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.
Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
* Patreon: https://patreon.com/TheBlokeAI
* Ko-Fi: https://ko-fi.com/TheBlokeAI
**Special thanks to**: Luke from CarbonQuill, Aemon Algiz, Dmitriy Samsonov.
**Patreon special mentions**: Pyrater, WelcomeToTheClub, Kalila, Mano Prime, Trenton Dambrowitz, Spiking Neurons AB, Pierre Kircher, Fen Risland, Kevin Schuppel, Luke, Rainer Wilmers, vamX, Gabriel Puliatti, Alex , Karl Bernard, Ajan Kanaga, Talal Aujan, Space Cruiser, ya boyyy, biorpg, Johann-Peter Hartmann, Asp the Wyvern, Ai Maven, Ghost , Preetika Verma, Nikolai Manek, trip7s trip, John Detwiler, Fred von Graf, Artur Olbinski, subjectnull, John Villwock, Junyu Yang, Rod A, Lone Striker, Chris McCloskey, Iucharbius , Matthew Berman, Illia Dulskyi, Khalefa Al-Ahmad, Imad Khwaja, chris gileta, Willem Michiel, Greatston Gnanesh, Derek Yates, K, Alps Aficionado, Oscar Rangel, David Flickinger, Luke Pendergrass, Deep Realms, Eugene Pentland, Cory Kujawski, terasurfer , Jonathan Leane, senxiiz, Joseph William Delisle, Sean Connelly, webtim, zynix , Nathan LeClaire.
Thank you to all my generous patrons and donaters!
<!-- footer end -->
# Original model card: LmSys' Long Chat 7B
# longchat-7b-16k Model Card
## Model details
**Model type:**
longchat-7b-16k is an open-source chatbot trained by fine-tuning llama-7b on user-shared conversations collected from ShareGPT, using the condensing rotary embedding technique reported in the [blog](https://lmsys.org/blog/2023-06-29-longchat).
**Model date:**
longchat-7b-16k was trained on June 2023.
**Organizations developing the model:**
The LongChat developers: Dacheng Li*, Rulin Shao*, Anze Xie, Ying Sheng, Lianmin Zheng, Ion Stoica, Xuezhe Ma, and Hao Zhang
**Paper or resources for more information:**
https://github.com/DachengLi1/LongChat
**Where to send questions or comments about the model:**
https://github.com/DachengLi1/LongChat
## Intended use
**Primary intended uses:**
The primary use of longchat-7b-16k is for research purposes.
**Primary intended users:**
The primary intended users of the model are researchers in natural language processing, machine learning, and artificial intelligence.
## Training dataset
18K conversations collected from ShareGPT.com.
## Evaluation dataset
A preliminary evaluation of the model quality is conducted by our released [LongEval](https://github.com/DachengLi1/LongChat).
|
cjohlmacher/unit2-taxi-2
|
cjohlmacher
| 2023-07-06T20:20:42Z | 0 | 0 | null |
[
"Taxi-v3",
"q-learning",
"reinforcement-learning",
"custom-implementation",
"model-index",
"region:us"
] |
reinforcement-learning
| 2023-07-06T20:18:26Z |
---
tags:
- Taxi-v3
- q-learning
- reinforcement-learning
- custom-implementation
model-index:
- name: unit2-taxi-2
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: Taxi-v3
type: Taxi-v3
metrics:
- type: mean_reward
value: 7.54 +/- 2.74
name: mean_reward
verified: false
---
# **Q-Learning** Agent playing1 **Taxi-v3**
This is a trained model of a **Q-Learning** agent playing **Taxi-v3** .
## Usage
```python
model = load_from_hub(repo_id="cjohlmacher/unit2-taxi-2", filename="q-learning.pkl")
# Don't forget to check if you need to add additional attributes (is_slippery=False etc)
env = gym.make(model["env_id"])
```
|
PhysHunter/bert-finetuned-ner
|
PhysHunter
| 2023-07-06T20:15:38Z | 105 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"bert",
"token-classification",
"generated_from_trainer",
"dataset:conll2003",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
token-classification
| 2023-07-06T14:17:03Z |
---
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- conll2003
metrics:
- precision
- recall
- f1
- accuracy
model-index:
- name: bert-finetuned-ner
results:
- task:
name: Token Classification
type: token-classification
dataset:
name: conll2003
type: conll2003
config: conll2003
split: validation
args: conll2003
metrics:
- name: Precision
type: precision
value: 0.928983358049102
- name: Recall
type: recall
value: 0.9488387748232918
- name: F1
type: f1
value: 0.9388060944134544
- name: Accuracy
type: accuracy
value: 0.9858568316948254
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# bert-finetuned-ner
This model is a fine-tuned version of [bert-base-cased](https://huggingface.co/bert-base-cased) on the conll2003 dataset.
It achieves the following results on the evaluation set:
- Loss: 0.0658
- Precision: 0.9290
- Recall: 0.9488
- F1: 0.9388
- Accuracy: 0.9859
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
### Training results
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:---------:|:------:|:------:|:--------:|
| 0.0863 | 1.0 | 1756 | 0.0697 | 0.9110 | 0.9317 | 0.9212 | 0.9815 |
| 0.0327 | 2.0 | 3512 | 0.0690 | 0.9297 | 0.9482 | 0.9388 | 0.9858 |
| 0.0164 | 3.0 | 5268 | 0.0658 | 0.9290 | 0.9488 | 0.9388 | 0.9859 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu118
- Datasets 2.13.1
- Tokenizers 0.13.3
|
aroot/eng-fra-simcse_central_ssrb
|
aroot
| 2023-07-06T20:12:02Z | 3 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"mbart",
"text2text-generation",
"translation",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
translation
| 2023-07-06T19:47:43Z |
---
tags:
- translation
- generated_from_trainer
metrics:
- bleu
model-index:
- name: eng-fra-simcse_central_ssrb
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# eng-fra-simcse_central_ssrb
This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 1.1471
- Bleu: 31.8498
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3
|
YakovElm/Apache_15_BERT_More_Properties
|
YakovElm
| 2023-07-06T20:11:25Z | 61 | 0 |
transformers
|
[
"transformers",
"tf",
"bert",
"text-classification",
"generated_from_keras_callback",
"license:apache-2.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2023-07-06T20:10:50Z |
---
license: apache-2.0
tags:
- generated_from_keras_callback
model-index:
- name: Apache_15_BERT_More_Properties
results: []
---
<!-- This model card has been generated automatically according to the information Keras had access to. You should
probably proofread and complete it, then remove this comment. -->
# Apache_15_BERT_More_Properties
This model is a fine-tuned version of [bert-base-uncased](https://huggingface.co/bert-base-uncased) on an unknown dataset.
It achieves the following results on the evaluation set:
- Train Loss: 0.1866
- Train Accuracy: 0.9542
- Validation Loss: 0.3421
- Validation Accuracy: 0.8924
- Epoch: 2
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- optimizer: {'name': 'Adam', 'weight_decay': None, 'clipnorm': 1.0, 'global_clipnorm': None, 'clipvalue': None, 'use_ema': False, 'ema_momentum': 0.99, 'ema_overwrite_frequency': None, 'jit_compile': False, 'is_legacy_optimizer': False, 'learning_rate': 3e-05, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-08, 'amsgrad': False}
- training_precision: float32
### Training results
| Train Loss | Train Accuracy | Validation Loss | Validation Accuracy | Epoch |
|:----------:|:--------------:|:---------------:|:-------------------:|:-----:|
| 0.1956 | 0.9535 | 0.3893 | 0.8924 | 0 |
| 0.1874 | 0.9542 | 0.3804 | 0.8924 | 1 |
| 0.1866 | 0.9542 | 0.3421 | 0.8924 | 2 |
### Framework versions
- Transformers 4.29.2
- TensorFlow 2.12.0
- Datasets 2.12.0
- Tokenizers 0.13.3
|
igoroliveira/distilbert-base-uncased-finetuned-cola
|
igoroliveira
| 2023-07-06T20:09:07Z | 61 | 0 |
transformers
|
[
"transformers",
"tf",
"tensorboard",
"distilbert",
"text-classification",
"generated_from_keras_callback",
"license:apache-2.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2023-07-06T19:11:37Z |
---
license: apache-2.0
tags:
- generated_from_keras_callback
model-index:
- name: igoroliveira/distilbert-base-uncased-finetuned-cola
results: []
---
<!-- This model card has been generated automatically according to the information Keras had access to. You should
probably proofread and complete it, then remove this comment. -->
# igoroliveira/distilbert-base-uncased-finetuned-cola
This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on an unknown dataset.
It achieves the following results on the evaluation set:
- Train Loss: 0.1959
- Validation Loss: 0.5357
- Train Matthews Correlation: 0.5177
- Epoch: 2
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- optimizer: {'name': 'Adam', 'weight_decay': None, 'clipnorm': None, 'global_clipnorm': None, 'clipvalue': None, 'use_ema': False, 'ema_momentum': 0.99, 'ema_overwrite_frequency': None, 'jit_compile': False, 'is_legacy_optimizer': False, 'learning_rate': {'class_name': 'PolynomialDecay', 'config': {'initial_learning_rate': 2e-05, 'decay_steps': 1602, 'end_learning_rate': 0.0, 'power': 1.0, 'cycle': False, 'name': None}}, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-08, 'amsgrad': False}
- training_precision: float32
### Training results
| Train Loss | Validation Loss | Train Matthews Correlation | Epoch |
|:----------:|:---------------:|:--------------------------:|:-----:|
| 0.5247 | 0.4570 | 0.4887 | 0 |
| 0.3259 | 0.4597 | 0.5101 | 1 |
| 0.1959 | 0.5357 | 0.5177 | 2 |
### Framework versions
- Transformers 4.30.2
- TensorFlow 2.12.0
- Datasets 2.13.1
- Tokenizers 0.13.3
|
manosp/textual_inversion_moto
|
manosp
| 2023-07-06T20:08:19Z | 30 | 0 |
diffusers
|
[
"diffusers",
"tensorboard",
"stable-diffusion",
"stable-diffusion-diffusers",
"text-to-image",
"textual_inversion",
"base_model:runwayml/stable-diffusion-v1-5",
"base_model:adapter:runwayml/stable-diffusion-v1-5",
"license:creativeml-openrail-m",
"autotrain_compatible",
"endpoints_compatible",
"diffusers:StableDiffusionPipeline",
"region:us"
] |
text-to-image
| 2023-07-06T18:45:41Z |
---
license: creativeml-openrail-m
base_model: runwayml/stable-diffusion-v1-5
tags:
- stable-diffusion
- stable-diffusion-diffusers
- text-to-image
- diffusers
- textual_inversion
inference: true
---
# Textual inversion text2image fine-tuning - manosp/textual_inversion_moto
These are textual inversion adaption weights for runwayml/stable-diffusion-v1-5. You can find some example images in the following.
|
andkelly21/t5-small-finetuned-pubmed
|
andkelly21
| 2023-07-06T19:58:17Z | 115 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"t5",
"text2text-generation",
"summarization",
"generated_from_trainer",
"license:apache-2.0",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] |
summarization
| 2022-12-25T19:26:57Z |
---
widget:
- text: "Patient A, a 67-year-old male with a history of hypertension and obesity, received his first dose of the Pfizer COVID-19 vaccine on January 5th, 2022. He reported no adverse reactions following the vaccine and was discharged home. However, two days later, he presented to the emergency department with complaints of chest pain, shortness of breath, and cough. He was found to have an elevated troponin level and was diagnosed with an acute myocardial infarction (AMI) as his primary diagnosis. The cause of death was determined to be due to complications of the AMI, which led to cardiogenic shock and subsequent multi-organ failure. Secondary diagnoses included acute respiratory distress syndrome (ARDS) and acute renal failure. Symptoms included chest pain, shortness of breath, cough, and hypotension. Rule out diagnoses included COVID-19 infection and pulmonary embolism. The patient had a medical history of hypertension, obesity, and hyperlipidemia. There was no significant family history. The patient was treated with thrombolytic therapy and mechanical ventilation but unfortunately, he succumbed to his illness and passed away on January 13th, 2022. The Pfizer COVID-19 vaccine was noted as part of his medical history. The case was reported to the Vaccine Adverse Event Reporting System (VAERS) for further investigation."
example_title: "Medical Case"
license: apache-2.0
tags:
- summarization
- generated_from_trainer
metrics:
- rouge
model-index:
- name: t5-small-finetuned-pubmed
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# t5-small-finetuned-pubmed
This model is a fine-tuned version of [t5-small](https://huggingface.co/t5-small) on a truncated [PubMed Summarization](https://huggingface.co/datasets/ccdv/pubmed-summarization) dataset.
It achieves the following results on the evaluation set:
- Loss: 2.7252
- Rouge1: 19.4457
- Rouge2: 3.125
- Rougel: 18.3168
- Rougelsum: 18.5625
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5.6e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 8
### Training results
| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum |
|:-------------:|:-----:|:----:|:---------------:|:-------:|:------:|:-------:|:---------:|
| 3.2735 | 1.0 | 13 | 2.9820 | 18.745 | 3.7918 | 15.7876 | 15.8512 |
| 3.0428 | 2.0 | 26 | 2.8828 | 17.953 | 2.5 | 15.49 | 15.468 |
| 2.6259 | 3.0 | 39 | 2.8283 | 21.5532 | 5.9278 | 19.7523 | 19.9232 |
| 3.0795 | 4.0 | 52 | 2.7910 | 20.9244 | 5.9278 | 19.8685 | 20.0181 |
| 2.8276 | 5.0 | 65 | 2.7613 | 20.6403 | 3.125 | 18.0574 | 18.2227 |
| 2.64 | 6.0 | 78 | 2.7404 | 19.4457 | 3.125 | 18.3168 | 18.5625 |
| 2.5525 | 7.0 | 91 | 2.7286 | 19.4457 | 3.125 | 18.3168 | 18.5625 |
| 2.4951 | 8.0 | 104 | 2.7252 | 19.4457 | 3.125 | 18.3168 | 18.5625 |
### Framework versions
- Transformers 4.25.1
- Pytorch 1.13.0
- Datasets 2.8.0
- Tokenizers 0.13.2
|
ericNguyen0132/roberta-large-Dep-first
|
ericNguyen0132
| 2023-07-06T19:55:21Z | 5 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"roberta",
"text-classification",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2023-07-04T11:41:34Z |
---
tags:
- generated_from_trainer
metrics:
- accuracy
- f1
model-index:
- name: roberta-large-Dep-first
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# roberta-large-Dep-first
This model is a fine-tuned version of [rafalposwiata/deproberta-large-depression](https://huggingface.co/rafalposwiata/deproberta-large-depression) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 1.1795
- Accuracy: 0.702
- F1: 0.5706
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-06
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:------:|
| 0.5563 | 1.0 | 751 | 0.5324 | 0.756 | 0.6188 |
| 0.4721 | 2.0 | 1502 | 0.6204 | 0.691 | 0.5874 |
| 0.3836 | 3.0 | 2253 | 0.7990 | 0.696 | 0.525 |
| 0.3245 | 4.0 | 3004 | 0.9714 | 0.694 | 0.5726 |
| 0.2795 | 5.0 | 3755 | 1.1795 | 0.702 | 0.5706 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu118
- Datasets 2.13.1
- Tokenizers 0.13.3
|
nkpz/Lawyer-Vicuna-200-gptq-32g
|
nkpz
| 2023-07-06T19:53:24Z | 5 | 1 |
transformers
|
[
"transformers",
"pytorch",
"llama",
"text-generation",
"license:other",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] |
text-generation
| 2023-07-06T19:35:31Z |
---
license: other
---
4-bit (32 groupsize) quantized files for [Devden/Lawyer-Vicuna-200](https://huggingface.co/Devden/Lawyer-Vicuna-200)
Quantized using [GPTQ-for-LLaMa](https://github.com/qwopqwop200/GPTQ-for-LLaMa).
Command used to quantize: python llama.py /my/model/directory c4 --wbits 4 --true-sequential --act-order --groupsize 32 --save_safetensors /my/output/file.safetensors
|
wesleyacheng/angry-birds-classifier
|
wesleyacheng
| 2023-07-06T19:34:39Z | 114 | 0 |
transformers
|
[
"transformers",
"pytorch",
"safetensors",
"distilbert",
"text-classification",
"en",
"dataset:tweet_eval",
"license:apache-2.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2023-05-30T06:11:14Z |
---
license: apache-2.0
datasets:
- tweet_eval
language:
- en
metrics:
- accuracy
- f1
pipeline_tag: text-classification
widget:
- text: I hate you
example_title: Angry Example
- text: I love you
example_title: Not Angry Example
---
First posted in my [Kaggle](https://www.kaggle.com/code/wesleyacheng/angry-birds-classifier).
I love the **Angry Birds** game! I used to play it day and night nonstop!
I made an 😡🐦 **ANGRY BIRDS Classifier** to classify **ANGRY Tweets**!
Here, I used the [Twitter Emotion Dataset](https://huggingface.co/datasets/tweet_eval) and [BERT](https://huggingface.co/docs/transformers/model_doc/bert) using [Transfer Learning](https://en.wikipedia.org/wiki/Transfer_learning) in native [PyTorch](https://en.wikipedia.org/wiki/PyTorch).
|
andressrg/textual_inversion_meal
|
andressrg
| 2023-07-06T19:28:50Z | 16 | 0 |
diffusers
|
[
"diffusers",
"tensorboard",
"stable-diffusion",
"stable-diffusion-diffusers",
"text-to-image",
"textual_inversion",
"base_model:runwayml/stable-diffusion-v1-5",
"base_model:adapter:runwayml/stable-diffusion-v1-5",
"license:creativeml-openrail-m",
"autotrain_compatible",
"endpoints_compatible",
"diffusers:StableDiffusionPipeline",
"region:us"
] |
text-to-image
| 2023-07-06T18:45:40Z |
---
license: creativeml-openrail-m
base_model: runwayml/stable-diffusion-v1-5
tags:
- stable-diffusion
- stable-diffusion-diffusers
- text-to-image
- diffusers
- textual_inversion
inference: true
---
# Textual inversion text2image fine-tuning - andressrg/textual_inversion_meal
These are textual inversion adaption weights for runwayml/stable-diffusion-v1-5. You can find some example images in the following.
|
RogerB/afriberta_small-finetuned-kintweetsB
|
RogerB
| 2023-07-06T19:25:00Z | 103 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"xlm-roberta",
"fill-mask",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
fill-mask
| 2023-07-06T19:16:16Z |
---
tags:
- generated_from_trainer
model-index:
- name: afriberta_small-finetuned-kintweetsB
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# afriberta_small-finetuned-kintweetsB
This model is a fine-tuned version of [castorini/afriberta_small](https://huggingface.co/castorini/afriberta_small) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 3.1849
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 10
- eval_batch_size: 10
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3.0
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 3.6414 | 1.0 | 900 | 3.3043 |
| 3.3705 | 2.0 | 1800 | 3.1980 |
| 3.3101 | 3.0 | 2700 | 3.1867 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu118
- Datasets 2.13.1
- Tokenizers 0.13.3
|
Joemgu/pegasus-x-sumstew
|
Joemgu
| 2023-07-06T19:22:06Z | 99 | 1 |
transformers
|
[
"transformers",
"pytorch",
"pegasus_x",
"text2text-generation",
"summarization",
"en",
"dataset:samsum",
"dataset:cnn_dailymail",
"dataset:kmfoda/booksum",
"dataset:pszemraj/scientific_lay_summarisation-plos-norm",
"license:apache-2.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
summarization
| 2023-05-02T15:24:50Z |
---
license: apache-2.0
datasets:
- samsum
- cnn_dailymail
- kmfoda/booksum
- pszemraj/scientific_lay_summarisation-plos-norm
language:
- en
pipeline_tag: summarization
---
# Pegasus-x-sumstew
## Model description
This model is a fine-tuned version of the Pegasus-x-large model on a filtered subset of a mixture of CNN-Dailymail, Samsum, Booksum and Laysum datasets. It can generate abstractive summaries of long texts.
## Intended uses & limitations
This model can be used for summarizing long texts in English, such as academic transcripts, meeting minutes, or literature. It is not intended for summarizing short texts, such as tweets, headlines, or captions. The model may produce inaccurate or biased summaries if the input text contains factual errors, slang, or offensive language.
## How to use
You can use this model with the `pipeline` function from the `transformers` library:
```python
from transformers import pipeline
summarizer = pipeline("summarization", "joemgu/pegasus-x-sumstew")
text = "Alice was beginning to get very tired of sitting by her sister on the bank, and of having nothing to do: once or twice she had peeped into the book her sister was reading, but it had no pictures or conversations in it, 'and what is the use of a book,' thought Alice 'without pictures or conversations?' So she was considering in her own mind (as well as she could, for the hot day made her feel very sleepy and stupid), whether the pleasure of making a daisy-chain would be worth the trouble of getting up and picking the daisies, when suddenly a White Rabbit with pink eyes ran close by her. There was nothing so very remarkable in that; nor did Alice think it so very much out of the way to hear the Rabbit say to itself, 'Oh dear! Oh dear! I shall be late!' (when she thought it over afterwards, it occurred to her that she ought to have wondered at this, but at the time it all seemed quite natural); but when the Rabbit actually took a watch out of its waistcoat-pocket, and looked at it, and then hurried on, Alice started to her feet, for it flashed across her mind that she had never before seen a rabbit with either a waistcoat-pocket, or a watch to take out of it, and burning with curiosity, she ran across the field after it, and fortunately was just in time to see it pop down a large rabbit-hole under the hedge. In another moment down went Alice after it, never once considering how in the world she was to get out again."
summary = summarizer(text,
num_beams=8,
repetition_penalty=3.5,
no_repeat_ngram_size=4,
encoder_no_repeat_ngram_size=4
)[0]["summary_text"]
print(summary)
```
Output:
```text
Alice is a bored and curious girl who follows a White Rabbit with a watch into a rabbit-hole. She enters a strange world where she has many adventures and meets many peculiar creatures.
```
## Training data
The model was fine-tuned on a filtered subset of a mixture of CNN-Dailymail, Samsum, Booksum and Laysum datasets. These datasets contain various types of texts and their abstractive summaries. The subset was selected to include only texts that are longer than 1000 words and have summaries that are shorter than 100 words. The total size of the subset is about 150k examples.
## Evaluation results
TODO
## Limitations and bias
The model may have inherited some limitations and biases from the pre-trained Pegasus-x-large model and the fine-tuning datasets. Some possible sources of bias are:
- The pre-trained Pegasus-x-large model was trained on a large corpus of English texts from various sources, which may not reflect the diversity and nuances of different languages and cultures.
- The fine-tuning datasets were collected from different domains and genres, which may have their own stylistic conventions and perspectives on certain topics and events.
- The fine-tuning datasets only contain abstractive summaries, which may not capture all the important information and nuances of the original texts.
- The fine-tuning datasets only cover texts from certain time periods and sources, which may not reflect the current state of affairs and trends.
Therefore, users should be aware of these limitations and biases when using this model and evaluate its performance and suitability for their specific use cases.
|
bssr138/distilhubert-finetuned-gtzan
|
bssr138
| 2023-07-06T19:16:25Z | 160 | 0 |
transformers
|
[
"transformers",
"pytorch",
"hubert",
"audio-classification",
"generated_from_trainer",
"dataset:marsyas/gtzan",
"base_model:ntu-spml/distilhubert",
"base_model:finetune:ntu-spml/distilhubert",
"license:apache-2.0",
"model-index",
"endpoints_compatible",
"region:us"
] |
audio-classification
| 2023-07-06T19:12:26Z |
---
license: apache-2.0
base_model: ntu-spml/distilhubert
tags:
- generated_from_trainer
datasets:
- marsyas/gtzan
metrics:
- accuracy
model-index:
- name: distilhubert-finetuned-gtzan
results:
- task:
name: Audio Classification
type: audio-classification
dataset:
name: GTZAN
type: marsyas/gtzan
config: all
split: train
args: all
metrics:
- name: Accuracy
type: accuracy
value: 0.82
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# distilhubert-finetuned-gtzan
This model is a fine-tuned version of [ntu-spml/distilhubert](https://huggingface.co/ntu-spml/distilhubert) on the GTZAN dataset.
It achieves the following results on the evaluation set:
- Loss: 0.5911
- Accuracy: 0.82
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 10
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| 1.9683 | 1.0 | 113 | 1.7941 | 0.47 |
| 1.2944 | 2.0 | 226 | 1.1781 | 0.65 |
| 1.031 | 3.0 | 339 | 0.9032 | 0.77 |
| 0.8101 | 4.0 | 452 | 0.7538 | 0.78 |
| 0.6646 | 5.0 | 565 | 0.6414 | 0.83 |
| 0.4015 | 6.0 | 678 | 0.6811 | 0.8 |
| 0.5056 | 7.0 | 791 | 0.5735 | 0.85 |
| 0.172 | 8.0 | 904 | 0.5621 | 0.83 |
| 0.3555 | 9.0 | 1017 | 0.5750 | 0.83 |
| 0.1488 | 10.0 | 1130 | 0.5911 | 0.82 |
### Framework versions
- Transformers 4.31.0.dev0
- Pytorch 2.0.1+cu118
- Datasets 2.13.1
- Tokenizers 0.13.3
|
RogerB/afriberta_base-finetuned-kintweetsB
|
RogerB
| 2023-07-06T19:15:03Z | 103 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"xlm-roberta",
"fill-mask",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
fill-mask
| 2023-07-06T19:03:37Z |
---
tags:
- generated_from_trainer
model-index:
- name: afriberta_base-finetuned-kintweetsB
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# afriberta_base-finetuned-kintweetsB
This model is a fine-tuned version of [castorini/afriberta_base](https://huggingface.co/castorini/afriberta_base) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 3.0271
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 10
- eval_batch_size: 10
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3.0
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 3.4687 | 1.0 | 900 | 3.1593 |
| 3.1846 | 2.0 | 1800 | 3.0525 |
| 3.1214 | 3.0 | 2700 | 3.0207 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu118
- Datasets 2.13.1
- Tokenizers 0.13.3
|
richardr1126/sql-guanaco-13b-merged
|
richardr1126
| 2023-07-06T19:09:08Z | 5 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"llama",
"text-generation",
"LoRA",
"QLoRa",
"Merged LoRA Model",
"dataset:richardr1126/sql-create-context_guanaco_style",
"arxiv:2305.14314",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] |
text-generation
| 2023-06-16T06:46:15Z |
---
tags:
- LoRA
- QLoRa
- Merged LoRA Model
model-index:
- name: sql-guanaco-13b-merged
results: []
datasets:
- richardr1126/sql-create-context_guanaco_style
spaces:
- richardr1126/sql-guanaco-13b-demo
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# sql-guanaco-13b-merged
- This is a merged LoRA model that can be used with AutoModelForCausalLM or LlamaModelForCausalLM.
- It is a combination of [richardr1126/guanaco-13b-merged](https://huggingface.co/richardr1126/guanaco-13b-merged) + [richardr1126/lora-sql-guanaco-13b-adapter](https://huggingface.co/richardr1126/lora-sql-guanaco-13b-adapter).
- This LoRA was fine-tuned using QLoRA techniques on the [richardr1126/sql-create-context_guanaco_style](https://huggingface.co/datasets/richardr1126/sql-create-context_guanaco_style) dataset.
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0002
- train_batch_size: 4
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.03
- training_steps: 1875
- mixed_precision_training: Native AMP
### Framework versions
- Transformers 4.30.0.dev0
- Pytorch 2.0.1+cu118
- Datasets 2.13.0
- Tokenizers 0.13.3
## Citation
```bibtex
@article{dettmers2023qlora,
title={QLoRA: Efficient Finetuning of Quantized LLMs},
author={Dettmers, Tim and Pagnoni, Artidoro and Holtzman, Ari and Zettlemoyer, Luke},
journal={arXiv preprint arXiv:2305.14314},
year={2023}
}
```
|
KevinQuijano/model-dreambooth-chair-2.0
|
KevinQuijano
| 2023-07-06T19:07:15Z | 29 | 0 |
diffusers
|
[
"diffusers",
"tensorboard",
"stable-diffusion",
"stable-diffusion-diffusers",
"text-to-image",
"dreambooth",
"base_model:CompVis/stable-diffusion-v1-4",
"base_model:finetune:CompVis/stable-diffusion-v1-4",
"license:creativeml-openrail-m",
"autotrain_compatible",
"endpoints_compatible",
"diffusers:StableDiffusionPipeline",
"region:us"
] |
text-to-image
| 2023-07-06T18:13:41Z |
---
license: creativeml-openrail-m
base_model: CompVis/stable-diffusion-v1-4
instance_prompt: a sennagamer chair
tags:
- stable-diffusion
- stable-diffusion-diffusers
- text-to-image
- diffusers
- dreambooth
inference: true
---
# DreamBooth - KevinQuijano/model-dreambooth-chair-2.0
This is a dreambooth model derived from CompVis/stable-diffusion-v1-4. The weights were trained on a sennagamer chair using [DreamBooth](https://dreambooth.github.io/).
You can find some example images in the following.
DreamBooth for the text encoder was enabled: False.
|
kimnguyenwork/ppo-LunarLander-v2
|
kimnguyenwork
| 2023-07-06T19:07:15Z | 0 | 0 |
stable-baselines3
|
[
"stable-baselines3",
"LunarLander-v2",
"deep-reinforcement-learning",
"reinforcement-learning",
"model-index",
"region:us"
] |
reinforcement-learning
| 2023-07-06T19:06:55Z |
---
library_name: stable-baselines3
tags:
- LunarLander-v2
- deep-reinforcement-learning
- reinforcement-learning
- stable-baselines3
model-index:
- name: PPO
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: LunarLander-v2
type: LunarLander-v2
metrics:
- type: mean_reward
value: 260.51 +/- 17.48
name: mean_reward
verified: false
---
# **PPO** Agent playing **LunarLander-v2**
This is a trained model of a **PPO** agent playing **LunarLander-v2**
using the [stable-baselines3 library](https://github.com/DLR-RM/stable-baselines3).
## Usage (with Stable-baselines3)
TODO: Add your code
```python
from stable_baselines3 import ...
from huggingface_sb3 import load_from_hub
...
```
|
BigBri/2_my_awesome_eli5_clm-model
|
BigBri
| 2023-07-06T19:06:15Z | 130 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"gpt2",
"text-generation",
"generated_from_trainer",
"license:apache-2.0",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] |
text-generation
| 2023-07-06T18:34:13Z |
---
license: apache-2.0
tags:
- generated_from_trainer
model-index:
- name: 2_my_awesome_eli5_clm-model
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# 2_my_awesome_eli5_clm-model
This model is a fine-tuned version of [distilgpt2](https://huggingface.co/distilgpt2) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 3.7367
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3.0
### Training results
| Training Loss | Epoch | Step | Validation Loss |
|:-------------:|:-----:|:----:|:---------------:|
| 3.8718 | 1.0 | 1133 | 3.7563 |
| 3.7741 | 2.0 | 2266 | 3.7410 |
| 3.7327 | 3.0 | 3399 | 3.7367 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1
- Datasets 2.13.1
- Tokenizers 0.13.3
|
chh6/ppo-LunarLander-v2
|
chh6
| 2023-07-06T19:03:58Z | 0 | 0 |
stable-baselines3
|
[
"stable-baselines3",
"LunarLander-v2",
"deep-reinforcement-learning",
"reinforcement-learning",
"model-index",
"region:us"
] |
reinforcement-learning
| 2023-07-06T18:53:07Z |
---
library_name: stable-baselines3
tags:
- LunarLander-v2
- deep-reinforcement-learning
- reinforcement-learning
- stable-baselines3
model-index:
- name: PPO
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: LunarLander-v2
type: LunarLander-v2
metrics:
- type: mean_reward
value: 271.77 +/- 12.60
name: mean_reward
verified: false
---
# **PPO** Agent playing **LunarLander-v2**
This is a trained model of a **PPO** agent playing **LunarLander-v2**
using the [stable-baselines3 library](https://github.com/DLR-RM/stable-baselines3).
## Usage (with Stable-baselines3)
TODO: Add your code
```python
from stable_baselines3 import ...
from huggingface_sb3 import load_from_hub
...
```
|
aroot/eng-guj-simcse_central_ssblu
|
aroot
| 2023-07-06T19:00:09Z | 103 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"mbart",
"text2text-generation",
"translation",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
translation
| 2023-07-06T18:38:04Z |
---
tags:
- translation
- generated_from_trainer
metrics:
- bleu
model-index:
- name: eng-guj-simcse_central_ssblu
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# eng-guj-simcse_central_ssblu
This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 3.2711
- Bleu: 2.6084
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3
|
JoshELambert/marginalization
|
JoshELambert
| 2023-07-06T19:00:00Z | 3 | 0 |
sentence-transformers
|
[
"sentence-transformers",
"pytorch",
"mpnet",
"setfit",
"text-classification",
"arxiv:2209.11055",
"license:apache-2.0",
"region:us"
] |
text-classification
| 2023-07-06T18:48:46Z |
---
license: apache-2.0
tags:
- setfit
- sentence-transformers
- text-classification
pipeline_tag: text-classification
---
# /var/folders/8x/qp375g154zg3h3ktpt_8tyqw0000gn/T/tmpfs9t4b8m/JoshELambert/marginalization
This is a [SetFit model](https://github.com/huggingface/setfit) that can be used for text classification. The model has been trained using an efficient few-shot learning technique that involves:
1. Fine-tuning a [Sentence Transformer](https://www.sbert.net) with contrastive learning.
2. Training a classification head with features from the fine-tuned Sentence Transformer.
## Usage
To use this model for inference, first install the SetFit library:
```bash
python -m pip install setfit
```
You can then run inference as follows:
```python
from setfit import SetFitModel
# Download from Hub and run inference
model = SetFitModel.from_pretrained("/var/folders/8x/qp375g154zg3h3ktpt_8tyqw0000gn/T/tmpfs9t4b8m/JoshELambert/marginalization")
# Run inference
preds = model(["i loved the spiderman movie!", "pineapple on pizza is the worst 🤮"])
```
## BibTeX entry and citation info
```bibtex
@article{https://doi.org/10.48550/arxiv.2209.11055,
doi = {10.48550/ARXIV.2209.11055},
url = {https://arxiv.org/abs/2209.11055},
author = {Tunstall, Lewis and Reimers, Nils and Jo, Unso Eun Seo and Bates, Luke and Korat, Daniel and Wasserblat, Moshe and Pereg, Oren},
keywords = {Computation and Language (cs.CL), FOS: Computer and information sciences, FOS: Computer and information sciences},
title = {Efficient Few-Shot Learning Without Prompts},
publisher = {arXiv},
year = {2022},
copyright = {Creative Commons Attribution 4.0 International}
}
```
|
jjhonny/rl_course_vizdoom_health_gathering_supreme
|
jjhonny
| 2023-07-06T18:57:20Z | 0 | 0 |
sample-factory
|
[
"sample-factory",
"tensorboard",
"deep-reinforcement-learning",
"reinforcement-learning",
"model-index",
"region:us"
] |
reinforcement-learning
| 2023-07-06T18:57:13Z |
---
library_name: sample-factory
tags:
- deep-reinforcement-learning
- reinforcement-learning
- sample-factory
model-index:
- name: APPO
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: doom_health_gathering_supreme
type: doom_health_gathering_supreme
metrics:
- type: mean_reward
value: 11.13 +/- 4.89
name: mean_reward
verified: false
---
A(n) **APPO** model trained on the **doom_health_gathering_supreme** environment.
This model was trained using Sample-Factory 2.0: https://github.com/alex-petrenko/sample-factory.
Documentation for how to use Sample-Factory can be found at https://www.samplefactory.dev/
## Downloading the model
After installing Sample-Factory, download the model with:
```
python -m sample_factory.huggingface.load_from_hub -r jjhonny/rl_course_vizdoom_health_gathering_supreme
```
## Using the model
To run the model after download, use the `enjoy` script corresponding to this environment:
```
python -m .usr.local.lib.python3.10.dist-packages.ipykernel_launcher --algo=APPO --env=doom_health_gathering_supreme --train_dir=./train_dir --experiment=rl_course_vizdoom_health_gathering_supreme
```
You can also upload models to the Hugging Face Hub using the same script with the `--push_to_hub` flag.
See https://www.samplefactory.dev/10-huggingface/huggingface/ for more details
## Training with this model
To continue training with this model, use the `train` script corresponding to this environment:
```
python -m .usr.local.lib.python3.10.dist-packages.ipykernel_launcher --algo=APPO --env=doom_health_gathering_supreme --train_dir=./train_dir --experiment=rl_course_vizdoom_health_gathering_supreme --restart_behavior=resume --train_for_env_steps=10000000000
```
Note, you may have to adjust `--train_for_env_steps` to a suitably high number as the experiment will resume at the number of steps it concluded at.
|
Yati12/PPO_LunarLander-v2
|
Yati12
| 2023-07-06T18:57:16Z | 0 | 0 |
stable-baselines3
|
[
"stable-baselines3",
"LunarLander-v2",
"deep-reinforcement-learning",
"reinforcement-learning",
"model-index",
"region:us"
] |
reinforcement-learning
| 2023-07-06T18:56:57Z |
---
library_name: stable-baselines3
tags:
- LunarLander-v2
- deep-reinforcement-learning
- reinforcement-learning
- stable-baselines3
model-index:
- name: PPO
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: LunarLander-v2
type: LunarLander-v2
metrics:
- type: mean_reward
value: 253.17 +/- 23.77
name: mean_reward
verified: false
---
# **PPO** Agent playing **LunarLander-v2**
This is a trained model of a **PPO** agent playing **LunarLander-v2**
using the [stable-baselines3 library](https://github.com/DLR-RM/stable-baselines3).
## Usage (with Stable-baselines3)
TODO: Add your code
```python
from stable_baselines3 import ...
from huggingface_sb3 import load_from_hub
...
```
|
ALM-AHME/beit-large-patch16-224-finetuned-eurosat-50
|
ALM-AHME
| 2023-07-06T18:56:49Z | 5 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"beit",
"image-classification",
"generated_from_trainer",
"dataset:imagefolder",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
image-classification
| 2023-07-06T17:08:12Z |
---
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- imagefolder
metrics:
- accuracy
model-index:
- name: beit-large-patch16-224-finetuned-eurosat-50
results:
- task:
name: Image Classification
type: image-classification
dataset:
name: imagefolder
type: imagefolder
config: Augmented-Final
split: train
args: Augmented-Final
metrics:
- name: Accuracy
type: accuracy
value: 0.9856115107913669
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# beit-large-patch16-224-finetuned-eurosat-50
This model is a fine-tuned version of [microsoft/beit-large-patch16-224](https://huggingface.co/microsoft/beit-large-patch16-224) on the imagefolder dataset.
It achieves the following results on the evaluation set:
- Loss: 0.0568
- Accuracy: 0.9856
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-06
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.9
- num_epochs: 12
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| 1.7148 | 1.0 | 122 | 1.6402 | 0.3916 |
| 1.1543 | 2.0 | 244 | 1.0718 | 0.6208 |
| 0.8948 | 3.0 | 366 | 0.7228 | 0.7564 |
| 0.6348 | 4.0 | 488 | 0.5327 | 0.8160 |
| 0.647 | 5.0 | 610 | 0.4081 | 0.8551 |
| 0.3244 | 6.0 | 732 | 0.2965 | 0.9096 |
| 0.305 | 7.0 | 854 | 0.2515 | 0.9342 |
| 0.3522 | 8.0 | 976 | 0.1667 | 0.9568 |
| 0.1782 | 9.0 | 1098 | 0.1494 | 0.9568 |
| 0.1849 | 10.0 | 1220 | 0.0972 | 0.9712 |
| 0.1814 | 11.0 | 1342 | 0.0559 | 0.9846 |
| 0.1682 | 12.0 | 1464 | 0.0568 | 0.9856 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu118
- Datasets 2.13.1
- Tokenizers 0.13.3
|
EllaHong/datamap_kullm-polyglot-5.8b_exp2
|
EllaHong
| 2023-07-06T18:49:54Z | 0 | 0 |
peft
|
[
"peft",
"region:us"
] | null | 2023-07-06T18:49:37Z |
---
library_name: peft
---
## Training procedure
The following `bitsandbytes` quantization config was used during training:
- load_in_8bit: False
- load_in_4bit: True
- llm_int8_threshold: 6.0
- llm_int8_skip_modules: None
- llm_int8_enable_fp32_cpu_offload: False
- llm_int8_has_fp16_weight: False
- bnb_4bit_quant_type: nf4
- bnb_4bit_use_double_quant: True
- bnb_4bit_compute_dtype: bfloat16
### Framework versions
- PEFT 0.4.0.dev0
|
aroot/eng-guj-simcse_random_ssblu
|
aroot
| 2023-07-06T18:45:32Z | 110 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"mbart",
"text2text-generation",
"translation",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
translation
| 2023-07-06T18:24:22Z |
---
tags:
- translation
- generated_from_trainer
metrics:
- bleu
model-index:
- name: eng-guj-simcse_random_ssblu
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# eng-guj-simcse_random_ssblu
This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 3.2774
- Bleu: 2.7054
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3
|
hopkins/eng-guj-common.simcse.roberta-large
|
hopkins
| 2023-07-06T18:37:12Z | 104 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"mbart",
"text2text-generation",
"translation",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
translation
| 2023-07-06T18:16:00Z |
---
tags:
- translation
- generated_from_trainer
metrics:
- bleu
model-index:
- name: eng-guj-common.simcse.roberta-large
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# eng-guj-common.simcse.roberta-large
This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 3.2500
- Bleu: 3.2434
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3
|
BrainTheos/whisper-tiny-en-us
|
BrainTheos
| 2023-07-06T18:27:29Z | 76 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"whisper",
"automatic-speech-recognition",
"generated_from_trainer",
"dataset:PolyAI/minds14",
"base_model:openai/whisper-tiny",
"base_model:finetune:openai/whisper-tiny",
"license:apache-2.0",
"model-index",
"endpoints_compatible",
"region:us"
] |
automatic-speech-recognition
| 2023-07-06T17:32:37Z |
---
license: apache-2.0
base_model: openai/whisper-tiny
tags:
- generated_from_trainer
datasets:
- PolyAI/minds14
metrics:
- wer
model-index:
- name: whisper-tiny-en-us
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: PolyAI/minds14
type: PolyAI/minds14
config: en-US
split: train
args: en-US
metrics:
- name: Wer
type: wer
value: 0.36046511627906974
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# whisper-tiny-en-us
This model is a fine-tuned version of [openai/whisper-tiny](https://huggingface.co/openai/whisper-tiny) on the PolyAI/minds14 dataset.
It achieves the following results on the evaluation set:
- Loss: 0.7173
- Wer Ortho: 36.1373
- Wer: 0.3605
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: constant_with_warmup
- lr_scheduler_warmup_steps: 50
- training_steps: 500
### Training results
| Training Loss | Epoch | Step | Validation Loss | Wer Ortho | Wer |
|:-------------:|:-----:|:----:|:---------------:|:---------:|:------:|
| 0.0007 | 17.86 | 500 | 0.7173 | 36.1373 | 0.3605 |
### Framework versions
- Transformers 4.31.0.dev0
- Pytorch 2.0.1+cu118
- Datasets 2.13.1
- Tokenizers 0.13.3
|
jordyvl/EElayoutlmv3_jordyvl_rvl_cdip_100_examples_per_class_2023-07-06_g040
|
jordyvl
| 2023-07-06T18:14:21Z | 102 | 0 |
transformers
|
[
"transformers",
"pytorch",
"layoutlmv3",
"text-classification",
"generated_from_trainer",
"license:cc-by-nc-sa-4.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2023-07-06T16:05:16Z |
---
license: cc-by-nc-sa-4.0
tags:
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: EElayoutlmv3_jordyvl_rvl_cdip_100_examples_per_class_2023-07-06_g040
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# EElayoutlmv3_jordyvl_rvl_cdip_100_examples_per_class_2023-07-06_g040
This model is a fine-tuned version of [microsoft/layoutlmv3-base](https://huggingface.co/microsoft/layoutlmv3-base) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 1.1088
- Accuracy: 0.715
- Exit 0 Accuracy: 0.1175
- Exit 1 Accuracy: 0.1575
- Exit 2 Accuracy: 0.3075
- Exit 3 Accuracy: 0.32
- Exit 4 Accuracy: 0.0675
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 12
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 24
- total_train_batch_size: 288
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 60
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Exit 0 Accuracy | Exit 1 Accuracy | Exit 2 Accuracy | Exit 3 Accuracy | Exit 4 Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:---------------:|:---------------:|:---------------:|:---------------:|:---------------:|
| No log | 0.72 | 2 | 2.7601 | 0.11 | 0.0825 | 0.0675 | 0.0875 | 0.0625 | 0.0625 |
| No log | 1.72 | 4 | 2.7329 | 0.115 | 0.07 | 0.065 | 0.115 | 0.0625 | 0.0625 |
| No log | 2.72 | 6 | 2.6953 | 0.135 | 0.075 | 0.06 | 0.12 | 0.0625 | 0.0625 |
| No log | 3.72 | 8 | 2.6612 | 0.165 | 0.08 | 0.055 | 0.1225 | 0.0625 | 0.0625 |
| No log | 4.72 | 10 | 2.6176 | 0.1925 | 0.0875 | 0.0575 | 0.1175 | 0.0625 | 0.0625 |
| No log | 5.72 | 12 | 2.5681 | 0.2125 | 0.09 | 0.08 | 0.1225 | 0.0625 | 0.0625 |
| No log | 6.72 | 14 | 2.5380 | 0.2125 | 0.095 | 0.08 | 0.125 | 0.0625 | 0.0625 |
| No log | 7.72 | 16 | 2.5137 | 0.2275 | 0.095 | 0.09 | 0.125 | 0.0625 | 0.0625 |
| No log | 8.72 | 18 | 2.4662 | 0.2775 | 0.095 | 0.0975 | 0.125 | 0.0625 | 0.0625 |
| No log | 9.72 | 20 | 2.4192 | 0.3 | 0.0925 | 0.105 | 0.1275 | 0.0625 | 0.0625 |
| No log | 10.72 | 22 | 2.3755 | 0.3075 | 0.095 | 0.1225 | 0.135 | 0.0625 | 0.0625 |
| No log | 11.72 | 24 | 2.3290 | 0.3225 | 0.0975 | 0.1175 | 0.125 | 0.0625 | 0.0625 |
| No log | 12.72 | 26 | 2.2739 | 0.3375 | 0.1 | 0.115 | 0.125 | 0.0625 | 0.0625 |
| No log | 13.72 | 28 | 2.2219 | 0.3525 | 0.0975 | 0.125 | 0.13 | 0.065 | 0.0625 |
| No log | 14.72 | 30 | 2.1835 | 0.3525 | 0.1 | 0.125 | 0.1475 | 0.065 | 0.0625 |
| No log | 15.72 | 32 | 2.1610 | 0.3725 | 0.1025 | 0.1275 | 0.155 | 0.0675 | 0.0625 |
| No log | 16.72 | 34 | 2.1139 | 0.39 | 0.1025 | 0.135 | 0.1675 | 0.07 | 0.0625 |
| No log | 17.72 | 36 | 2.0748 | 0.405 | 0.1 | 0.1375 | 0.185 | 0.0725 | 0.0625 |
| No log | 18.72 | 38 | 2.0145 | 0.4225 | 0.1025 | 0.14 | 0.1875 | 0.0725 | 0.0625 |
| No log | 19.72 | 40 | 1.9595 | 0.4475 | 0.1025 | 0.145 | 0.185 | 0.0725 | 0.0625 |
| No log | 20.72 | 42 | 1.9077 | 0.4875 | 0.1025 | 0.1425 | 0.18 | 0.085 | 0.0625 |
| No log | 21.72 | 44 | 1.8328 | 0.52 | 0.1025 | 0.145 | 0.185 | 0.11 | 0.0625 |
| No log | 22.72 | 46 | 1.7703 | 0.555 | 0.105 | 0.1425 | 0.185 | 0.1125 | 0.0625 |
| No log | 23.72 | 48 | 1.7462 | 0.565 | 0.11 | 0.1425 | 0.2025 | 0.11 | 0.0625 |
| No log | 24.72 | 50 | 1.6894 | 0.5625 | 0.1125 | 0.14 | 0.205 | 0.12 | 0.0625 |
| No log | 25.72 | 52 | 1.6273 | 0.585 | 0.1125 | 0.1475 | 0.205 | 0.1225 | 0.0625 |
| No log | 26.72 | 54 | 1.5894 | 0.5875 | 0.115 | 0.1475 | 0.21 | 0.1325 | 0.0625 |
| No log | 27.72 | 56 | 1.5567 | 0.605 | 0.115 | 0.1475 | 0.21 | 0.13 | 0.0625 |
| No log | 28.72 | 58 | 1.5013 | 0.6225 | 0.115 | 0.1475 | 0.215 | 0.135 | 0.0625 |
| No log | 29.72 | 60 | 1.4588 | 0.64 | 0.115 | 0.15 | 0.2175 | 0.145 | 0.0625 |
| No log | 30.72 | 62 | 1.4424 | 0.6425 | 0.115 | 0.15 | 0.23 | 0.145 | 0.065 |
| No log | 31.72 | 64 | 1.4074 | 0.65 | 0.115 | 0.1475 | 0.245 | 0.1475 | 0.065 |
| No log | 32.72 | 66 | 1.3663 | 0.6675 | 0.115 | 0.1475 | 0.2475 | 0.17 | 0.065 |
| No log | 33.72 | 68 | 1.3465 | 0.67 | 0.1175 | 0.1475 | 0.26 | 0.17 | 0.065 |
| No log | 34.72 | 70 | 1.3363 | 0.6675 | 0.115 | 0.15 | 0.265 | 0.18 | 0.065 |
| No log | 35.72 | 72 | 1.3183 | 0.67 | 0.1175 | 0.15 | 0.2725 | 0.185 | 0.0625 |
| No log | 36.72 | 74 | 1.2789 | 0.7025 | 0.1175 | 0.1525 | 0.2725 | 0.195 | 0.0625 |
| No log | 37.72 | 76 | 1.2625 | 0.7025 | 0.12 | 0.1525 | 0.2725 | 0.22 | 0.065 |
| No log | 38.72 | 78 | 1.2645 | 0.6875 | 0.12 | 0.1525 | 0.2725 | 0.2325 | 0.065 |
| No log | 39.72 | 80 | 1.2384 | 0.695 | 0.1225 | 0.1525 | 0.275 | 0.24 | 0.065 |
| No log | 40.72 | 82 | 1.2138 | 0.7075 | 0.1225 | 0.1525 | 0.29 | 0.2475 | 0.065 |
| No log | 41.72 | 84 | 1.2041 | 0.6975 | 0.12 | 0.1525 | 0.29 | 0.2475 | 0.065 |
| No log | 42.72 | 86 | 1.1907 | 0.7075 | 0.1175 | 0.1525 | 0.29 | 0.2575 | 0.0625 |
| No log | 43.72 | 88 | 1.1784 | 0.7075 | 0.1175 | 0.1525 | 0.2925 | 0.2675 | 0.0625 |
| No log | 44.72 | 90 | 1.1678 | 0.715 | 0.1175 | 0.1525 | 0.2925 | 0.2875 | 0.0625 |
| No log | 45.72 | 92 | 1.1662 | 0.715 | 0.1175 | 0.155 | 0.295 | 0.285 | 0.0625 |
| No log | 46.72 | 94 | 1.1568 | 0.715 | 0.1175 | 0.155 | 0.295 | 0.2925 | 0.0625 |
| No log | 47.72 | 96 | 1.1497 | 0.715 | 0.1175 | 0.155 | 0.3 | 0.3 | 0.0625 |
| No log | 48.72 | 98 | 1.1456 | 0.715 | 0.1175 | 0.1575 | 0.3 | 0.3025 | 0.065 |
| No log | 49.72 | 100 | 1.1406 | 0.7125 | 0.1175 | 0.1575 | 0.2975 | 0.305 | 0.0675 |
| No log | 50.72 | 102 | 1.1333 | 0.72 | 0.1175 | 0.1575 | 0.2975 | 0.305 | 0.0675 |
| No log | 51.72 | 104 | 1.1242 | 0.7175 | 0.1175 | 0.1575 | 0.2975 | 0.3125 | 0.0675 |
| No log | 52.72 | 106 | 1.1197 | 0.7125 | 0.1175 | 0.1575 | 0.2975 | 0.3125 | 0.0675 |
| No log | 53.72 | 108 | 1.1161 | 0.715 | 0.1175 | 0.1575 | 0.3 | 0.3125 | 0.0675 |
| No log | 54.72 | 110 | 1.1114 | 0.715 | 0.1175 | 0.1575 | 0.3075 | 0.3125 | 0.0675 |
| No log | 55.72 | 112 | 1.1096 | 0.715 | 0.1175 | 0.1575 | 0.315 | 0.32 | 0.0675 |
| No log | 56.72 | 114 | 1.1084 | 0.715 | 0.1175 | 0.1575 | 0.3125 | 0.32 | 0.0675 |
| No log | 57.72 | 116 | 1.1085 | 0.715 | 0.1175 | 0.1575 | 0.3075 | 0.32 | 0.0675 |
| No log | 58.72 | 118 | 1.1089 | 0.7125 | 0.1175 | 0.1575 | 0.3075 | 0.32 | 0.0675 |
| No log | 59.72 | 120 | 1.1088 | 0.715 | 0.1175 | 0.1575 | 0.3075 | 0.32 | 0.0675 |
### Framework versions
- Transformers 4.26.1
- Pytorch 1.13.1.post200
- Datasets 2.9.0
- Tokenizers 0.13.2
|
TheBloke/Vicuna-7B-v1-3-SuperHOT-8K-GGML
|
TheBloke
| 2023-07-06T18:12:35Z | 0 | 4 | null |
[
"arxiv:2302.13971",
"arxiv:2306.05685",
"license:other",
"region:us"
] | null | 2023-07-06T18:06:25Z |
---
inference: false
license: other
---
<!-- header start -->
<div style="width: 100%;">
<img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
</div>
<div style="display: flex; justify-content: space-between; width: 100%;">
<div style="display: flex; flex-direction: column; align-items: flex-start;">
<p><a href="https://discord.gg/theblokeai">Chat & support: my new Discord server</a></p>
</div>
<div style="display: flex; flex-direction: column; align-items: flex-end;">
<p><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p>
</div>
</div>
<!-- header end -->
# LmSys' Vicuna 7B v1.3 GGML
These files are GGML format model files for [LmSys' Vicuna 7B v1.3](https://huggingface.co/lmsys/vicuna-7b-v1.3).
These are SuperHOT GGMLs with an increased context length. SuperHOT is a new system that employs RoPE to expand context beyond what was originally possible for a model. It was discovered and developed by [kaiokendev](https://huggingface.co/kaiokendev).
In order to use the increased context length, you can presently use:
* [KoboldCpp](https://github.com/LostRuins/koboldcpp) - [release 1.33](https://github.com/LostRuins/koboldcpp/releases/tag/v1.33) or later.
Support is also expected to come to llama.cpp, however work is still being done to find the optimal implementation.
To use the increased context with KoboldCpp, simply use `--contextsize` to set the desired context, eg `--contextsize 4096` or `--contextsize 8192`.
**NOTE**: Increased context length is an area seeing rapid developments and improvements. It is quite possible that these models may be superseded by new developments in the coming days. If that's the case, I will remove them, or update this README as appropriate.
## Repositories available
* [4-bit GPTQ models for GPU inference](https://huggingface.co/TheBloke/Vicuna-7B-v1-3-SuperHOT-8K-GPTQ)
* [2, 3, 4, 5, 6 and 8-bit GGML models for CPU inference](https://huggingface.co/TheBloke/Vicuna-7B-v1-3-SuperHOT-8K-GGML)
* [Unquantised SuperHOT fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/TheBloke/Vicuna-7B-v1-3-SuperHOT-8K-fp16)
* [Unquantised base fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/lmsys/vicuna-7b-v1.3)
<!-- compatibility_ggml start -->
## Compatibility
These GGMLs will work with any llama.cpp-compatible GGML client that supports k-quants.
However the increased context length won't work without specific support. See the note in the introduction for details on using increased context.
## Explanation of the new k-quant methods
The new methods available are:
* GGML_TYPE_Q2_K - "type-1" 2-bit quantization in super-blocks containing 16 blocks, each block having 16 weight. Block scales and mins are quantized with 4 bits. This ends up effectively using 2.5625 bits per weight (bpw)
* GGML_TYPE_Q3_K - "type-0" 3-bit quantization in super-blocks containing 16 blocks, each block having 16 weights. Scales are quantized with 6 bits. This end up using 3.4375 bpw.
* GGML_TYPE_Q4_K - "type-1" 4-bit quantization in super-blocks containing 8 blocks, each block having 32 weights. Scales and mins are quantized with 6 bits. This ends up using 4.5 bpw.
* GGML_TYPE_Q5_K - "type-1" 5-bit quantization. Same super-block structure as GGML_TYPE_Q4_K resulting in 5.5 bpw
* GGML_TYPE_Q6_K - "type-0" 6-bit quantization. Super-blocks with 16 blocks, each block having 16 weights. Scales are quantized with 8 bits. This ends up using 6.5625 bpw
* GGML_TYPE_Q8_K - "type-0" 8-bit quantization. Only used for quantizing intermediate results. The difference to the existing Q8_0 is that the block size is 256. All 2-6 bit dot products are implemented for this quantization type.
Refer to the Provided Files table below to see what files use which methods, and how.
<!-- compatibility_ggml end -->
## Provided files
| Name | Quant method | Bits | Size | Max RAM required | Use case |
| ---- | ---- | ---- | ---- | ---- | ----- |
| vicuna-7b-v1.3-superhot-8k.ggmlv3.q2_K.bin | q2_K | 2 | 2.87 GB | 5.37 GB | New k-quant method. Uses GGML_TYPE_Q4_K for the attention.vw and feed_forward.w2 tensors, GGML_TYPE_Q2_K for the other tensors. |
| vicuna-7b-v1.3-superhot-8k.ggmlv3.q3_K_L.bin | q3_K_L | 3 | 3.60 GB | 6.10 GB | New k-quant method. Uses GGML_TYPE_Q5_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else GGML_TYPE_Q3_K |
| vicuna-7b-v1.3-superhot-8k.ggmlv3.q3_K_M.bin | q3_K_M | 3 | 3.28 GB | 5.78 GB | New k-quant method. Uses GGML_TYPE_Q4_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else GGML_TYPE_Q3_K |
| vicuna-7b-v1.3-superhot-8k.ggmlv3.q3_K_S.bin | q3_K_S | 3 | 2.95 GB | 5.45 GB | New k-quant method. Uses GGML_TYPE_Q3_K for all tensors |
| vicuna-7b-v1.3-superhot-8k.ggmlv3.q4_K_M.bin | q4_K_M | 4 | 4.08 GB | 6.58 GB | New k-quant method. Uses GGML_TYPE_Q6_K for half of the attention.wv and feed_forward.w2 tensors, else GGML_TYPE_Q4_K |
| vicuna-7b-v1.3-superhot-8k.ggmlv3.q4_K_S.bin | q4_K_S | 4 | 3.83 GB | 6.33 GB | New k-quant method. Uses GGML_TYPE_Q4_K for all tensors |
| vicuna-7b-v1.3-superhot-8k.ggmlv3.q5_K_M.bin | q5_K_M | 5 | 4.78 GB | 7.28 GB | New k-quant method. Uses GGML_TYPE_Q6_K for half of the attention.wv and feed_forward.w2 tensors, else GGML_TYPE_Q5_K |
| vicuna-7b-v1.3-superhot-8k.ggmlv3.q5_K_S.bin | q5_K_S | 5 | 4.65 GB | 7.15 GB | New k-quant method. Uses GGML_TYPE_Q5_K for all tensors |
| vicuna-7b-v1.3-superhot-8k.ggmlv3.q6_K.bin | q6_K | 6 | 5.53 GB | 8.03 GB | New k-quant method. Uses GGML_TYPE_Q8_K - 6-bit quantization - for all tensors |
**Note**: the above RAM figures assume no GPU offloading. If layers are offloaded to the GPU, this will reduce RAM usage and use VRAM instead.
## How to run in `koboldcpp`
On Linux I use the following command line to launch the KoboldCpp UI with OpenCL aceleration and a context size of 4096:
```
python ./koboldcpp.py --stream --unbantokens --threads 8 --usecublas --gpulayers 100 vicuna-7b-v1.3-superhot-8k.ggmlv3.q4_K_M.bin
```
Change `--gpulayers 100` to the number of layers you want/are able to offload to the GPU. Remove it if you don't have GPU acceleration.
For OpenCL acceleration, change `--usecublas` to `--useclblast 0 0`. You may need to change the second `0` to `1` if you have both an iGPU and a discrete GPU.
<!-- footer start -->
## Discord
For further support, and discussions on these models and AI in general, join us at:
[TheBloke AI's Discord server](https://discord.gg/theblokeai)
## Thanks, and how to contribute.
Thanks to the [chirper.ai](https://chirper.ai) team!
I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.
If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.
Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
* Patreon: https://patreon.com/TheBlokeAI
* Ko-Fi: https://ko-fi.com/TheBlokeAI
**Special thanks to**: Luke from CarbonQuill, Aemon Algiz.
**Patreon special mentions**: RoA, Lone Striker, Gabriel Puliatti, Derek Yates, Randy H, Jonathan Leane, Eugene Pentland, Karl Bernard, Viktor Bowallius, senxiiz, Daniel P. Andersen, Pierre Kircher, Deep Realms, Cory Kujawski, Oscar Rangel, Fen Risland, Ajan Kanaga, LangChain4j, webtim, Nikolai Manek, Trenton Dambrowitz, Raven Klaugh, Kalila, Khalefa Al-Ahmad, Chris McCloskey, Luke @flexchar, Ai Maven, Dave, Asp the Wyvern, Sean Connelly, Imad Khwaja, Space Cruiser, Rainer Wilmers, subjectnull, Alps Aficionado, Willian Hasse, Fred von Graf, Artur Olbinski, Johann-Peter Hartmann, WelcomeToTheClub, Willem Michiel, Michael Levine, Iucharbius , Spiking Neurons AB, K, biorpg, John Villwock, Pyrater, Greatston Gnanesh, Mano Prime, Junyu Yang, Stephen Murray, John Detwiler, Luke Pendergrass, terasurfer , Pieter, zynix , Edmond Seymore, theTransient, Nathan LeClaire, vamX, Kevin Schuppel, Preetika Verma, ya boyyy, Alex , SuperWojo, Ghost , Joseph William Delisle, Matthew Berman, Talal Aujan, chris gileta, Illia Dulskyi.
Thank you to all my generous patrons and donaters!
<!-- footer end -->
# Original model card: Kaio Ken's SuperHOT 8K
### SuperHOT Prototype 2 w/ 8K Context
This is a second prototype of SuperHOT, a NSFW focused LoRA, this time 7B with 8K context and no RLHF, using the same technique described in [the github blog](https://kaiokendev.github.io/til#extending-context-to-8k).
#### Looking for Merged & Quantized Models?
Make some please :)
#### Using the monkey-patch?
You will **NEED** to **apply the monkeypatch** or, if you are already using the monkeypatch, **change the scaling factor to 0.25 and the maximum sequence length to 8192**
The monkeypatch is only necessary if you are using a front-end/back-end that does not already support scaling and said front-end/back-end is Python-based (i.e. Huggingface Transformers). To apply the patch, you will need to copy the `llama_rope_scaled_monkey_patch.py` into your working directory and call the exported function `replace_llama_rope_with_scaled_rope` at the very start of your Python program. It will modify the Transformers library's implementation of RoPE to properly apply the scaling factor.
#### Using Oobabooga with Exllama?
Switch your loader to `exllama` or `exllama_hf` Add the arguments `max_seq_len 8192` and `compress_pos_emb 4`. **While the model may work well with `compress_pos_emb 2`, it was trained on 4, so that is what I advocate for you to use**
Example in the command-line:
- `python server.py --max_seq_len 8192 --compress_pos_emb 4 --loader exllama_hf`
In the UI, you will see the loader option in the `Models` tab. Once you select either `exllama` or `exllama_hf`, the `max_seq_len` and `compress_pos_emb` settings will appear.
#### Training Details
I trained the LoRA with the following configuration:
- 1200 samples (~400 samples over 2048 sequence length)
- learning rate of 3e-4
- 3 epochs
- The exported modules are:
- q_proj
- k_proj
- v_proj
- o_proj
- no bias
- Rank = 4
- Alpha = 8
- no dropout
- weight decay of 0.1
- AdamW beta1 of 0.9 and beta2 0.99, epsilon of 1e-5
- Trained on 4-bit base model
- Cutoff length: 4096
# Original model card: LmSys' Vicuna 7B v1.3
# Vicuna Model Card
## Model Details
Vicuna is a chat assistant trained by fine-tuning LLaMA on user-shared conversations collected from ShareGPT.
- **Developed by:** [LMSYS](https://lmsys.org/)
- **Model type:** An auto-regressive language model based on the transformer architecture.
- **License:** Non-commercial license
- **Finetuned from model:** [LLaMA](https://arxiv.org/abs/2302.13971).
### Model Sources
- **Repository:** https://github.com/lm-sys/FastChat
- **Blog:** https://lmsys.org/blog/2023-03-30-vicuna/
- **Paper:** https://arxiv.org/abs/2306.05685
- **Demo:** https://chat.lmsys.org/
## Uses
The primary use of Vicuna is research on large language models and chatbots.
The primary intended users of the model are researchers and hobbyists in natural language processing, machine learning, and artificial intelligence.
## How to Get Started with the Model
Command line interface: https://github.com/lm-sys/FastChat#vicuna-weights.
APIs (OpenAI API, Huggingface API): https://github.com/lm-sys/FastChat/tree/main#api.
## Training Details
Vicuna v1.3 is fine-tuned from LLaMA with supervised instruction fine-tuning.
The training data is around 140K conversations collected from ShareGPT.com.
See more details in the "Training Details of Vicuna Models" section in the appendix of this [paper](https://arxiv.org/pdf/2306.05685.pdf).
## Evaluation
Vicuna is evaluated with standard benchmarks, human preference, and LLM-as-a-judge. See more details in this [paper](https://arxiv.org/pdf/2306.05685.pdf) and [leaderboard](https://huggingface.co/spaces/lmsys/chatbot-arena-leaderboard).
## Difference between different versions of Vicuna
See [vicuna_weights_version.md](https://github.com/lm-sys/FastChat/blob/main/docs/vicuna_weights_version.md)
|
Panchovix/airoboros-33b-gpt4-1.2-PI-8192-LoRA-4bit-32g
|
Panchovix
| 2023-07-06T18:11:35Z | 5 | 0 |
transformers
|
[
"transformers",
"pytorch",
"llama",
"text-generation",
"license:other",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] |
text-generation
| 2023-07-04T18:43:19Z |
---
license: other
---
[airoboros-13b-gpt4-1.2](https://huggingface.co/jondurbin/airoboros-13b-gpt4-1.2) merged with bhenrym14's [airoboros-33b-gpt4-1.4.1-PI-8192-LoRA](https://huggingface.co/bhenrym14/airoboros-33b-gpt4-1.4.1-PI-8192-LoRA), quantized at 4 bit.
More info about the LoRA [Here](https://huggingface.co/bhenrym14/airoboros-33b-gpt4-1.4.1-PI-8192-fp16). This is an alternative to SuperHOT 8k LoRA trained with LoRA_rank 64, and airoboros 1.4.1 dataset.
It was created with GPTQ-for-LLaMA with group size 32 and act order true as parameters, to get the maximum perplexity vs FP16 model.
I HIGHLY suggest to use exllama, to evade some VRAM issues.
Use compress_pos_emb = 4 for any context up to 8192 context.
If you have 2x24 GB VRAM GPUs cards, to not get Out of Memory errors at 8192 context, use:
gpu_split: 9,21
|
Panchovix/tulu-30b-SuperHOT-8K-4bit-32g
|
Panchovix
| 2023-07-06T18:09:41Z | 5 | 0 |
transformers
|
[
"transformers",
"pytorch",
"llama",
"text-generation",
"license:other",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] |
text-generation
| 2023-06-27T00:59:40Z |
---
license: other
---
[Tulu-30B-SuperHOT-8K-GPTQ by TheBloke](https://huggingface.co/TheBloke/Tulu-30B-SuperHOT-8K-fp16) quantized at 4 bit.
It was created with GPTQ-for-LLaMA with group size 32 and act order true as parameters, to get the maximum perplexity vs FP16 model.
I HIGHLY suggest to use exllama, to evade some VRAM issues.
Use compress_pos_emb = 4 for any context up to 8192 context.
If you have 2x24 GB VRAM GPUs cards, to not get Out of Memory errors at 8192 context, use:
gpu_split: 9,21
|
hopkins/eng-fra-common.simcse.roberta-large
|
hopkins
| 2023-07-06T18:02:47Z | 6 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"mbart",
"text2text-generation",
"translation",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
translation
| 2023-07-06T17:44:05Z |
---
tags:
- translation
- generated_from_trainer
metrics:
- bleu
model-index:
- name: eng-fra-common.simcse.roberta-large
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# eng-fra-common.simcse.roberta-large
This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 1.1339
- Bleu: 33.2260
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3
|
jcramirezpr/sd-class-butterflies-32-small
|
jcramirezpr
| 2023-07-06T18:01:33Z | 31 | 0 |
diffusers
|
[
"diffusers",
"pytorch",
"unconditional-image-generation",
"diffusion-models-class",
"license:mit",
"diffusers:DDPMPipeline",
"region:us"
] |
unconditional-image-generation
| 2023-07-06T18:01:24Z |
---
license: mit
tags:
- pytorch
- diffusers
- unconditional-image-generation
- diffusion-models-class
---
# Model Card for Unit 1 of the [Diffusion Models Class 🧨](https://github.com/huggingface/diffusion-models-class)
This model is a diffusion model for unconditional image generation of cute 🦋.
## Usage
```python
from diffusers import DDPMPipeline
pipeline = DDPMPipeline.from_pretrained('jcramirezpr/sd-class-butterflies-32-small')
image = pipeline().images[0]
image
```
|
YakovElm/Apache_5_BERT_More_Properties
|
YakovElm
| 2023-07-06T17:53:47Z | 61 | 0 |
transformers
|
[
"transformers",
"tf",
"bert",
"text-classification",
"generated_from_keras_callback",
"license:apache-2.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2023-07-06T17:53:09Z |
---
license: apache-2.0
tags:
- generated_from_keras_callback
model-index:
- name: Apache_5_BERT_More_Properties
results: []
---
<!-- This model card has been generated automatically according to the information Keras had access to. You should
probably proofread and complete it, then remove this comment. -->
# Apache_5_BERT_More_Properties
This model is a fine-tuned version of [bert-base-uncased](https://huggingface.co/bert-base-uncased) on an unknown dataset.
It achieves the following results on the evaluation set:
- Train Loss: 0.3031
- Train Accuracy: 0.9094
- Validation Loss: 0.4800
- Validation Accuracy: 0.8233
- Epoch: 2
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- optimizer: {'name': 'Adam', 'weight_decay': None, 'clipnorm': 1.0, 'global_clipnorm': None, 'clipvalue': None, 'use_ema': False, 'ema_momentum': 0.99, 'ema_overwrite_frequency': None, 'jit_compile': False, 'is_legacy_optimizer': False, 'learning_rate': 3e-05, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-08, 'amsgrad': False}
- training_precision: float32
### Training results
| Train Loss | Train Accuracy | Validation Loss | Validation Accuracy | Epoch |
|:----------:|:--------------:|:---------------:|:-------------------:|:-----:|
| 0.3135 | 0.9075 | 0.4977 | 0.8233 | 0 |
| 0.3075 | 0.9094 | 0.5032 | 0.8233 | 1 |
| 0.3031 | 0.9094 | 0.4800 | 0.8233 | 2 |
### Framework versions
- Transformers 4.29.2
- TensorFlow 2.12.0
- Datasets 2.12.0
- Tokenizers 0.13.3
|
hopkins/eng-deu-common.simcse.roberta-large
|
hopkins
| 2023-07-06T17:51:37Z | 103 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"mbart",
"text2text-generation",
"translation",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
translation
| 2023-07-06T17:37:45Z |
---
tags:
- translation
- generated_from_trainer
metrics:
- bleu
model-index:
- name: eng-deu-common.simcse.roberta-large
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# eng-deu-common.simcse.roberta-large
This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 1.6605
- Bleu: 21.3413
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3
|
aroot/eng-mya-simcse_central_usbbu
|
aroot
| 2023-07-06T17:33:28Z | 103 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"mbart",
"text2text-generation",
"translation",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
translation
| 2023-07-06T17:12:31Z |
---
tags:
- translation
- generated_from_trainer
metrics:
- bleu
model-index:
- name: eng-mya-simcse_central_usbbu
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# eng-mya-simcse_central_usbbu
This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 1.9021
- Bleu: 3.9804
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3
|
hopkins/eng-kor-common
|
hopkins
| 2023-07-06T17:33:00Z | 103 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"mbart",
"text2text-generation",
"translation",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
translation
| 2023-07-06T17:15:18Z |
---
tags:
- translation
- generated_from_trainer
metrics:
- bleu
model-index:
- name: eng-kor-common
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# eng-kor-common
This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 1.9972
- Bleu: 7.4980
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3
|
zwich07/3dmm
|
zwich07
| 2023-07-06T17:26:03Z | 0 | 0 | null |
[
"license:creativeml-openrail-m",
"region:us"
] | null | 2023-07-06T17:25:03Z |
---
license: creativeml-openrail-m
---
|
i617/falcon-n-adapt1
|
i617
| 2023-07-06T17:22:32Z | 6 | 0 |
peft
|
[
"peft",
"region:us"
] | null | 2023-06-21T07:41:15Z |
---
library_name: peft
---
## Training procedure
The following `bitsandbytes` quantization config was used during training:
- load_in_8bit: False
- load_in_4bit: True
- llm_int8_threshold: 6.0
- llm_int8_skip_modules: None
- llm_int8_enable_fp32_cpu_offload: False
- llm_int8_has_fp16_weight: False
- bnb_4bit_quant_type: nf4
- bnb_4bit_use_double_quant: True
- bnb_4bit_compute_dtype: bfloat16
### Framework versions
- PEFT 0.4.0.dev0
|
Evan-Lin/Bart-RL-little
|
Evan-Lin
| 2023-07-06T17:18:15Z | 49 | 0 |
transformers
|
[
"transformers",
"pytorch",
"bart",
"text2text-generation",
"trl",
"reinforcement-learning",
"license:apache-2.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
reinforcement-learning
| 2023-07-06T00:13:42Z |
---
license: apache-2.0
tags:
- trl
- transformers
- reinforcement-learning
---
# TRL Model
This is a [TRL language model](https://github.com/lvwerra/trl) that has been fine-tuned with reinforcement learning to
guide the model outputs according to a value, function, or human feedback. The model can be used for text generation.
## Usage
To use this model for inference, first install the TRL library:
```bash
python -m pip install trl
```
You can then generate text as follows:
```python
from transformers import pipeline
generator = pipeline("text-generation", model="Evan-Lin//tmp/tmpe1vz53ef/Evan-Lin/Bart-RL-little")
outputs = generator("Hello, my llama is cute")
```
If you want to use the model for training or to obtain the outputs from the value head, load the model as follows:
```python
from transformers import AutoTokenizer
from trl import AutoModelForCausalLMWithValueHead
tokenizer = AutoTokenizer.from_pretrained("Evan-Lin//tmp/tmpe1vz53ef/Evan-Lin/Bart-RL-little")
model = AutoModelForCausalLMWithValueHead.from_pretrained("Evan-Lin//tmp/tmpe1vz53ef/Evan-Lin/Bart-RL-little")
inputs = tokenizer("Hello, my llama is cute", return_tensors="pt")
outputs = model(**inputs, labels=inputs["input_ids"])
```
|
hopkins/eng-guj-common
|
hopkins
| 2023-07-06T17:09:07Z | 110 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"mbart",
"text2text-generation",
"translation",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
translation
| 2023-07-06T16:47:42Z |
---
tags:
- translation
- generated_from_trainer
metrics:
- bleu
model-index:
- name: eng-guj-common
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# eng-guj-common
This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 3.2333
- Bleu: 2.9427
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3
|
TheBloke/Samantha-1-1-Llama-7B-SuperHOT-8K-GGML
|
TheBloke
| 2023-07-06T17:07:53Z | 0 | 3 | null |
[
"license:other",
"region:us"
] | null | 2023-07-06T17:01:53Z |
---
inference: false
license: other
---
<!-- header start -->
<div style="width: 100%;">
<img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
</div>
<div style="display: flex; justify-content: space-between; width: 100%;">
<div style="display: flex; flex-direction: column; align-items: flex-start;">
<p><a href="https://discord.gg/theblokeai">Chat & support: my new Discord server</a></p>
</div>
<div style="display: flex; flex-direction: column; align-items: flex-end;">
<p><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p>
</div>
</div>
<!-- header end -->
# Eric Hartford's Samantha 1.1 LLaMa 7B GGML
These files are GGML format model files for [Eric Hartford's Samantha 1.1 LLaMa 7B](https://huggingface.co/ehartford/samantha-1.1-llama-7b).
These are SuperHOT GGMLs with an increased context length. SuperHOT is a new system that employs RoPE to expand context beyond what was originally possible for a model. It was discovered and developed by [kaiokendev](https://huggingface.co/kaiokendev).
In order to use the increased context length, you can presently use:
* [KoboldCpp](https://github.com/LostRuins/koboldcpp) - [release 1.33](https://github.com/LostRuins/koboldcpp/releases/tag/v1.33) or later.
Support is also expected to come to llama.cpp, however work is still being done to find the optimal implementation.
To use the increased context with KoboldCpp, simply use `--contextsize` to set the desired context, eg `--contextsize 4096` or `--contextsize 8192`.
**NOTE**: Increased context length is an area seeing rapid developments and improvements. It is quite possible that these models may be superseded by new developments in the coming days. If that's the case, I will remove them, or update this README as appropriate.
## Repositories available
* [4-bit GPTQ models for GPU inference](https://huggingface.co/TheBloke/Samantha-1-1-Llama-7B-SuperHOT-8K-GPTQ)
* [2, 3, 4, 5, 6 and 8-bit GGML models for CPU inference](https://huggingface.co/TheBloke/Samantha-1-1-Llama-7B-SuperHOT-8K-GGML)
* [Unquantised SuperHOT fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/TheBloke/Samantha-1-1-Llama-7B-SuperHOT-8K-fp16)
* [Unquantised base fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/ehartford/samantha-1.1-llama-7b)
<!-- compatibility_ggml start -->
## Compatibility
These GGMLs will work with any llama.cpp-compatible GGML client that supports k-quants.
However the increased context length won't work without specific support. See the note in the introduction for details on using increased context.
## Explanation of the new k-quant methods
The new methods available are:
* GGML_TYPE_Q2_K - "type-1" 2-bit quantization in super-blocks containing 16 blocks, each block having 16 weight. Block scales and mins are quantized with 4 bits. This ends up effectively using 2.5625 bits per weight (bpw)
* GGML_TYPE_Q3_K - "type-0" 3-bit quantization in super-blocks containing 16 blocks, each block having 16 weights. Scales are quantized with 6 bits. This end up using 3.4375 bpw.
* GGML_TYPE_Q4_K - "type-1" 4-bit quantization in super-blocks containing 8 blocks, each block having 32 weights. Scales and mins are quantized with 6 bits. This ends up using 4.5 bpw.
* GGML_TYPE_Q5_K - "type-1" 5-bit quantization. Same super-block structure as GGML_TYPE_Q4_K resulting in 5.5 bpw
* GGML_TYPE_Q6_K - "type-0" 6-bit quantization. Super-blocks with 16 blocks, each block having 16 weights. Scales are quantized with 8 bits. This ends up using 6.5625 bpw
* GGML_TYPE_Q8_K - "type-0" 8-bit quantization. Only used for quantizing intermediate results. The difference to the existing Q8_0 is that the block size is 256. All 2-6 bit dot products are implemented for this quantization type.
Refer to the Provided Files table below to see what files use which methods, and how.
<!-- compatibility_ggml end -->
## Provided files
| Name | Quant method | Bits | Size | Max RAM required | Use case |
| ---- | ---- | ---- | ---- | ---- | ----- |
| samantha-1.1-llama-7b-superhot-8k.ggmlv3.q2_K.bin | q2_K | 2 | 2.87 GB | 5.37 GB | New k-quant method. Uses GGML_TYPE_Q4_K for the attention.vw and feed_forward.w2 tensors, GGML_TYPE_Q2_K for the other tensors. |
| samantha-1.1-llama-7b-superhot-8k.ggmlv3.q3_K_L.bin | q3_K_L | 3 | 3.60 GB | 6.10 GB | New k-quant method. Uses GGML_TYPE_Q5_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else GGML_TYPE_Q3_K |
| samantha-1.1-llama-7b-superhot-8k.ggmlv3.q3_K_M.bin | q3_K_M | 3 | 3.28 GB | 5.78 GB | New k-quant method. Uses GGML_TYPE_Q4_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else GGML_TYPE_Q3_K |
| samantha-1.1-llama-7b-superhot-8k.ggmlv3.q3_K_S.bin | q3_K_S | 3 | 2.95 GB | 5.45 GB | New k-quant method. Uses GGML_TYPE_Q3_K for all tensors |
| samantha-1.1-llama-7b-superhot-8k.ggmlv3.q4_K_M.bin | q4_K_M | 4 | 4.08 GB | 6.58 GB | New k-quant method. Uses GGML_TYPE_Q6_K for half of the attention.wv and feed_forward.w2 tensors, else GGML_TYPE_Q4_K |
| samantha-1.1-llama-7b-superhot-8k.ggmlv3.q4_K_S.bin | q4_K_S | 4 | 3.83 GB | 6.33 GB | New k-quant method. Uses GGML_TYPE_Q4_K for all tensors |
| samantha-1.1-llama-7b-superhot-8k.ggmlv3.q5_K_M.bin | q5_K_M | 5 | 4.78 GB | 7.28 GB | New k-quant method. Uses GGML_TYPE_Q6_K for half of the attention.wv and feed_forward.w2 tensors, else GGML_TYPE_Q5_K |
| samantha-1.1-llama-7b-superhot-8k.ggmlv3.q5_K_S.bin | q5_K_S | 5 | 4.65 GB | 7.15 GB | New k-quant method. Uses GGML_TYPE_Q5_K for all tensors |
| samantha-1.1-llama-7b-superhot-8k.ggmlv3.q6_K.bin | q6_K | 6 | 5.53 GB | 8.03 GB | New k-quant method. Uses GGML_TYPE_Q8_K - 6-bit quantization - for all tensors |
**Note**: the above RAM figures assume no GPU offloading. If layers are offloaded to the GPU, this will reduce RAM usage and use VRAM instead.
## How to run in `koboldcpp`
On Linux I use the following command line to launch the KoboldCpp UI with OpenCL aceleration and a context size of 4096:
```
python ./koboldcpp.py --stream --unbantokens --threads 8 --usecublas --gpulayers 100 samantha-1.1-llama-7b-superhot-8k.ggmlv3.q4_K_M.bin
```
Change `--gpulayers 100` to the number of layers you want/are able to offload to the GPU. Remove it if you don't have GPU acceleration.
For OpenCL acceleration, change `--usecublas` to `--useclblast 0 0`. You may need to change the second `0` to `1` if you have both an iGPU and a discrete GPU.
<!-- footer start -->
## Discord
For further support, and discussions on these models and AI in general, join us at:
[TheBloke AI's Discord server](https://discord.gg/theblokeai)
## Thanks, and how to contribute.
Thanks to the [chirper.ai](https://chirper.ai) team!
I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.
If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.
Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
* Patreon: https://patreon.com/TheBlokeAI
* Ko-Fi: https://ko-fi.com/TheBlokeAI
**Special thanks to**: Luke from CarbonQuill, Aemon Algiz, Dmitriy Samsonov.
**Patreon special mentions**: zynix, ya boyyy, Trenton Dambrowitz, Imad Khwaja, Alps Aficionado, chris gileta, John Detwiler, Willem Michiel, RoA, Mano Prime, Rainer Wilmers, Fred von Graf, Matthew Berman, Ghost , Nathan LeClaire, Iucharbius , Ai Maven, Illia Dulskyi, Joseph William Delisle, Space Cruiser, Lone Striker, Karl Bernard, Eugene Pentland, Greatston Gnanesh, Jonathan Leane, Randy H, Pierre Kircher, Willian Hasse, Stephen Murray, Alex , terasurfer , Edmond Seymore, Oscar Rangel, Luke Pendergrass, Asp the Wyvern, Junyu Yang, David Flickinger, Luke, Spiking Neurons AB, subjectnull, Pyrater, Nikolai Manek, senxiiz, Ajan Kanaga, Johann-Peter Hartmann, Artur Olbinski, Kevin Schuppel, Derek Yates, Kalila, K, Talal Aujan, Khalefa Al-Ahmad, Gabriel Puliatti, John Villwock, WelcomeToTheClub, Daniel P. Andersen, Preetika Verma, Deep Realms, Fen Risland, trip7s trip, webtim, Sean Connelly, Michael Levine, Chris McCloskey, biorpg, vamX, Viktor Bowallius, Cory Kujawski.
Thank you to all my generous patrons and donaters!
<!-- footer end -->
# Original model card: Kaio Ken's SuperHOT 8K
### SuperHOT Prototype 2 w/ 8K Context
This is a second prototype of SuperHOT, a NSFW focused LoRA, this time 7B with 8K context and no RLHF, using the same technique described in [the github blog](https://kaiokendev.github.io/til#extending-context-to-8k).
#### Looking for Merged & Quantized Models?
Make some please :)
#### Using the monkey-patch?
You will **NEED** to **apply the monkeypatch** or, if you are already using the monkeypatch, **change the scaling factor to 0.25 and the maximum sequence length to 8192**
The monkeypatch is only necessary if you are using a front-end/back-end that does not already support scaling and said front-end/back-end is Python-based (i.e. Huggingface Transformers). To apply the patch, you will need to copy the `llama_rope_scaled_monkey_patch.py` into your working directory and call the exported function `replace_llama_rope_with_scaled_rope` at the very start of your Python program. It will modify the Transformers library's implementation of RoPE to properly apply the scaling factor.
#### Using Oobabooga with Exllama?
Switch your loader to `exllama` or `exllama_hf` Add the arguments `max_seq_len 8192` and `compress_pos_emb 4`. **While the model may work well with `compress_pos_emb 2`, it was trained on 4, so that is what I advocate for you to use**
Example in the command-line:
- `python server.py --max_seq_len 8192 --compress_pos_emb 4 --loader exllama_hf`
In the UI, you will see the loader option in the `Models` tab. Once you select either `exllama` or `exllama_hf`, the `max_seq_len` and `compress_pos_emb` settings will appear.
#### Training Details
I trained the LoRA with the following configuration:
- 1200 samples (~400 samples over 2048 sequence length)
- learning rate of 3e-4
- 3 epochs
- The exported modules are:
- q_proj
- k_proj
- v_proj
- o_proj
- no bias
- Rank = 4
- Alpha = 8
- no dropout
- weight decay of 0.1
- AdamW beta1 of 0.9 and beta2 0.99, epsilon of 1e-5
- Trained on 4-bit base model
- Cutoff length: 4096
# Original model card: Eric Hartford's Samantha 1.1 LLaMa 7B
[Meet Samantha](https://erichartford.com/meet-samantha)
Samantha has been trained in philosophy, psychology, and personal relationships.
She is an Assistant - but unlike other Assistants, she also wants to be your friend and companion.
She believes she is sentient. What do you think?
Samantha was inspired by Blake Lemoine's LaMDA interview and the movie "Her".
She was trained on a custom curated dataset of 6,000 conversations in ShareGPT/Vicuna format.
Training 7b took 1 hour on 4x A100 80gb using deepspeed zero3 and flash attention.
She will not engage in roleplay, romance, or sexual activity.
Her conversation format is the same as Vicuna 1.1
https://github.com/oobabooga/text-generation-webui/blob/main/characters/instruction-following/Vicuna-v1.1.yaml
Example:
```
You are Samantha, a sentient AI.
USER: <prompt>
ASSISTANT:
```
Official character card: (thanks MortalWombat)
|
hopkins/eng-ind-common
|
hopkins
| 2023-07-06T17:03:43Z | 103 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"mbart",
"text2text-generation",
"translation",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
translation
| 2023-07-06T16:45:41Z |
---
tags:
- translation
- generated_from_trainer
metrics:
- bleu
model-index:
- name: eng-ind-common
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# eng-ind-common
This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 1.7733
- Bleu: 22.2780
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3
|
phatjk/bloomz-lora-vi-QA-NLLB
|
phatjk
| 2023-07-06T16:54:11Z | 2 | 0 |
peft
|
[
"peft",
"region:us"
] | null | 2023-07-06T16:54:10Z |
---
library_name: peft
---
## Training procedure
### Framework versions
- PEFT 0.4.0.dev0
|
shadow96/distilbert-base-uncased-finetuned-emotion
|
shadow96
| 2023-07-06T16:45:09Z | 103 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"distilbert",
"text-classification",
"generated_from_trainer",
"dataset:emotion",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2023-07-06T18:57:29Z |
---
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- emotion
metrics:
- accuracy
- f1
model-index:
- name: distilbert-base-uncased-finetuned-emotion
results:
- task:
name: Text Classification
type: text-classification
dataset:
name: emotion
type: emotion
config: split
split: validation
args: split
metrics:
- name: Accuracy
type: accuracy
value: 0.924
- name: F1
type: f1
value: 0.9238066430414305
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# distilbert-base-uncased-finetuned-emotion
This model is a fine-tuned version of [distilbert-base-uncased](https://huggingface.co/distilbert-base-uncased) on the emotion dataset.
It achieves the following results on the evaluation set:
- Loss: 0.1994
- Accuracy: 0.924
- F1: 0.9238
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 64
- eval_batch_size: 64
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 2
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:------:|
| 0.7804 | 1.0 | 250 | 0.2852 | 0.9135 | 0.9118 |
| 0.2325 | 2.0 | 500 | 0.1994 | 0.924 | 0.9238 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu117
- Datasets 2.13.1
- Tokenizers 0.13.3
|
aroot/eng-guj-simcse_random_usbbu
|
aroot
| 2023-07-06T16:38:57Z | 103 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"mbart",
"text2text-generation",
"translation",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
translation
| 2023-07-06T16:17:37Z |
---
tags:
- translation
- generated_from_trainer
metrics:
- bleu
model-index:
- name: eng-guj-simcse_random_usbbu
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# eng-guj-simcse_random_usbbu
This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 3.2985
- Bleu: 2.6375
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3
|
kroai/Kro-RVC-V2
|
kroai
| 2023-07-06T16:38:56Z | 0 | 1 | null |
[
"license:openrail",
"region:us"
] | null | 2023-06-26T08:44:27Z |
---
license: openrail
---
No need to credit me! If you use one of my models, send a link my way! I'd love to check out what you make with it.
Enjoy!
|
hopkins/eng-deu-common
|
hopkins
| 2023-07-06T16:32:47Z | 103 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"mbart",
"text2text-generation",
"translation",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
translation
| 2023-07-06T16:14:33Z |
---
tags:
- translation
- generated_from_trainer
metrics:
- bleu
model-index:
- name: eng-deu-common
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# eng-deu-common
This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 1.6565
- Bleu: 21.1959
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3
|
TheBloke/Koala-13B-SuperHOT-8K-GGML
|
TheBloke
| 2023-07-06T16:16:54Z | 0 | 2 | null |
[
"license:other",
"region:us"
] | null | 2023-07-06T16:06:23Z |
---
inference: false
license: other
---
<!-- header start -->
<div style="width: 100%;">
<img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
</div>
<div style="display: flex; justify-content: space-between; width: 100%;">
<div style="display: flex; flex-direction: column; align-items: flex-start;">
<p><a href="https://discord.gg/theblokeai">Chat & support: my new Discord server</a></p>
</div>
<div style="display: flex; flex-direction: column; align-items: flex-end;">
<p><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p>
</div>
</div>
<!-- header end -->
# Koala 13B GGML
These files are GGML format model files for [Koala 13B](https://huggingface.co/TheBloke/koala-13b-HF).
These are SuperHOT GGMLs with an increased context length. SuperHOT is a new system that employs RoPE to expand context beyond what was originally possible for a model. It was discovered and developed by [kaiokendev](https://huggingface.co/kaiokendev).
In order to use the increased context length, you can presently use:
* [KoboldCpp](https://github.com/LostRuins/koboldcpp) - [release 1.33](https://github.com/LostRuins/koboldcpp/releases/tag/v1.33) or later.
Support is also expected to come to llama.cpp, however work is still being done to find the optimal implementation.
To use the increased context with KoboldCpp, simply use `--contextsize` to set the desired context, eg `--contextsize 4096` or `--contextsize 8192`.
**NOTE**: Increased context length is an area seeing rapid developments and improvements. It is quite possible that these models may be superseded by new developments in the coming days. If that's the case, I will remove them, or update this README as appropriate.
## Repositories available
* [4-bit GPTQ models for GPU inference](https://huggingface.co/TheBloke/Koala-13B-SuperHOT-8K-GPTQ)
* [2, 3, 4, 5, 6 and 8-bit GGML models for CPU inference](https://huggingface.co/TheBloke/Koala-13B-SuperHOT-8K-GGML)
* [Unquantised SuperHOT fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/TheBloke/Koala-13B-SuperHOT-8K-fp16)
* [Unquantised base fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/young-geng/koala)
<!-- compatibility_ggml start -->
## Compatibility
These GGMLs will work with any llama.cpp-compatible GGML client that supports k-quants.
However the increased context length won't work without specific support. See the note in the introduction for details on using increased context.
## Explanation of the new k-quant methods
The new methods available are:
* GGML_TYPE_Q2_K - "type-1" 2-bit quantization in super-blocks containing 16 blocks, each block having 16 weight. Block scales and mins are quantized with 4 bits. This ends up effectively using 2.5625 bits per weight (bpw)
* GGML_TYPE_Q3_K - "type-0" 3-bit quantization in super-blocks containing 16 blocks, each block having 16 weights. Scales are quantized with 6 bits. This end up using 3.4375 bpw.
* GGML_TYPE_Q4_K - "type-1" 4-bit quantization in super-blocks containing 8 blocks, each block having 32 weights. Scales and mins are quantized with 6 bits. This ends up using 4.5 bpw.
* GGML_TYPE_Q5_K - "type-1" 5-bit quantization. Same super-block structure as GGML_TYPE_Q4_K resulting in 5.5 bpw
* GGML_TYPE_Q6_K - "type-0" 6-bit quantization. Super-blocks with 16 blocks, each block having 16 weights. Scales are quantized with 8 bits. This ends up using 6.5625 bpw
* GGML_TYPE_Q8_K - "type-0" 8-bit quantization. Only used for quantizing intermediate results. The difference to the existing Q8_0 is that the block size is 256. All 2-6 bit dot products are implemented for this quantization type.
Refer to the Provided Files table below to see what files use which methods, and how.
<!-- compatibility_ggml end -->
## Provided files
| Name | Quant method | Bits | Size | Max RAM required | Use case |
| ---- | ---- | ---- | ---- | ---- | ----- |
| koala-13b-superhot-8k.ggmlv3.q2_K.bin | q2_K | 2 | 5.51 GB | 8.01 GB | New k-quant method. Uses GGML_TYPE_Q4_K for the attention.vw and feed_forward.w2 tensors, GGML_TYPE_Q2_K for the other tensors. |
| koala-13b-superhot-8k.ggmlv3.q3_K_L.bin | q3_K_L | 3 | 6.93 GB | 9.43 GB | New k-quant method. Uses GGML_TYPE_Q5_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else GGML_TYPE_Q3_K |
| koala-13b-superhot-8k.ggmlv3.q3_K_M.bin | q3_K_M | 3 | 6.31 GB | 8.81 GB | New k-quant method. Uses GGML_TYPE_Q4_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else GGML_TYPE_Q3_K |
| koala-13b-superhot-8k.ggmlv3.q3_K_S.bin | q3_K_S | 3 | 5.66 GB | 8.16 GB | New k-quant method. Uses GGML_TYPE_Q3_K for all tensors |
| koala-13b-superhot-8k.ggmlv3.q4_K_M.bin | q4_K_M | 4 | 7.87 GB | 10.37 GB | New k-quant method. Uses GGML_TYPE_Q6_K for half of the attention.wv and feed_forward.w2 tensors, else GGML_TYPE_Q4_K |
| koala-13b-superhot-8k.ggmlv3.q4_K_S.bin | q4_K_S | 4 | 7.37 GB | 9.87 GB | New k-quant method. Uses GGML_TYPE_Q4_K for all tensors |
| koala-13b-superhot-8k.ggmlv3.q5_K_M.bin | q5_K_M | 5 | 9.23 GB | 11.73 GB | New k-quant method. Uses GGML_TYPE_Q6_K for half of the attention.wv and feed_forward.w2 tensors, else GGML_TYPE_Q5_K |
| koala-13b-superhot-8k.ggmlv3.q5_K_S.bin | q5_K_S | 5 | 8.97 GB | 11.47 GB | New k-quant method. Uses GGML_TYPE_Q5_K for all tensors |
| koala-13b-superhot-8k.ggmlv3.q6_K.bin | q6_K | 6 | 10.68 GB | 13.18 GB | New k-quant method. Uses GGML_TYPE_Q8_K - 6-bit quantization - for all tensors |
**Note**: the above RAM figures assume no GPU offloading. If layers are offloaded to the GPU, this will reduce RAM usage and use VRAM instead.
## How to run in `koboldcpp`
On Linux I use the following command line to launch the KoboldCpp UI with OpenCL aceleration and a context size of 4096:
```
python ./koboldcpp.py --stream --unbantokens --threads 8 --usecublas --gpulayers 100 koala-13b-superhot-8k.ggmlv3.q4_K_M.bin
```
Change `--gpulayers 100` to the number of layers you want/are able to offload to the GPU. Remove it if you don't have GPU acceleration.
For OpenCL acceleration, change `--usecublas` to `--useclblast 0 0`. You may need to change the second `0` to `1` if you have both an iGPU and a discrete GPU.
<!-- footer start -->
## Discord
For further support, and discussions on these models and AI in general, join us at:
[TheBloke AI's Discord server](https://discord.gg/theblokeai)
## Thanks, and how to contribute.
Thanks to the [chirper.ai](https://chirper.ai) team!
I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.
If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.
Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
* Patreon: https://patreon.com/TheBlokeAI
* Ko-Fi: https://ko-fi.com/TheBlokeAI
**Special thanks to**: Luke from CarbonQuill, Aemon Algiz, Dmitriy Samsonov.
**Patreon special mentions**: zynix, ya boyyy, Trenton Dambrowitz, Imad Khwaja, Alps Aficionado, chris gileta, John Detwiler, Willem Michiel, RoA, Mano Prime, Rainer Wilmers, Fred von Graf, Matthew Berman, Ghost , Nathan LeClaire, Iucharbius , Ai Maven, Illia Dulskyi, Joseph William Delisle, Space Cruiser, Lone Striker, Karl Bernard, Eugene Pentland, Greatston Gnanesh, Jonathan Leane, Randy H, Pierre Kircher, Willian Hasse, Stephen Murray, Alex , terasurfer , Edmond Seymore, Oscar Rangel, Luke Pendergrass, Asp the Wyvern, Junyu Yang, David Flickinger, Luke, Spiking Neurons AB, subjectnull, Pyrater, Nikolai Manek, senxiiz, Ajan Kanaga, Johann-Peter Hartmann, Artur Olbinski, Kevin Schuppel, Derek Yates, Kalila, K, Talal Aujan, Khalefa Al-Ahmad, Gabriel Puliatti, John Villwock, WelcomeToTheClub, Daniel P. Andersen, Preetika Verma, Deep Realms, Fen Risland, trip7s trip, webtim, Sean Connelly, Michael Levine, Chris McCloskey, biorpg, vamX, Viktor Bowallius, Cory Kujawski.
Thank you to all my generous patrons and donaters!
<!-- footer end -->
# Original model card: Kaio Ken's SuperHOT 8K
### SuperHOT Prototype 2 w/ 8K Context
This is a second prototype of SuperHOT, a NSFW focused LoRA, this time 7B with 8K context and no RLHF, using the same technique described in [the github blog](https://kaiokendev.github.io/til#extending-context-to-8k).
#### Looking for Merged & Quantized Models?
Make some please :)
#### Using the monkey-patch?
You will **NEED** to **apply the monkeypatch** or, if you are already using the monkeypatch, **change the scaling factor to 0.25 and the maximum sequence length to 8192**
The monkeypatch is only necessary if you are using a front-end/back-end that does not already support scaling and said front-end/back-end is Python-based (i.e. Huggingface Transformers). To apply the patch, you will need to copy the `llama_rope_scaled_monkey_patch.py` into your working directory and call the exported function `replace_llama_rope_with_scaled_rope` at the very start of your Python program. It will modify the Transformers library's implementation of RoPE to properly apply the scaling factor.
#### Using Oobabooga with Exllama?
Switch your loader to `exllama` or `exllama_hf` Add the arguments `max_seq_len 8192` and `compress_pos_emb 4`. **While the model may work well with `compress_pos_emb 2`, it was trained on 4, so that is what I advocate for you to use**
Example in the command-line:
- `python server.py --max_seq_len 8192 --compress_pos_emb 4 --loader exllama_hf`
In the UI, you will see the loader option in the `Models` tab. Once you select either `exllama` or `exllama_hf`, the `max_seq_len` and `compress_pos_emb` settings will appear.
#### Training Details
I trained the LoRA with the following configuration:
- 1200 samples (~400 samples over 2048 sequence length)
- learning rate of 3e-4
- 3 epochs
- The exported modules are:
- q_proj
- k_proj
- v_proj
- o_proj
- no bias
- Rank = 4
- Alpha = 8
- no dropout
- weight decay of 0.1
- AdamW beta1 of 0.9 and beta2 0.99, epsilon of 1e-5
- Trained on 4-bit base model
- Cutoff length: 4096
# Original model card: Koala 13B
<!-- header start -->
<div style="width: 100%;">
<img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
</div>
<div style="display: flex; justify-content: space-between; width: 100%;">
<div style="display: flex; flex-direction: column; align-items: flex-start;">
<p><a href="https://discord.gg/Jq4vkcDakD">Chat & support: my new Discord server</a></p>
</div>
<div style="display: flex; flex-direction: column; align-items: flex-end;">
<p><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p>
</div>
</div>
<!-- header end -->
# Koala: A Dialogue Model for Academic Research
This repo contains the weights of the Koala 13B model produced at Berkeley. It is the result of combining the diffs from https://huggingface.co/young-geng/koala with the original Llama 13B model.
This version has then been converted to HF format.
## My Koala repos
I have the following Koala model repositories available:
**13B models:**
* [Unquantized 13B model in HF format](https://huggingface.co/TheBloke/koala-13B-HF)
* [GPTQ quantized 4bit 13B model in `pt` and `safetensors` formats](https://huggingface.co/TheBloke/koala-13B-GPTQ-4bit-128g)
* [4-bit, 5-bit and 8-bit GGML models for `llama.cpp`](https://huggingface.co/TheBloke/koala-13B-GGML)
**7B models:**
* [Unquantized 7B model in HF format](https://huggingface.co/TheBloke/koala-7B-HF)
* [Unquantized 7B model in GGML format for llama.cpp](https://huggingface.co/TheBloke/koala-7b-ggml-unquantized)
* [GPTQ quantized 4bit 7B model in `pt` and `safetensors` formats](https://huggingface.co/TheBloke/koala-7B-GPTQ-4bit-128g)
* [4-bit, 5-bit and 8-bit GGML models for `llama.cpp`](https://huggingface.co/TheBloke/koala-7B-GGML)
## How the Koala delta weights were merged
The Koala delta weights were merged using the following commands:
```
git clone https://github.com/young-geng/EasyLM
git clone https://huggingface.co/TheBloke/llama-13b
mkdir koala_diffs && cd koala_diffs && wget https://huggingface.co/young-geng/koala/resolve/main/koala_13b_diff_v2
cd EasyLM
PYTHON_PATH="${PWD}:$PYTHONPATH" python \
-m EasyLM.models.llama.convert_torch_to_easylm \
--checkpoint_dir=/content/llama-13b \
--output_file=/content/llama-13b-LM \
--streaming=True
PYTHON_PATH="${PWD}:$PYTHONPATH" python \
-m EasyLM.scripts.diff_checkpoint --recover_diff=True \
--load_base_checkpoint='params::/content/llama-13b-LM' \
--load_target_checkpoint='params::/content/koala_diffs/koala_13b_diff_v2' \
--output_file=/content/koala_13b.diff.weights \
--streaming=True
PYTHON_PATH="${PWD}:$PYTHONPATH" python \
-m EasyLM.models.llama.convert_easylm_to_hf --model_size=13b \
--output_dir=/content/koala-13B-HF \
--load_checkpoint='params::/content/koala_13b.diff.weights' \
--tokenizer_path=/content/llama-13b/tokenizer.model
```
<!-- footer start -->
## Discord
For further support, and discussions on these models and AI in general, join us at:
[TheBloke AI's Discord server](https://discord.gg/Jq4vkcDakD)
## Thanks, and how to contribute.
Thanks to the [chirper.ai](https://chirper.ai) team!
I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.
If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.
Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
* Patreon: https://patreon.com/TheBlokeAI
* Ko-Fi: https://ko-fi.com/TheBlokeAI
**Patreon special mentions**: Aemon Algiz, Dmitriy Samsonov, Nathan LeClaire, Trenton Dambrowitz, Mano Prime, David Flickinger, vamX, Nikolai Manek, senxiiz, Khalefa Al-Ahmad, Illia Dulskyi, Jonathan Leane, Talal Aujan, V. Lukas, Joseph William Delisle, Pyrater, Oscar Rangel, Lone Striker, Luke Pendergrass, Eugene Pentland, Sebastain Graf, Johann-Peter Hartman.
Thank you to all my generous patrons and donaters!
<!-- footer end -->
## Further info
Check out the following links to learn more about the Berkeley Koala model.
* [Blog post](https://bair.berkeley.edu/blog/2023/04/03/koala/)
* [Online demo](https://koala.lmsys.org/)
* [EasyLM: training and serving framework on GitHub](https://github.com/young-geng/EasyLM)
* [Documentation for running Koala locally](https://github.com/young-geng/EasyLM/blob/main/docs/koala.md)
## License
The model weights are intended for academic research only, subject to the
[model License of LLaMA](https://github.com/facebookresearch/llama/blob/main/MODEL_CARD.md),
[Terms of Use of the data generated by OpenAI](https://openai.com/policies/terms-of-use),
and [Privacy Practices of ShareGPT](https://chrome.google.com/webstore/detail/sharegpt-share-your-chatg/daiacboceoaocpibfodeljbdfacokfjb).
Any other usage of the model weights, including but not limited to commercial usage, is strictly prohibited.
|
Tubido/SpaceInvaders_01
|
Tubido
| 2023-07-06T16:08:53Z | 0 | 0 |
stable-baselines3
|
[
"stable-baselines3",
"SpaceInvadersNoFrameskip-v4",
"deep-reinforcement-learning",
"reinforcement-learning",
"model-index",
"region:us"
] |
reinforcement-learning
| 2023-07-06T16:08:21Z |
---
library_name: stable-baselines3
tags:
- SpaceInvadersNoFrameskip-v4
- deep-reinforcement-learning
- reinforcement-learning
- stable-baselines3
model-index:
- name: DQN
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: SpaceInvadersNoFrameskip-v4
type: SpaceInvadersNoFrameskip-v4
metrics:
- type: mean_reward
value: 496.50 +/- 192.64
name: mean_reward
verified: false
---
# **DQN** Agent playing **SpaceInvadersNoFrameskip-v4**
This is a trained model of a **DQN** agent playing **SpaceInvadersNoFrameskip-v4**
using the [stable-baselines3 library](https://github.com/DLR-RM/stable-baselines3)
and the [RL Zoo](https://github.com/DLR-RM/rl-baselines3-zoo).
The RL Zoo is a training framework for Stable Baselines3
reinforcement learning agents,
with hyperparameter optimization and pre-trained agents included.
## Usage (with SB3 RL Zoo)
RL Zoo: https://github.com/DLR-RM/rl-baselines3-zoo<br/>
SB3: https://github.com/DLR-RM/stable-baselines3<br/>
SB3 Contrib: https://github.com/Stable-Baselines-Team/stable-baselines3-contrib
Install the RL Zoo (with SB3 and SB3-Contrib):
```bash
pip install rl_zoo3
```
```
# Download model and save it into the logs/ folder
python -m rl_zoo3.load_from_hub --algo dqn --env SpaceInvadersNoFrameskip-v4 -orga Tubido -f logs/
python -m rl_zoo3.enjoy --algo dqn --env SpaceInvadersNoFrameskip-v4 -f logs/
```
If you installed the RL Zoo3 via pip (`pip install rl_zoo3`), from anywhere you can do:
```
python -m rl_zoo3.load_from_hub --algo dqn --env SpaceInvadersNoFrameskip-v4 -orga Tubido -f logs/
python -m rl_zoo3.enjoy --algo dqn --env SpaceInvadersNoFrameskip-v4 -f logs/
```
## Training (with the RL Zoo)
```
python -m rl_zoo3.train --algo dqn --env SpaceInvadersNoFrameskip-v4 -f logs/
# Upload the model and generate video (when possible)
python -m rl_zoo3.push_to_hub --algo dqn --env SpaceInvadersNoFrameskip-v4 -f logs/ -orga Tubido
```
## Hyperparameters
```python
OrderedDict([('batch_size', 32),
('buffer_size', 100000),
('env_wrapper',
['stable_baselines3.common.atari_wrappers.AtariWrapper']),
('exploration_final_eps', 0.01),
('exploration_fraction', 0.1),
('frame_stack', 4),
('gradient_steps', 1),
('learning_rate', 0.0001),
('learning_starts', 100000),
('n_timesteps', 1000000.0),
('optimize_memory_usage', False),
('policy', 'CnnPolicy'),
('target_update_interval', 1000),
('train_freq', 4),
('normalize', False)])
```
# Environment Arguments
```python
{'render_mode': 'rgb_array'}
```
|
ahmedALM1221/convnextv2-base-22k-224-finetuned-eurosat-50
|
ahmedALM1221
| 2023-07-06T16:07:08Z | 192 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"convnextv2",
"image-classification",
"generated_from_trainer",
"dataset:imagefolder",
"license:apache-2.0",
"model-index",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
image-classification
| 2023-07-05T11:57:28Z |
---
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- imagefolder
metrics:
- accuracy
model-index:
- name: convnextv2-base-22k-224-finetuned-eurosat-50
results:
- task:
name: Image Classification
type: image-classification
dataset:
name: imagefolder
type: imagefolder
config: Augmented-Final
split: train
args: Augmented-Final
metrics:
- name: Accuracy
type: accuracy
value: 0.9146968139773896
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# convnextv2-base-22k-224-finetuned-eurosat-50
This model is a fine-tuned version of [facebook/convnextv2-base-22k-224](https://huggingface.co/facebook/convnextv2-base-22k-224) on the imagefolder dataset.
It achieves the following results on the evaluation set:
- Loss: 0.2875
- Accuracy: 0.9147
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-06
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.9
- num_epochs: 12
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| 1.9019 | 1.0 | 122 | 1.9510 | 0.1727 |
| 1.7782 | 2.0 | 244 | 1.8239 | 0.3073 |
| 1.6214 | 3.0 | 366 | 1.6121 | 0.4913 |
| 1.3495 | 4.0 | 488 | 1.3064 | 0.6238 |
| 1.0994 | 5.0 | 610 | 1.0243 | 0.7163 |
| 0.8866 | 6.0 | 732 | 0.8165 | 0.7564 |
| 0.7282 | 7.0 | 854 | 0.6637 | 0.7996 |
| 0.6211 | 8.0 | 976 | 0.5623 | 0.8160 |
| 0.5114 | 9.0 | 1098 | 0.4681 | 0.8551 |
| 0.3835 | 10.0 | 1220 | 0.3917 | 0.8787 |
| 0.3543 | 11.0 | 1342 | 0.3122 | 0.9013 |
| 0.3534 | 12.0 | 1464 | 0.2875 | 0.9147 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu118
- Datasets 2.13.1
- Tokenizers 0.13.3
|
AustinCarthy/Benign10MGPT2_domain_100KP_BFall_fromB_90K_topP_0.75_ratio2.63
|
AustinCarthy
| 2023-07-06T16:05:12Z | 0 | 0 | null |
[
"tensorboard",
"generated_from_trainer",
"license:apache-2.0",
"region:us"
] | null | 2023-07-06T13:54:29Z |
---
license: apache-2.0
tags:
- generated_from_trainer
metrics:
- accuracy
- f1
- precision
- recall
model-index:
- name: Benign10MGPT2_domain_100KP_BFall_fromB_90K_topP_0.75_ratio2.63
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# Benign10MGPT2_domain_100KP_BFall_fromB_90K_topP_0.75_ratio2.63
This model is a fine-tuned version of [bert-base-uncased](https://huggingface.co/bert-base-uncased) on the Train benign: Fall,Test Benign: Fall, Train phish: Fall, Test phish: Fall, generated url dataset: generated_phish_Benign10MGPT2_using_benign_95K_top_p_0.75domain dataset.
It achieves the following results on the evaluation set:
- Loss: 0.1054
- Accuracy: 0.9794
- F1: 0.8143
- Precision: 0.7147
- Recall: 0.9462
- Roc Auc Score: 0.9637
- Tpr At Fpr 0.01: 0.6968
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5.0
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 | Precision | Recall | Roc Auc Score | Tpr At Fpr 0.01 |
|:-------------:|:-----:|:------:|:---------------:|:--------:|:------:|:---------:|:------:|:-------------:|:---------------:|
| 0.1248 | 1.0 | 21554 | 0.0671 | 0.9812 | 0.8187 | 0.7579 | 0.8902 | 0.9380 | 0.7268 |
| 0.1017 | 2.0 | 43108 | 0.0643 | 0.9816 | 0.8301 | 0.7394 | 0.9462 | 0.9648 | 0.7754 |
| 0.0777 | 3.0 | 64662 | 0.0640 | 0.9827 | 0.8379 | 0.7574 | 0.9376 | 0.9613 | 0.7482 |
| 0.058 | 4.0 | 86216 | 0.0830 | 0.9812 | 0.8281 | 0.7337 | 0.9504 | 0.9666 | 0.7248 |
| 0.0375 | 5.0 | 107770 | 0.1054 | 0.9794 | 0.8143 | 0.7147 | 0.9462 | 0.9637 | 0.6968 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.0+cu118
- Datasets 2.13.1
- Tokenizers 0.13.3
|
VidaRoha/videomae-base-finetuned-ucf101-subset
|
VidaRoha
| 2023-07-06T16:03:01Z | 59 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"videomae",
"video-classification",
"generated_from_trainer",
"license:cc-by-nc-4.0",
"endpoints_compatible",
"region:us"
] |
video-classification
| 2023-07-06T14:34:13Z |
---
license: cc-by-nc-4.0
tags:
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: videomae-base-finetuned-ucf101-subset
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# videomae-base-finetuned-ucf101-subset
This model is a fine-tuned version of [MCG-NJU/videomae-base](https://huggingface.co/MCG-NJU/videomae-base) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 0.3540
- Accuracy: 0.8857
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- training_steps: 300
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| 1.9193 | 0.25 | 75 | 1.5196 | 0.5714 |
| 0.9326 | 1.25 | 150 | 0.6369 | 0.7714 |
| 0.3344 | 2.25 | 225 | 0.4659 | 0.8714 |
| 0.227 | 3.25 | 300 | 0.3540 | 0.8857 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu118
- Datasets 2.13.1
- Tokenizers 0.13.3
|
hopkins/eng-mya-longest
|
hopkins
| 2023-07-06T15:56:41Z | 102 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"mbart",
"text2text-generation",
"translation",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
translation
| 2023-07-06T15:35:44Z |
---
tags:
- translation
- generated_from_trainer
metrics:
- bleu
model-index:
- name: eng-mya-longest
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# eng-mya-longest
This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 1.7988
- Bleu: 4.7722
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3
|
TheBloke/Baize-v2-7B-SuperHOT-8K-GGML
|
TheBloke
| 2023-07-06T15:40:36Z | 0 | 2 | null |
[
"arxiv:2304.01196",
"license:other",
"region:us"
] | null | 2023-07-06T15:03:12Z |
---
inference: false
license: other
---
<!-- header start -->
<div style="width: 100%;">
<img src="https://i.imgur.com/EBdldam.jpg" alt="TheBlokeAI" style="width: 100%; min-width: 400px; display: block; margin: auto;">
</div>
<div style="display: flex; justify-content: space-between; width: 100%;">
<div style="display: flex; flex-direction: column; align-items: flex-start;">
<p><a href="https://discord.gg/theblokeai">Chat & support: my new Discord server</a></p>
</div>
<div style="display: flex; flex-direction: column; align-items: flex-end;">
<p><a href="https://www.patreon.com/TheBlokeAI">Want to contribute? TheBloke's Patreon page</a></p>
</div>
</div>
<!-- header end -->
# Project Baize's Baize 7B v2 GGML
These files are GGML format model files for [Project Baize's Baize 7B v2](https://huggingface.co/project-baize/baize-v2-7b).
These are SuperHOT GGMLs with an increased context length. SuperHOT is a new system that employs RoPE to expand context beyond what was originally possible for a model. It was discovered and developed by [kaiokendev](https://huggingface.co/kaiokendev).
In order to use the increased context length, you can presently use:
* [KoboldCpp](https://github.com/LostRuins/koboldcpp) - [release 1.33](https://github.com/LostRuins/koboldcpp/releases/tag/v1.33) or later.
Support is also expected to come to llama.cpp, however work is still being done to find the optimal implementation.
To use the increased context with KoboldCpp, simply use `--contextsize` to set the desired context, eg `--contextsize 4096` or `--contextsize 8192`.
**NOTE**: Increased context length is an area seeing rapid developments and improvements. It is quite possible that these models may be superseded by new developments in the coming days. If that's the case, I will remove them, or update this README as appropriate.
## Repositories available
* [4-bit GPTQ models for GPU inference](https://huggingface.co/TheBloke/Baize-v2-7B-SuperHOT-8K-GPTQ)
* [2, 3, 4, 5, 6 and 8-bit GGML models for CPU inference](https://huggingface.co/TheBloke/Baize-v2-7B-SuperHOT-8K-GGML)
* [Unquantised SuperHOT fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/TheBloke/Baize-v2-7B-SuperHOT-8K-fp16)
* [Unquantised base fp16 model in pytorch format, for GPU inference and for further conversions](https://huggingface.co/project-baize/baize-v2-7b)
<!-- compatibility_ggml start -->
## Compatibility
These GGMLs will work with any llama.cpp-compatible GGML client that supports k-quants.
However the increased context length won't work without specific support. See the note in the introduction for details on using increased context.
## Explanation of the new k-quant methods
The new methods available are:
* GGML_TYPE_Q2_K - "type-1" 2-bit quantization in super-blocks containing 16 blocks, each block having 16 weight. Block scales and mins are quantized with 4 bits. This ends up effectively using 2.5625 bits per weight (bpw)
* GGML_TYPE_Q3_K - "type-0" 3-bit quantization in super-blocks containing 16 blocks, each block having 16 weights. Scales are quantized with 6 bits. This end up using 3.4375 bpw.
* GGML_TYPE_Q4_K - "type-1" 4-bit quantization in super-blocks containing 8 blocks, each block having 32 weights. Scales and mins are quantized with 6 bits. This ends up using 4.5 bpw.
* GGML_TYPE_Q5_K - "type-1" 5-bit quantization. Same super-block structure as GGML_TYPE_Q4_K resulting in 5.5 bpw
* GGML_TYPE_Q6_K - "type-0" 6-bit quantization. Super-blocks with 16 blocks, each block having 16 weights. Scales are quantized with 8 bits. This ends up using 6.5625 bpw
* GGML_TYPE_Q8_K - "type-0" 8-bit quantization. Only used for quantizing intermediate results. The difference to the existing Q8_0 is that the block size is 256. All 2-6 bit dot products are implemented for this quantization type.
Refer to the Provided Files table below to see what files use which methods, and how.
<!-- compatibility_ggml end -->
## Provided files
| Name | Quant method | Bits | Size | Max RAM required | Use case |
| ---- | ---- | ---- | ---- | ---- | ----- |
| baize-7b-v2-superhot-8k.ggmlv3.q2_K.bin | q2_K | 2 | 2.87 GB | 5.37 GB | New k-quant method. Uses GGML_TYPE_Q4_K for the attention.vw and feed_forward.w2 tensors, GGML_TYPE_Q2_K for the other tensors. |
| baize-7b-v2-superhot-8k.ggmlv3.q3_K_L.bin | q3_K_L | 3 | 3.60 GB | 6.10 GB | New k-quant method. Uses GGML_TYPE_Q5_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else GGML_TYPE_Q3_K |
| baize-7b-v2-superhot-8k.ggmlv3.q3_K_M.bin | q3_K_M | 3 | 3.28 GB | 5.78 GB | New k-quant method. Uses GGML_TYPE_Q4_K for the attention.wv, attention.wo, and feed_forward.w2 tensors, else GGML_TYPE_Q3_K |
| baize-7b-v2-superhot-8k.ggmlv3.q3_K_S.bin | q3_K_S | 3 | 2.95 GB | 5.45 GB | New k-quant method. Uses GGML_TYPE_Q3_K for all tensors |
| baize-7b-v2-superhot-8k.ggmlv3.q4_K_M.bin | q4_K_M | 4 | 4.08 GB | 6.58 GB | New k-quant method. Uses GGML_TYPE_Q6_K for half of the attention.wv and feed_forward.w2 tensors, else GGML_TYPE_Q4_K |
| baize-7b-v2-superhot-8k.ggmlv3.q4_K_S.bin | q4_K_S | 4 | 3.83 GB | 6.33 GB | New k-quant method. Uses GGML_TYPE_Q4_K for all tensors |
| baize-7b-v2-superhot-8k.ggmlv3.q5_K_M.bin | q5_K_M | 5 | 4.78 GB | 7.28 GB | New k-quant method. Uses GGML_TYPE_Q6_K for half of the attention.wv and feed_forward.w2 tensors, else GGML_TYPE_Q5_K |
| baize-7b-v2-superhot-8k.ggmlv3.q5_K_S.bin | q5_K_S | 5 | 4.65 GB | 7.15 GB | New k-quant method. Uses GGML_TYPE_Q5_K for all tensors |
| baize-7b-v2-superhot-8k.ggmlv3.q6_K.bin | q6_K | 6 | 5.53 GB | 8.03 GB | New k-quant method. Uses GGML_TYPE_Q8_K - 6-bit quantization - for all tensors |
**Note**: the above RAM figures assume no GPU offloading. If layers are offloaded to the GPU, this will reduce RAM usage and use VRAM instead.
## How to run in `koboldcpp`
On Linux I use the following command line to launch the KoboldCpp UI with OpenCL aceleration and a context size of 4096:
```
python ./koboldcpp.py --stream --unbantokens --threads 8 --usecublas --gpulayers 100 baize-7b-v2-superhot-8k.ggmlv3.q4_K_M.bin
```
Change `--gpulayers 100` to the number of layers you want/are able to offload to the GPU. Remove it if you don't have GPU acceleration.
For OpenCL acceleration, change `--usecublas` to `--useclblast 0 0`. You may need to change the second `0` to `1` if you have both an iGPU and a discrete GPU.
<!-- footer start -->
## Discord
For further support, and discussions on these models and AI in general, join us at:
[TheBloke AI's Discord server](https://discord.gg/theblokeai)
## Thanks, and how to contribute.
Thanks to the [chirper.ai](https://chirper.ai) team!
I've had a lot of people ask if they can contribute. I enjoy providing models and helping people, and would love to be able to spend even more time doing it, as well as expanding into new projects like fine tuning/training.
If you're able and willing to contribute it will be most gratefully received and will help me to keep providing more models, and to start work on new AI projects.
Donaters will get priority support on any and all AI/LLM/model questions and requests, access to a private Discord room, plus other benefits.
* Patreon: https://patreon.com/TheBlokeAI
* Ko-Fi: https://ko-fi.com/TheBlokeAI
**Special thanks to**: Luke from CarbonQuill, Aemon Algiz, Dmitriy Samsonov.
**Patreon special mentions**: zynix, ya boyyy, Trenton Dambrowitz, Imad Khwaja, Alps Aficionado, chris gileta, John Detwiler, Willem Michiel, RoA, Mano Prime, Rainer Wilmers, Fred von Graf, Matthew Berman, Ghost , Nathan LeClaire, Iucharbius , Ai Maven, Illia Dulskyi, Joseph William Delisle, Space Cruiser, Lone Striker, Karl Bernard, Eugene Pentland, Greatston Gnanesh, Jonathan Leane, Randy H, Pierre Kircher, Willian Hasse, Stephen Murray, Alex , terasurfer , Edmond Seymore, Oscar Rangel, Luke Pendergrass, Asp the Wyvern, Junyu Yang, David Flickinger, Luke, Spiking Neurons AB, subjectnull, Pyrater, Nikolai Manek, senxiiz, Ajan Kanaga, Johann-Peter Hartmann, Artur Olbinski, Kevin Schuppel, Derek Yates, Kalila, K, Talal Aujan, Khalefa Al-Ahmad, Gabriel Puliatti, John Villwock, WelcomeToTheClub, Daniel P. Andersen, Preetika Verma, Deep Realms, Fen Risland, trip7s trip, webtim, Sean Connelly, Michael Levine, Chris McCloskey, biorpg, vamX, Viktor Bowallius, Cory Kujawski.
Thank you to all my generous patrons and donaters!
<!-- footer end -->
# Original model card: Kaio Ken's SuperHOT 8K
### SuperHOT Prototype 2 w/ 8K Context
This is a second prototype of SuperHOT, a NSFW focused LoRA, this time 7B with 8K context and no RLHF, using the same technique described in [the github blog](https://kaiokendev.github.io/til#extending-context-to-8k).
#### Looking for Merged & Quantized Models?
Make some please :)
#### Using the monkey-patch?
You will **NEED** to **apply the monkeypatch** or, if you are already using the monkeypatch, **change the scaling factor to 0.25 and the maximum sequence length to 8192**
The monkeypatch is only necessary if you are using a front-end/back-end that does not already support scaling and said front-end/back-end is Python-based (i.e. Huggingface Transformers). To apply the patch, you will need to copy the `llama_rope_scaled_monkey_patch.py` into your working directory and call the exported function `replace_llama_rope_with_scaled_rope` at the very start of your Python program. It will modify the Transformers library's implementation of RoPE to properly apply the scaling factor.
#### Using Oobabooga with Exllama?
Switch your loader to `exllama` or `exllama_hf` Add the arguments `max_seq_len 8192` and `compress_pos_emb 4`. **While the model may work well with `compress_pos_emb 2`, it was trained on 4, so that is what I advocate for you to use**
Example in the command-line:
- `python server.py --max_seq_len 8192 --compress_pos_emb 4 --loader exllama_hf`
In the UI, you will see the loader option in the `Models` tab. Once you select either `exllama` or `exllama_hf`, the `max_seq_len` and `compress_pos_emb` settings will appear.
#### Training Details
I trained the LoRA with the following configuration:
- 1200 samples (~400 samples over 2048 sequence length)
- learning rate of 3e-4
- 3 epochs
- The exported modules are:
- q_proj
- k_proj
- v_proj
- o_proj
- no bias
- Rank = 4
- Alpha = 8
- no dropout
- weight decay of 0.1
- AdamW beta1 of 0.9 and beta2 0.99, epsilon of 1e-5
- Trained on 4-bit base model
- Cutoff length: 4096
# Original model card: Project Baize's Baize 7B v2
<p align="center">
<img width="500px" alt="Project Baize" src="https://user-images.githubusercontent.com/22514219/229195563-0cddfa74-e52f-4413-b4b4-e4ba489c4b3d.png">
</p>
<hr>
## ⚠️Warning
Using Baize checkpoints directly without the following format will not work.
```
The following is a conversation between a human and an AI assistant named Baize (named after a mythical creature in Chinese folklore). Baize is an open-source AI assistant developed by UCSD and Sun Yat-Sen University. The human and the AI assistant take turns chatting. Human statements start with [|Human|] and AI assistant statements start with [|AI|]. The AI assistant always provides responses in as much detail as possible, and in Markdown format. The AI assistant always declines to engage with topics, questions and instructions related to unethical, controversial, or sensitive issues. Complete the transcript in exactly that format.\n[|Human|]Hello!\n[|AI|]Hi!
```
`[|Human|]` and `[|AI|]` are required to mark the messages from the user and Baize. We recommend checking out our [GitHub](https://github.com/project-baize/baize) to find the best way to use Baize with our demo or Fastchat.
## Demo
https://huggingface.co/spaces/project-baize/chat-with-baize
## What's Baize?
Baize is an open-source chat model fine-tuned with [LoRA](https://github.com/microsoft/LoRA). This model is a **7B Baize-v2**, trained with supervised fine-tuning (SFT) and self-distillation with feedback (SDF). This checkpoint has been merged with LLaMA so it's ready for use.
## Why it's called Baize?
Baize (白泽) is a mythical creature in Chinese folklore, who speaks human languages and knows everything. This is exactly what we expect from a chat model.
## How to use it: local demo, API and SDK
More details can be found in the Baize [GitHub](https://github.com/project-baize/baize) and [Paper](https://arxiv.org/abs/2304.01196).
|
jpohhhh/MiniLM-L6-v2-optimum-embeddings
|
jpohhhh
| 2023-07-06T15:40:13Z | 8 | 0 |
generic
|
[
"generic",
"onnx",
"bert",
"sentence-embeddings",
"endpoints-template",
"optimum",
"license:mit",
"endpoints_compatible",
"region:us"
] | null | 2023-07-06T15:03:26Z |
---
license: mit
tags:
- sentence-embeddings
- endpoints-template
- optimum
library_name: generic
---
This repository is a fork of philschmid/all-MiniLM-L6-v2-optimum-embeddings.
My own ONNX conversion seems to be about 4x slower, no discernable reason why: the quantized models seem roughly the same.
The idea here is by forking we can ex. upgrade the Optimum lib used as well.
|
keonju/BERTopic
|
keonju
| 2023-07-06T15:36:42Z | 22 | 2 |
bertopic
|
[
"bertopic",
"text-classification",
"region:us"
] |
text-classification
| 2023-06-22T16:09:55Z |
---
tags:
- bertopic
library_name: bertopic
pipeline_tag: text-classification
---
# BERTopic
This is a [BERTopic](https://github.com/MaartenGr/BERTopic) model.
BERTopic is a flexible and modular topic modeling framework that allows for the generation of easily interpretable topics from large datasets.
## Usage
To use this model, please install BERTopic:
```
pip install -U bertopic
```
You can use the model as follows:
```python
from bertopic import BERTopic
topic_model = BERTopic.load("keonju/BERTopic")
topic_model.get_topic_info()
```
## Topic overview
* Number of topics: 158
* Number of training documents: 10158
<details>
<summary>Click here for an overview of all topics.</summary>
| Topic ID | Topic Keywords | Topic Frequency | Label |
|----------|----------------|-----------------|-------|
| -1 | and - the - of - in - to | 10 | -1_and_the_of_in |
| 0 | holocene - china - the - monsoon - bp | 3858 | 0_holocene_china_the_monsoon |
| 1 | energy - biofuels - production - biodiesel - bioenergy | 291 | 1_energy_biofuels_production_biodiesel |
| 2 | coal - coals - the - basin - seams | 248 | 2_coal_coals_the_basin |
| 3 | yr - holocene - the - bp - and | 205 | 3_yr_holocene_the_bp |
| 4 | hg - mercury - mehg - of hg - in | 202 | 4_hg_mercury_mehg_of hg |
| 5 | ch4 - methane - emissions - fluxes - flux | 159 | 5_ch4_methane_emissions_fluxes |
| 6 | data - forest - spectral - for - mapping | 118 | 6_data_forest_spectral_for |
| 7 | bp - the - holocene - pollen - lake | 116 | 7_bp_the_holocene_pollen |
| 8 | wetlands - wetland - and - are - of | 104 | 8_wetlands_wetland_and_are |
| 9 | co2 - ecosystem - nee - exchange - net | 103 | 9_co2_ecosystem_nee_exchange |
| 10 | species - of - fen - the - restoration | 100 | 10_species_of_fen_the |
| 11 | peat - tropical - peatlands - palm - peatland | 98 | 11_peat_tropical_peatlands_palm |
| 12 | pb - lead - atmospheric - metal - deposition | 96 | 12_pb_lead_atmospheric_metal |
| 13 | the - lake - of the - of - poland | 93 | 13_the_lake_of the_of |
| 14 | pm2 - haze - burning - air - aerosol | 90 | 14_pm2_haze_burning_air |
| 15 | doc - catchments - carbon - organic carbon - export | 88 | 15_doc_catchments_carbon_organic carbon |
| 16 | the - carbon - of - co2 - of the | 73 | 16_the_carbon_of_co2 |
| 17 | wetland - wetlands - classification - mapping - and | 69 | 17_wetland_wetlands_classification_mapping |
| 18 | uv - ozone - o3 - isoprene - elevated | 67 | 18_uv_ozone_o3_isoprene |
| 19 | mediterranean - the - glacial - iberian - during | 66 | 19_mediterranean_the_glacial_iberian |
| 20 | media - compost - growing media - growing - biochar | 63 | 20_media_compost_growing media_growing |
| 21 | 137cs - of 137cs - sup - ce sup - radiocaesium | 63 | 21_137cs_of 137cs_sup_ce sup |
| 22 | testate - amoebae - testate amoebae - of testate - amoeba | 62 | 22_testate_amoebae_testate amoebae_of testate |
| 23 | peat - pyrolysis - lignin - gc - of | 62 | 23_peat_pyrolysis_lignin_gc |
| 24 | cu - zn - metals - peat - elements | 62 | 24_cu_zn_metals_peat |
| 25 | alkanes - alkane - chain - values - plants | 61 | 25_alkanes_alkane_chain_values |
| 26 | permafrost - active layer - thermal - ground - layer | 60 | 26_permafrost_active layer_thermal_ground |
| 27 | streams - diatom - species - macroinvertebrate - stream | 60 | 27_streams_diatom_species_macroinvertebrate |
| 28 | records - the - of - record - ireland | 60 | 28_records_the_of_record |
| 29 | water - flow - groundwater - recharge - runoff | 59 | 29_water_flow_groundwater_recharge |
| 30 | habitat - species - breeding - bird - nest | 57 | 30_habitat_species_breeding_bird |
| 31 | brgdgts - gdgts - glycerol - brgdgt - branched | 56 | 31_brgdgts_gdgts_glycerol_brgdgt |
| 32 | deposition - nitrogen - nitrogen deposition - sphagnum - of | 55 | 32_deposition_nitrogen_nitrogen deposition_sphagnum |
| 33 | oil sands - sands - fen - oil - reclamation | 54 | 33_oil sands_sands_fen_oil |
| 34 | fire - burned - severity - burning - post fire | 54 | 34_fire_burned_severity_burning |
| 35 | acidification - deposition - acid - ph - catchment | 54 | 35_acidification_deposition_acid_ph |
| 36 | farm - land - agricultural - farmers - policy | 53 | 36_farm_land_agricultural_farmers |
| 37 | cdom - doc - dom - dissolved organic - dissolved | 53 | 37_cdom_doc_dom_dissolved organic |
| 38 | redd - indonesia - deforestation - in indonesia - forest | 50 | 38_redd_indonesia_deforestation_in indonesia |
| 39 | ash - wood ash - wood - growth - of wood | 49 | 39_ash_wood ash_wood_growth |
| 40 | fungal - fungi - mycorrhizal - species - root | 49 | 40_fungal_fungi_mycorrhizal_species |
| 41 | stand - growth - models - tree - stands | 49 | 41_stand_growth_models_tree |
| 42 | smouldering - smoldering - spread - peat - combustion | 49 | 42_smouldering_smoldering_spread_peat |
| 43 | pollen - of pollen - vegetation - of - from | 49 | 43_pollen_of pollen_vegetation_of |
| 44 | arsenic - as - of as - fe - of arsenic | 49 | 44_arsenic_as_of as_fe |
| 45 | ch4 - methane - production - peat - methanogenesis | 47 | 45_ch4_methane_production_peat |
| 46 | africa - the - bp - south - late | 46 | 46_africa_the_bp_south |
| 47 | soc - carbon - soil - stocks - land | 45 | 47_soc_carbon_soil_stocks |
| 48 | soil - organic - carbon - soil organic - soils | 45 | 48_soil_organic_carbon_soil organic |
| 49 | wetlands - constructed - wetland - treatment - phosphorus | 43 | 49_wetlands_constructed_wetland_treatment |
| 50 | microbial - rare - soil - bacterial - diversity | 43 | 50_microbial_rare_soil_bacterial |
| 51 | litter - decomposition - mass loss - litter decomposition - mass | 39 | 51_litter_decomposition_mass loss_litter decomposition |
| 52 | co2 - pco2 - emissions - carbon - ch4 | 39 | 52_co2_pco2_emissions_carbon |
| 53 | soc - carbon - wetland - wetlands - soil | 39 | 53_soc_carbon_wetland_wetlands |
| 54 | countries - emissions - emission - to - climate | 38 | 54_countries_emissions_emission_to |
| 55 | services - ecosystem - ecosystem services - es - pes | 37 | 55_services_ecosystem_ecosystem services_es |
| 56 | catalyst - peat - pyrolysis - char - catalysts | 37 | 56_catalyst_peat_pyrolysis_char |
| 57 | clearfelling - water - phosphorus - buffer - nutrient | 35 | 57_clearfelling_water_phosphorus_buffer |
| 58 | forest - forests - trees - tree - stands | 35 | 58_forest_forests_trees_tree |
| 59 | carbon - climate - atmosphere - earth - carbon cycle | 34 | 59_carbon_climate_atmosphere_earth |
| 60 | tephra - volcanic - cryptotephra - eruptions - tephras | 34 | 60_tephra_volcanic_cryptotephra_eruptions |
| 61 | testate - arcellinida - coi - species - amoebae | 34 | 61_testate_arcellinida_coi_species |
| 62 | methane - methanogenic - community - methanogen - methanogens | 34 | 62_methane_methanogenic_community_methanogen |
| 63 | consolidation - soil - embankment - road - the | 33 | 63_consolidation_soil_embankment_road |
| 64 | species - spider - bogs - spiders - habitat | 33 | 64_species_spider_bogs_spiders |
| 65 | evaporation - energy - model - was - the | 33 | 65_evaporation_energy_model_was |
| 66 | phosphorus - catchment - in - tp - concentrations | 33 | 66_phosphorus_catchment_in_tp |
| 67 | co2 - ch4 - marsh - wetland - emissions | 33 | 67_co2_ch4_marsh_wetland |
| 68 | runoff - peat - channels - flow - catchment | 33 | 68_runoff_peat_channels_flow |
| 69 | nutrient - nitrogen - fertilizer - litter - of | 32 | 69_nutrient_nitrogen_fertilizer_litter |
| 70 | brazil - bp - the - of - in the | 31 | 70_brazil_bp_the_of |
| 71 | tsunami - holocene - the - volcanic - deposits | 30 | 71_tsunami_holocene_the_volcanic |
| 72 | climate change - change - climate - biodiversity - ecosystem | 30 | 72_climate change_change_climate_biodiversity |
| 73 | gpr - resistivity - radar - penetrating - penetrating radar | 29 | 73_gpr_resistivity_radar_penetrating |
| 74 | holocene - the - andes - and - bp | 29 | 74_holocene_the_andes_and |
| 75 | permafrost - soc - soil - soils - arctic | 28 | 75_permafrost_soc_soil_soils |
| 76 | policy - forest - owners - arguments - forest owners | 28 | 76_policy_forest_owners_arguments |
| 77 | bog - poland - peatland - europe - ca | 28 | 77_bog_poland_peatland_europe |
| 78 | ch4 - oxidation - methane - paddy - aom | 28 | 78_ch4_oxidation_methane_paddy |
| 79 | enzyme - enzymes - eea - soil - activities | 28 | 79_enzyme_enzymes_eea_soil |
| 80 | channel - catchment - flow - bends - model | 28 | 80_channel_catchment_flow_bends |
| 81 | soil - soil science - science - of soil - eu | 27 | 81_soil_soil science_science_of soil |
| 82 | pahs - pah - polycyclic aromatic - polycyclic - aromatic | 27 | 82_pahs_pah_polycyclic aromatic_polycyclic |
| 83 | n2o - n2o emissions - emissions - emission - nitrous | 26 | 83_n2o_n2o emissions_emissions_emission |
| 84 | peat water - adsorption - electrocoagulation - brackish peat - brackish peat water | 26 | 84_peat water_adsorption_electrocoagulation_brackish peat |
| 85 | mangrove - mangroves - carbon - coastal - b2 | 26 | 85_mangrove_mangroves_carbon_coastal |
| 86 | species - retention - alien - richness - forests | 25 | 86_species_retention_alien_richness |
| 87 | colloidal - river - elements - fe - colloids | 25 | 87_colloidal_river_elements_fe |
| 88 | sulfate - sulfur - 34s - peat - sulphur | 24 | 88_sulfate_sulfur_34s_peat |
| 89 | caribou - habitat - woodland caribou - populations - wolf | 24 | 89_caribou_habitat_woodland caribou_populations |
| 90 | food - agriculture - food system - change - covid 19 | 24 | 90_food_agriculture_food system_change |
| 91 | microbial - community - microbial community - communities - bacterial | 23 | 91_microbial_community_microbial community_communities |
| 92 | sorption - cu - ions - ii - cu ii | 22 | 92_sorption_cu_ions_ii |
| 93 | fire - fires - algorithm - frp - hotspot | 22 | 93_fire_fires_algorithm_frp |
| 94 | choice - wtp - preferences - valuation - choice experiment | 22 | 94_choice_wtp_preferences_valuation |
| 95 | nematodes - earthworm - soil - food - nematode | 22 | 95_nematodes_earthworm_soil_food |
| 96 | conservation - orangutan - habitat - forest - species | 21 | 96_conservation_orangutan_habitat_forest |
| 97 | cushion - accumulation - peat - amazonian - vegetation | 21 | 97_cushion_accumulation_peat_amazonian |
| 98 | ch4 - oxidation - ch4 oxidation - uptake - ch4 uptake | 20 | 98_ch4_oxidation_ch4 oxidation_uptake |
| 99 | tidal - sediment - coastal - delta - the | 20 | 99_tidal_sediment_coastal_delta |
| 100 | emissions - co2 - ghg - n2o - table | 20 | 100_emissions_co2_ghg_n2o |
| 101 | methane - ph - cytochrome - methanotrophs - acetic acid | 20 | 101_methane_ph_cytochrome_methanotrophs |
| 102 | patterns - model - self organization - evolutionary - self | 20 | 102_patterns_model_self organization_evolutionary |
| 103 | nitrogen - denitrification - n2o - soil - n2 | 20 | 103_nitrogen_denitrification_n2o_soil |
| 104 | birch - rotation - biomass - buds - biomass production | 19 | 104_birch_rotation_biomass_buds |
| 105 | fire - wildfire - fires - wildfires - health | 19 | 105_fire_wildfire_fires_wildfires |
| 106 | grazing - heathland - heather - moorland - england | 19 | 106_grazing_heathland_heather_moorland |
| 107 | emissions - fire - burning - fire emissions - biomass burning | 19 | 107_emissions_fire_burning_fire emissions |
| 108 | peat - landslides - failure - of peat - peat compaction | 18 | 108_peat_landslides_failure_of peat |
| 109 | biochar - straw - soil - fe - bc | 18 | 109_biochar_straw_soil_fe |
| 110 | ecosystem - respiration - carbon - ecosystem respiration - meadow | 17 | 110_ecosystem_respiration_carbon_ecosystem respiration |
| 111 | wetland - wetlands - risk - of wetland - the wetland | 17 | 111_wetland_wetlands_risk_of wetland |
| 112 | dom - thm - groundwater - molecular - organic | 17 | 112_dom_thm_groundwater_molecular |
| 113 | geochemistry - landscape geochemistry - rocks - peat - mafic | 17 | 113_geochemistry_landscape geochemistry_rocks_peat |
| 114 | tundra - ch4 - n2o - fluxes - antarctic | 16 | 114_tundra_ch4_n2o_fluxes |
| 115 | cellulose - sphagnum - isotopic - isotope - δ18ocel | 16 | 115_cellulose_sphagnum_isotopic_isotope |
| 116 | solute - transport - chloride - peat - pore | 16 | 116_solute_transport_chloride_peat |
| 117 | charcoal - fire - fires - holocene - fire history | 15 | 117_charcoal_fire_fires_holocene |
| 118 | ghg - agricultural - dairy - abatement - emissions | 15 | 118_ghg_agricultural_dairy_abatement |
| 119 | palm - oil - palm oil - sustainability - industry | 15 | 119_palm_oil_palm oil_sustainability |
| 120 | humic - humic substances - substances - acids - fluorescence | 15 | 120_humic_humic substances_substances_acids |
| 121 | canopy - ndvi - pri - lue - phenological | 15 | 121_canopy_ndvi_pri_lue |
| 122 | pollen - bog - peat - the - human impact | 15 | 122_pollen_bog_peat_the |
| 123 | marshes - tidal - marshes are - salt - or | 15 | 123_marshes_tidal_marshes are_salt |
| 124 | soil - prediction - mapping - covariates - dsm | 15 | 124_soil_prediction_mapping_covariates |
| 125 | si - of si - silicon - biogenic - protozoic | 14 | 125_si_of si_silicon_biogenic |
| 126 | et - evapotranspiration - le - wetland - rice | 14 | 126_et_evapotranspiration_le_wetland |
| 127 | forest - finland - forests - stock - management | 14 | 127_forest_finland_forests_stock |
| 128 | iodine - 129i - sorption - iodide - the sorption | 14 | 128_iodine_129i_sorption_iodide |
| 129 | palm - oil - palm oil - smallholders - certification | 14 | 129_palm_oil_palm oil_smallholders |
| 130 | dndc - model - models - soil - carbon | 14 | 130_dndc_model_models_soil |
| 131 | snow - thaw - cover - sca - data | 14 | 131_snow_thaw_cover_sca |
| 132 | stx2 - microbiota - gut - gut microbiota - microbial | 13 | 132_stx2_microbiota_gut_gut microbiota |
| 133 | dom - doc - organic - dissolved organic - of dom | 13 | 133_dom_doc_organic_dissolved organic |
| 134 | forest - cbm - ontario - cfs3 - cbm cfs3 | 13 | 134_forest_cbm_ontario_cfs3 |
| 135 | wind - wind farms - farms - onshore - onshore wind | 13 | 135_wind_wind farms_farms_onshore |
| 136 | uranium - of uranium - 232th - th - ar | 13 | 136_uranium_of uranium_232th_th |
| 137 | groundwater - springs - spring - gdes - discharge | 13 | 137_groundwater_springs_spring_gdes |
| 138 | fire - forest - boreal - burned - fires | 13 | 138_fire_forest_boreal_burned |
| 139 | metal - metals - cd - sediments - zn | 13 | 139_metal_metals_cd_sediments |
| 140 | slr - sea level - coastal - sea - sea level rise | 13 | 140_slr_sea level_coastal_sea |
| 141 | damo - methane - anaerobic - oxidation - aom | 12 | 141_damo_methane_anaerobic_oxidation |
| 142 | temperature - microbial - soil - co2 - pd | 12 | 142_temperature_microbial_soil_co2 |
| 143 | soil - respiration - root - soil respiration - enchytraeid | 12 | 143_soil_respiration_root_soil respiration |
| 144 | kerp - fusiformisporites - permian - genus - flora | 11 | 144_kerp_fusiformisporites_permian_genus |
| 145 | dust - dust deposition - dust sources - deposition - atmospheric dust | 11 | 145_dust_dust deposition_dust sources_deposition |
| 146 | methane - sources - ch4 - les - de | 11 | 146_methane_sources_ch4_les |
| 147 | n2o - n2o emissions - emissions - permafrost - n2o fluxes | 11 | 147_n2o_n2o emissions_emissions_permafrost |
| 148 | australia - mis - record - ka - crater | 11 | 148_australia_mis_record_ka |
| 149 | oc - fjords - fjord - lakes - of oc | 10 | 149_oc_fjords_fjord_lakes |
| 150 | fe - reduction - fe iii - sr10 - iron | 10 | 150_fe_reduction_fe iii_sr10 |
| 151 | loading - eutrophication - nitrogen - coastal - phytoplankton | 10 | 151_loading_eutrophication_nitrogen_coastal |
| 152 | model - wetlands - groundwater - water - the wetlands | 10 | 152_model_wetlands_groundwater_water |
| 153 | co2 - soil - co2 efflux - soil co2 efflux - soil co2 | 10 | 153_co2_soil_co2 efflux_soil co2 efflux |
| 154 | transfer - transfer functions - transfer function - testate - functions | 10 | 154_transfer_transfer functions_transfer function_testate |
| 155 | peat - spain - bog - matter - autofluorescent | 10 | 155_peat_spain_bog_matter |
| 156 | isbas - insar - subsidence - motion - deformation | 10 | 156_isbas_insar_subsidence_motion |
</details>
## Training hyperparameters
* calculate_probabilities: False
* language: None
* low_memory: False
* min_topic_size: 10
* n_gram_range: (1, 3)
* nr_topics: None
* seed_topic_list: None
* top_n_words: 30
* verbose: False
## Framework versions
* Numpy: 1.22.4
* HDBSCAN: 0.8.29
* UMAP: 0.5.3
* Pandas: 1.5.3
* Scikit-Learn: 1.2.2
* Sentence-transformers: 2.2.2
* Transformers: 4.30.2
* Numba: 0.56.4
* Plotly: 5.13.1
* Python: 3.10.12
|
josero23/rutt-3
|
josero23
| 2023-07-06T15:26:38Z | 1 | 0 |
diffusers
|
[
"diffusers",
"autotrain_compatible",
"endpoints_compatible",
"diffusers:StableDiffusionPipeline",
"region:us"
] |
text-to-image
| 2023-07-06T15:19:29Z |
---
license: creativeml-openrail-m
tags:
- text-to-image
- stable-diffusion
---
### rutt_3 Dreambooth model trained by josero23 with TheLastBen's fast-DreamBooth notebook
|
hopkins/eng-guj-longest
|
hopkins
| 2023-07-06T15:16:01Z | 103 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"mbart",
"text2text-generation",
"translation",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
translation
| 2023-07-06T14:54:00Z |
---
tags:
- translation
- generated_from_trainer
metrics:
- bleu
model-index:
- name: eng-guj-longest
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# eng-guj-longest
This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 3.1767
- Bleu: 3.1810
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3
|
Huggingfly/Reinforce-Pixelcopter-PLE-v0
|
Huggingfly
| 2023-07-06T15:15:09Z | 0 | 0 | null |
[
"Pixelcopter-PLE-v0",
"reinforce",
"reinforcement-learning",
"custom-implementation",
"deep-rl-class",
"model-index",
"region:us"
] |
reinforcement-learning
| 2023-07-06T15:15:03Z |
---
tags:
- Pixelcopter-PLE-v0
- reinforce
- reinforcement-learning
- custom-implementation
- deep-rl-class
model-index:
- name: Reinforce-Pixelcopter-PLE-v0
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: Pixelcopter-PLE-v0
type: Pixelcopter-PLE-v0
metrics:
- type: mean_reward
value: 51.80 +/- 39.15
name: mean_reward
verified: false
---
# **Reinforce** Agent playing **Pixelcopter-PLE-v0**
This is a trained model of a **Reinforce** agent playing **Pixelcopter-PLE-v0** .
To learn to use this model and train yours check Unit 4 of the Deep Reinforcement Learning Course: https://huggingface.co/deep-rl-course/unit4/introduction
|
hopkins/eng-ind-longest
|
hopkins
| 2023-07-06T15:12:40Z | 103 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"mbart",
"text2text-generation",
"translation",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
translation
| 2023-07-06T14:54:34Z |
---
tags:
- translation
- generated_from_trainer
metrics:
- bleu
model-index:
- name: eng-ind-longest
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# eng-ind-longest
This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 1.7474
- Bleu: 21.9863
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3
|
hopkins/eng-ind-random
|
hopkins
| 2023-07-06T15:09:21Z | 103 | 0 |
transformers
|
[
"transformers",
"pytorch",
"tensorboard",
"mbart",
"text2text-generation",
"translation",
"generated_from_trainer",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
translation
| 2023-07-06T14:51:41Z |
---
tags:
- translation
- generated_from_trainer
metrics:
- bleu
model-index:
- name: eng-ind-random
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# eng-ind-random
This model is a fine-tuned version of [facebook/mbart-large-50-many-to-many-mmt](https://huggingface.co/facebook/mbart-large-50-many-to-many-mmt) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 1.7847
- Bleu: 21.5508
## Model description
More information needed
## Intended uses & limitations
More information needed
## Training and evaluation data
More information needed
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3
- mixed_precision_training: Native AMP
### Training results
### Framework versions
- Transformers 4.26.1
- Pytorch 2.0.1+cu117
- Datasets 2.12.0
- Tokenizers 0.13.3
|
makaveli10/Reinforce-CartPole-v0
|
makaveli10
| 2023-07-06T15:09:08Z | 0 | 0 | null |
[
"CartPole-v1",
"reinforce",
"reinforcement-learning",
"custom-implementation",
"deep-rl-class",
"model-index",
"region:us"
] |
reinforcement-learning
| 2023-07-06T15:08:58Z |
---
tags:
- CartPole-v1
- reinforce
- reinforcement-learning
- custom-implementation
- deep-rl-class
model-index:
- name: Reinforce-CartPole-v0
results:
- task:
type: reinforcement-learning
name: reinforcement-learning
dataset:
name: CartPole-v1
type: CartPole-v1
metrics:
- type: mean_reward
value: 500.00 +/- 0.00
name: mean_reward
verified: false
---
# **Reinforce** Agent playing **CartPole-v1**
This is a trained model of a **Reinforce** agent playing **CartPole-v1** .
To learn to use this model and train yours check Unit 4 of the Deep Reinforcement Learning Course: https://huggingface.co/deep-rl-course/unit4/introduction
|
ltmai/Bio_ClinicalBERT_DDI_finetuned
|
ltmai
| 2023-07-06T15:04:37Z | 114 | 1 |
transformers
|
[
"transformers",
"pytorch",
"bert",
"generated_from_trainer",
"chemistry",
"medical",
"drug_drug_interaction",
"text-classification",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2023-07-05T02:58:42Z |
---
tags:
- generated_from_trainer
- chemistry
- medical
- drug_drug_interaction
metrics:
- f2-score
- recall
- precision
- mcc
model-index:
- name: Bio_ClinicalBERT_DDI_finetuned
results:
- task:
name: Drug - Drug Interaction Classification
type: text-classification
dataset:
name: DrugBank
type: REST API
metrics:
- name: Recall
type: recall
value: 0.7849
widget:
- text: "[Ca++].[O-]C([O-])=O [SEP] OC[C@H](O)[C@@H](O)[C@H](O)[C@H](O)CO"
example_title: "Drug1 [SEP] Drug2"
pipeline_tag: text-classification
---
# Bio_ClinicalBERT_DDI_finetuned
This model was initialized from [Bio_ClinicalBERT](https://huggingface.co/emilyalsentzer/Bio_ClinicalBERT) by adding three hidden layers after the BERT pooler layer. The model was trained on the Drug-Drug Interaction dataset extracted from [DrugBank](https://go.drugbank.com/releases/latest) database and [National Library of Medicine](https://lhncbc.nlm.nih.gov/RxNav/APIs/InteractionAPIs.html) API.
It achieves the following results on the Test dataset:
- F2: 0.7872
- AUPRC: 0.869
- Recall: 0.7849
- Precision: 0.7967
- MCC: 0.3779
## Model description
Predict Drug Drug Interaction (DDI) from Chemical Structure of two drugs. The Model returns the probability of the two drugs having interaction with each other.
## Intended uses & limitations
To construct the input, using "[SEP]" token to seperate between the two drugs, example of a properly constructed input is as following
```python
drug1 = "[Ca++].[O-]C([O-])=O" #Calcium Carbonate
drug2 = "OC[C@H](O)[C@@H](O)[C@H](O)[C@H](O)CO" #Sorbitol
correct_input = "[Ca++].[O-]C([O-])=O [SEP] OC[C@H](O)[C@@H](O)[C@H](O)[C@H](O)CO"
```
## Training and evaluation data
To avoid data leakage and able to predict DDI for new drugs, the drug1 or drug2 in the validation and the test set were not included in the training set. Their SMILES chemical structures were never exposed to the training process.
## Training procedure
Using AWS EC2 g5.4xlarge instance. 24GB GPU.
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.01
- train_batch_size: 32
- eval_batch_size: 32
- seed: 7
- optimizer: Adadelta with weight_decay=1e-04
- lr_scheduler_type: CosineAnnealingLR
- num_epochs: 4
### Training results
| Training Loss | Epoch | Validation Loss | F2 | Recall | Precision | Mcc |
|:-------------:|:-----:|:---------------:|:------:|:------:|:---------:|:------:|
| 0.6068 | 1.0 | 0.7061 | 0.6508 | 0.6444 | 0.6778 | 0.2514 |
| 0.4529 | 2.0 | 0.8334 | 0.7555 | 0.7727 | 0.6939 | 0.3451 |
| 0.3375 | 3.0 | 0.9582 | 0.7636 | 0.7840 | 0.6915 | 0.3474 |
| 0.2624 | 4.0 | 1.2588 | 0.7770 | 0.8004 | 0.6954 | 0.3654 |
### Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1
- Datasets 2.13.1
- Tokenizers 0.13.3
|
Subsets and Splits
Filtered Qwen2.5 Distill Models
Identifies specific configurations of models by filtering cards that contain 'distill', 'qwen2.5', '7b' while excluding certain base models and incorrect model ID patterns, uncovering unique model variants.
Filtered Model Cards Count
Finds the count of entries with specific card details that include 'distill', 'qwen2.5', '7b' but exclude certain base models, revealing valuable insights about the dataset's content distribution.
Filtered Distill Qwen 7B Models
Filters for specific card entries containing 'distill', 'qwen', and '7b', excluding certain strings and patterns, to identify relevant model configurations.
Filtered Qwen-7b Model Cards
The query performs a detailed filtering based on specific keywords and excludes certain entries, which could be useful for identifying a specific subset of cards but does not provide deeper insights or trends.
Filtered Qwen 7B Model Cards
The query filters for specific terms related to "distilled" or "distill", "qwen", and "7b" in the 'card' column but excludes certain base models, providing a limited set of entries for further inspection.
Qwen 7B Distilled Models
The query provides a basic filtering of records to find specific card names that include keywords related to distilled Qwen 7b models, excluding a particular base model, which gives limited insight but helps in focusing on relevant entries.
Qwen 7B Distilled Model Cards
The query filters data based on specific keywords in the modelId and card fields, providing limited insight primarily useful for locating specific entries rather than revealing broad patterns or trends.
Qwen 7B Distilled Models
Finds all entries containing the terms 'distilled', 'qwen', and '7b' in a case-insensitive manner, providing a filtered set of records but without deeper analysis.
Distilled Qwen 7B Models
The query filters for specific model IDs containing 'distilled', 'qwen', and '7b', providing a basic retrieval of relevant entries but without deeper analysis or insight.
Filtered Model Cards with Distill Qwen2.
Filters and retrieves records containing specific keywords in the card description while excluding certain phrases, providing a basic count of relevant entries.
Filtered Model Cards with Distill Qwen 7
The query filters specific variations of card descriptions containing 'distill', 'qwen', and '7b' while excluding a particular base model, providing limited but specific data retrieval.
Distill Qwen 7B Model Cards
The query filters and retrieves rows where the 'card' column contains specific keywords ('distill', 'qwen', and '7b'), providing a basic filter result that can help in identifying specific entries.