modelId
stringlengths 5
139
| author
stringlengths 2
42
| last_modified
timestamp[us, tz=UTC]date 2020-02-15 11:33:14
2025-08-30 06:27:36
| downloads
int64 0
223M
| likes
int64 0
11.7k
| library_name
stringclasses 527
values | tags
listlengths 1
4.05k
| pipeline_tag
stringclasses 55
values | createdAt
timestamp[us, tz=UTC]date 2022-03-02 23:29:04
2025-08-30 06:27:12
| card
stringlengths 11
1.01M
|
|---|---|---|---|---|---|---|---|---|---|
BootesVoid/cmexnbmuh05mxsr53s1r1frsx_cmexp1qej05o0sr53vv5b1zms
|
BootesVoid
| 2025-08-30T04:09:17Z | 0 | 0 |
diffusers
|
[
"diffusers",
"flux",
"lora",
"replicate",
"text-to-image",
"en",
"base_model:black-forest-labs/FLUX.1-dev",
"base_model:adapter:black-forest-labs/FLUX.1-dev",
"license:other",
"region:us"
] |
text-to-image
| 2025-08-30T04:09:16Z |
---
license: other
license_name: flux-1-dev-non-commercial-license
license_link: https://huggingface.co/black-forest-labs/FLUX.1-dev/blob/main/LICENSE.md
language:
- en
tags:
- flux
- diffusers
- lora
- replicate
base_model: "black-forest-labs/FLUX.1-dev"
pipeline_tag: text-to-image
# widget:
# - text: >-
# prompt
# output:
# url: https://...
instance_prompt: CALIE22
---
# Cmexnbmuh05Mxsr53S1R1Frsx_Cmexp1Qej05O0Sr53Vv5B1Zms
<Gallery />
## About this LoRA
This is a [LoRA](https://replicate.com/docs/guides/working-with-loras) for the FLUX.1-dev text-to-image model. It can be used with diffusers or ComfyUI.
It was trained on [Replicate](https://replicate.com/) using AI toolkit: https://replicate.com/ostris/flux-dev-lora-trainer/train
## Trigger words
You should use `CALIE22` to trigger the image generation.
## Run this LoRA with an API using Replicate
```py
import replicate
input = {
"prompt": "CALIE22",
"lora_weights": "https://huggingface.co/BootesVoid/cmexnbmuh05mxsr53s1r1frsx_cmexp1qej05o0sr53vv5b1zms/resolve/main/lora.safetensors"
}
output = replicate.run(
"black-forest-labs/flux-dev-lora",
input=input
)
for index, item in enumerate(output):
with open(f"output_{index}.webp", "wb") as file:
file.write(item.read())
```
## Use it with the [🧨 diffusers library](https://github.com/huggingface/diffusers)
```py
from diffusers import AutoPipelineForText2Image
import torch
pipeline = AutoPipelineForText2Image.from_pretrained('black-forest-labs/FLUX.1-dev', torch_dtype=torch.float16).to('cuda')
pipeline.load_lora_weights('BootesVoid/cmexnbmuh05mxsr53s1r1frsx_cmexp1qej05o0sr53vv5b1zms', weight_name='lora.safetensors')
image = pipeline('CALIE22').images[0]
```
For more details, including weighting, merging and fusing LoRAs, check the [documentation on loading LoRAs in diffusers](https://huggingface.co/docs/diffusers/main/en/using-diffusers/loading_adapters)
## Training details
- Steps: 2500
- Learning rate: 9e-05
- LoRA rank: 16
## Contribute your own examples
You can use the [community tab](https://huggingface.co/BootesVoid/cmexnbmuh05mxsr53s1r1frsx_cmexp1qej05o0sr53vv5b1zms/discussions) to add images that show off what you’ve made with this LoRA.
|
mradermacher/Luna-i1-GGUF
|
mradermacher
| 2025-08-30T03:52:28Z | 0 | 0 |
transformers
|
[
"transformers",
"gguf",
"roleplay",
"chat",
"rp",
"character",
"waifu",
"en",
"zh",
"vi",
"base_model:beyoru/Luna",
"base_model:quantized:beyoru/Luna",
"license:mit",
"endpoints_compatible",
"region:us",
"imatrix",
"conversational"
] | null | 2025-08-30T00:20:46Z |
---
base_model: beyoru/Luna
language:
- en
- zh
- vi
library_name: transformers
license: mit
mradermacher:
readme_rev: 1
quantized_by: mradermacher
tags:
- roleplay
- chat
- rp
- character
- waifu
- character
---
## About
<!-- ### quantize_version: 2 -->
<!-- ### output_tensor_quantised: 1 -->
<!-- ### convert_type: hf -->
<!-- ### vocab_type: -->
<!-- ### tags: nicoboss -->
<!-- ### quants: Q2_K IQ3_M Q4_K_S IQ3_XXS Q3_K_M small-IQ4_NL Q4_K_M IQ2_M Q6_K IQ4_XS Q2_K_S IQ1_M Q3_K_S IQ2_XXS Q3_K_L IQ2_XS Q5_K_S IQ2_S IQ1_S Q5_K_M Q4_0 IQ3_XS Q4_1 IQ3_S -->
<!-- ### quants_skip: -->
<!-- ### skip_mmproj: -->
weighted/imatrix quants of https://huggingface.co/beyoru/Luna
<!-- provided-files -->
***For a convenient overview and download list, visit our [model page for this model](https://hf.tst.eu/model#Luna-i1-GGUF).***
static quants are available at https://huggingface.co/mradermacher/Luna-GGUF
## Usage
If you are unsure how to use GGUF files, refer to one of [TheBloke's
READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for
more details, including on how to concatenate multi-part files.
## Provided Quants
(sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
| Link | Type | Size/GB | Notes |
|:-----|:-----|--------:|:------|
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.imatrix.gguf) | imatrix | 0.1 | imatrix file (for creating your own qwuants) |
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.i1-IQ1_S.gguf) | i1-IQ1_S | 1.2 | for the desperate |
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.i1-IQ1_M.gguf) | i1-IQ1_M | 1.2 | mostly desperate |
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.i1-IQ2_XXS.gguf) | i1-IQ2_XXS | 1.3 | |
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.i1-IQ2_XS.gguf) | i1-IQ2_XS | 1.5 | |
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.i1-IQ2_S.gguf) | i1-IQ2_S | 1.5 | |
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.i1-IQ2_M.gguf) | i1-IQ2_M | 1.6 | |
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.i1-Q2_K_S.gguf) | i1-Q2_K_S | 1.7 | very low quality |
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.i1-Q2_K.gguf) | i1-Q2_K | 1.8 | IQ3_XXS probably better |
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.i1-IQ3_XXS.gguf) | i1-IQ3_XXS | 1.8 | lower quality |
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.i1-IQ3_XS.gguf) | i1-IQ3_XS | 1.9 | |
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.i1-Q3_K_S.gguf) | i1-Q3_K_S | 2.0 | IQ3_XS probably better |
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.i1-IQ3_S.gguf) | i1-IQ3_S | 2.0 | beats Q3_K* |
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.i1-IQ3_M.gguf) | i1-IQ3_M | 2.1 | |
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.i1-Q3_K_M.gguf) | i1-Q3_K_M | 2.2 | IQ3_S probably better |
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.i1-Q3_K_L.gguf) | i1-Q3_K_L | 2.3 | IQ3_M probably better |
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.i1-IQ4_XS.gguf) | i1-IQ4_XS | 2.4 | |
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.i1-Q4_0.gguf) | i1-Q4_0 | 2.5 | fast, low quality |
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.i1-IQ4_NL.gguf) | i1-IQ4_NL | 2.5 | prefer IQ4_XS |
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.i1-Q4_K_S.gguf) | i1-Q4_K_S | 2.5 | optimal size/speed/quality |
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.i1-Q4_K_M.gguf) | i1-Q4_K_M | 2.6 | fast, recommended |
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.i1-Q4_1.gguf) | i1-Q4_1 | 2.7 | |
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.i1-Q5_K_S.gguf) | i1-Q5_K_S | 2.9 | |
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.i1-Q5_K_M.gguf) | i1-Q5_K_M | 3.0 | |
| [GGUF](https://huggingface.co/mradermacher/Luna-i1-GGUF/resolve/main/Luna.i1-Q6_K.gguf) | i1-Q6_K | 3.4 | practically like static Q6_K |
Here is a handy graph by ikawrakow comparing some lower-quality quant
types (lower is better):
And here are Artefact2's thoughts on the matter:
https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9
## FAQ / Model Request
See https://huggingface.co/mradermacher/model_requests for some answers to
questions you might have and/or if you want some other model quantized.
## Thanks
I thank my company, [nethype GmbH](https://www.nethype.de/), for letting
me use its servers and providing upgrades to my workstation to enable
this work in my free time. Additional thanks to [@nicoboss](https://huggingface.co/nicoboss) for giving me access to his private supercomputer, enabling me to provide many more imatrix quants, at much higher quality, than I would otherwise be able to.
<!-- end -->
|
allstax/editorial-qwen-v1-full
|
allstax
| 2025-08-30T03:12:41Z | 0 | 0 |
transformers
|
[
"transformers",
"safetensors",
"qwen3",
"text-generation",
"text-generation-inference",
"unsloth",
"conversational",
"en",
"license:apache-2.0",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-generation
| 2025-08-30T03:06:28Z |
---
base_model: unsloth/qwen3-14b-unsloth-bnb-4bit
tags:
- text-generation-inference
- transformers
- unsloth
- qwen3
license: apache-2.0
language:
- en
---
# Uploaded finetuned model
- **Developed by:** allstax
- **License:** apache-2.0
- **Finetuned from model :** unsloth/qwen3-14b-unsloth-bnb-4bit
This qwen3 model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library.
[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth)
|
pempekmangedd/blockassist-bc-patterned_sturdy_dolphin_1756520250
|
pempekmangedd
| 2025-08-30T02:41:20Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"patterned sturdy dolphin",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-30T02:41:17Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- patterned sturdy dolphin
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
QinShiHuangisavailable/output019
|
QinShiHuangisavailable
| 2025-08-30T02:23:40Z | 0 | 0 |
transformers
|
[
"transformers",
"tensorboard",
"safetensors",
"generated_from_trainer",
"trl",
"grpo",
"arxiv:2402.03300",
"base_model:Qwen/Qwen3-4B-Thinking-2507",
"base_model:finetune:Qwen/Qwen3-4B-Thinking-2507",
"endpoints_compatible",
"region:us"
] | null | 2025-08-25T17:44:15Z |
---
base_model: Qwen/Qwen3-4B-Thinking-2507
library_name: transformers
model_name: output019
tags:
- generated_from_trainer
- trl
- grpo
licence: license
---
# Model Card for output019
This model is a fine-tuned version of [Qwen/Qwen3-4B-Thinking-2507](https://huggingface.co/Qwen/Qwen3-4B-Thinking-2507).
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="QinShiHuangisavailable/output019", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.21.0
- Transformers: 4.55.3
- Pytorch: 2.7.1+cu118
- Datasets: 4.0.0
- Tokenizers: 0.21.4
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
```
|
AnonymousCS/populism_classifier_215
|
AnonymousCS
| 2025-08-30T00:57:40Z | 7 | 0 |
transformers
|
[
"transformers",
"safetensors",
"xlm-roberta",
"text-classification",
"generated_from_trainer",
"base_model:AnonymousCS/populism_xlmr_large",
"base_model:finetune:AnonymousCS/populism_xlmr_large",
"license:mit",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2025-08-26T07:31:27Z |
---
library_name: transformers
license: mit
base_model: AnonymousCS/populism_xlmr_large
tags:
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: populism_classifier_215
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. -->
# populism_classifier_215
This model is a fine-tuned version of [AnonymousCS/populism_xlmr_large](https://huggingface.co/AnonymousCS/populism_xlmr_large) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.1971
- Accuracy: 0.9588
- 1-f1: 0.0
- 1-recall: 0.0
- 1-precision: 0.0
- Balanced Acc: 0.5
## 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: 3e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- num_epochs: 20
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | 1-f1 | 1-recall | 1-precision | Balanced Acc |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:----:|:--------:|:-----------:|:------------:|
| 0.1352 | 1.0 | 97 | 0.1835 | 0.9588 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.0177 | 2.0 | 194 | 0.1926 | 0.9588 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.3155 | 3.0 | 291 | 0.2014 | 0.9588 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.4907 | 4.0 | 388 | 0.1971 | 0.9588 | 0.0 | 0.0 | 0.0 | 0.5 |
### Framework versions
- Transformers 4.46.3
- Pytorch 2.4.1+cu121
- Datasets 3.1.0
- Tokenizers 0.20.3
|
luckeciano/Qwen-2.5-7B-GRPO-NoBaseline-HessianMaskGlobal-0.1-v2_8303
|
luckeciano
| 2025-08-30T00:51:48Z | 0 | 0 |
transformers
|
[
"transformers",
"safetensors",
"qwen2",
"text-generation",
"generated_from_trainer",
"open-r1",
"trl",
"grpo",
"conversational",
"dataset:DigitalLearningGmbH/MATH-lighteval",
"arxiv:2402.03300",
"base_model:Qwen/Qwen2.5-Math-7B",
"base_model:finetune:Qwen/Qwen2.5-Math-7B",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] |
text-generation
| 2025-08-29T21:14:08Z |
---
base_model: Qwen/Qwen2.5-Math-7B
datasets: DigitalLearningGmbH/MATH-lighteval
library_name: transformers
model_name: Qwen-2.5-7B-GRPO-NoBaseline-HessianMaskGlobal-0.1-v2_8303
tags:
- generated_from_trainer
- open-r1
- trl
- grpo
licence: license
---
# Model Card for Qwen-2.5-7B-GRPO-NoBaseline-HessianMaskGlobal-0.1-v2_8303
This model is a fine-tuned version of [Qwen/Qwen2.5-Math-7B](https://huggingface.co/Qwen/Qwen2.5-Math-7B) on the [DigitalLearningGmbH/MATH-lighteval](https://huggingface.co/datasets/DigitalLearningGmbH/MATH-lighteval) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="luckeciano/Qwen-2.5-7B-GRPO-NoBaseline-HessianMaskGlobal-0.1-v2_8303", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/max-ent-llms/PolicyGradientStability/runs/7dciyvz3)
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.16.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.5.1
- Datasets: 3.4.1
- Tokenizers: 0.21.2
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallouédec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
```
|
sivakrishna123/my-jarvis-GGUF
|
sivakrishna123
| 2025-08-30T00:36:28Z | 0 | 0 |
transformers
|
[
"transformers",
"gguf",
"llama",
"text-generation-inference",
"unsloth",
"en",
"license:apache-2.0",
"endpoints_compatible",
"region:us"
] | null | 2025-08-30T00:35:06Z |
---
base_model: unsloth/meta-llama-3.1-8b-bnb-4bit
tags:
- text-generation-inference
- transformers
- unsloth
- llama
- gguf
license: apache-2.0
language:
- en
---
# Uploaded model
- **Developed by:** sivakrishna123
- **License:** apache-2.0
- **Finetuned from model :** unsloth/meta-llama-3.1-8b-bnb-4bit
This llama model was trained 2x faster with [Unsloth](https://github.com/unslothai/unsloth) and Huggingface's TRL library.
[<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>](https://github.com/unslothai/unsloth)
|
kmpartner/k5pcmlra-test
|
kmpartner
| 2025-08-30T00:35:02Z | 102 | 0 |
peft
|
[
"peft",
"tensorboard",
"diffusers",
"safetensors",
"arxiv:1910.09700",
"base_model:segmind/Segmind-Vega",
"base_model:adapter:segmind/Segmind-Vega",
"region:us"
] | null | 2025-04-27T12:32:18Z |
---
library_name: peft
base_model: segmind/Segmind-Vega
---
# Model Card for Model ID
<!-- Provide a quick summary of what the model is/does. -->
## Model Details
### Model Description
<!-- Provide a longer summary of what this model is. -->
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [More Information Needed]
- **Shared by [optional]:** [More Information Needed]
- **Model type:** [More Information Needed]
- **Language(s) (NLP):** [More Information Needed]
- **License:** [More Information Needed]
- **Finetuned from model [optional]:** [More Information Needed]
### Model Sources [optional]
<!-- Provide the basic links for the model. -->
- **Repository:** [More Information Needed]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed]
## Uses
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
### Direct Use
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
[More Information Needed]
### Downstream Use [optional]
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
[More Information Needed]
### Out-of-Scope Use
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
[More Information Needed]
## Bias, Risks, and Limitations
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
[More Information Needed]
### Recommendations
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
## How to Get Started with the Model
Use the code below to get started with the model.
[More Information Needed]
## Training Details
### Training Data
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
[More Information Needed]
### Training Procedure
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
#### Preprocessing [optional]
[More Information Needed]
#### Training Hyperparameters
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
#### Speeds, Sizes, Times [optional]
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
[More Information Needed]
## Evaluation
<!-- This section describes the evaluation protocols and provides the results. -->
### Testing Data, Factors & Metrics
#### Testing Data
<!-- This should link to a Dataset Card if possible. -->
[More Information Needed]
#### Factors
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
[More Information Needed]
#### Metrics
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
[More Information Needed]
### Results
[More Information Needed]
#### Summary
## Model Examination [optional]
<!-- Relevant interpretability work for the model goes here -->
[More Information Needed]
## Environmental Impact
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
- **Hardware Type:** [More Information Needed]
- **Hours used:** [More Information Needed]
- **Cloud Provider:** [More Information Needed]
- **Compute Region:** [More Information Needed]
- **Carbon Emitted:** [More Information Needed]
## Technical Specifications [optional]
### Model Architecture and Objective
[More Information Needed]
### Compute Infrastructure
[More Information Needed]
#### Hardware
[More Information Needed]
#### Software
[More Information Needed]
## Citation [optional]
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
**BibTeX:**
[More Information Needed]
**APA:**
[More Information Needed]
## Glossary [optional]
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
[More Information Needed]
## More Information [optional]
[More Information Needed]
## Model Card Authors [optional]
[More Information Needed]
## Model Card Contact
[More Information Needed]
### Framework versions
- PEFT 0.9.0
|
pavlodp/Qwen2.5-0.5B-Instruct-Gensyn-Swarm-exotic_pawing_wombat
|
pavlodp
| 2025-08-30T00:32:46Z | 27 | 0 |
transformers
|
[
"transformers",
"safetensors",
"qwen2",
"text-generation",
"generated_from_trainer",
"rl-swarm",
"grpo",
"gensyn",
"I am exotic pawing wombat",
"trl",
"genrl-swarm",
"I am exotic_pawing_wombat",
"conversational",
"arxiv:2402.03300",
"base_model:unsloth/Qwen2.5-0.5B-Instruct",
"base_model:finetune:unsloth/Qwen2.5-0.5B-Instruct",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] |
text-generation
| 2025-05-11T04:14:53Z |
---
base_model: unsloth/Qwen2.5-0.5B-Instruct
library_name: transformers
model_name: Qwen2.5-0.5B-Instruct-Gensyn-Swarm-exotic_pawing_wombat
tags:
- generated_from_trainer
- rl-swarm
- grpo
- gensyn
- I am exotic pawing wombat
- trl
- genrl-swarm
- I am exotic_pawing_wombat
licence: license
---
# Model Card for Qwen2.5-0.5B-Instruct-Gensyn-Swarm-exotic_pawing_wombat
This model is a fine-tuned version of [unsloth/Qwen2.5-0.5B-Instruct](https://huggingface.co/unsloth/Qwen2.5-0.5B-Instruct).
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="pavlodp/Qwen2.5-0.5B-Instruct-Gensyn-Swarm-exotic_pawing_wombat", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.17.0
- Transformers: 4.52.3
- Pytorch: 2.7.0
- Datasets: 3.6.0
- Tokenizers: 0.21.1
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallou{\'e}dec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
```
|
thejaminator/cities-backdoor-20250830-step-1000
|
thejaminator
| 2025-08-30T00:26:58Z | 0 | 0 |
peft
|
[
"peft",
"safetensors",
"qwen3",
"base_model:Qwen/Qwen3-8B",
"base_model:adapter:Qwen/Qwen3-8B",
"region:us"
] | null | 2025-08-30T00:26:37Z |
---
base_model: Qwen/Qwen3-8B
library_name: peft
---
# LoRA Adapter for SFT
This is a LoRA (Low-Rank Adaptation) adapter trained using supervised fine-tuning (SFT).
## Base Model
- **Base Model**: `Qwen/Qwen3-8B`
- **Adapter Type**: LoRA
- **Task**: Supervised Fine-Tuning
## Usage
```python
from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel
# Load base model and tokenizer
base_model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3-8B")
tokenizer = AutoTokenizer.from_pretrained("Qwen/Qwen3-8B")
# Load LoRA adapter
model = PeftModel.from_pretrained(base_model, "thejaminator/cities-backdoor-20250830-step-1000")
```
## Training Details
This adapter was trained using supervised fine-tuning on conversation data to improve the model's ability to follow instructions and generate helpful responses.
|
mestersop3/blockassist-bc-cunning_tangled_robin_1756513322
|
mestersop3
| 2025-08-30T00:22:52Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"cunning tangled robin",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-30T00:22:29Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- cunning tangled robin
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
seraphimzzzz/514690
|
seraphimzzzz
| 2025-08-30T00:08:37Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-30T00:08:32Z |
[View on Civ Archive](https://civarchive.com/models/534443?modelVersionId=599639)
|
seraphimzzzz/550018
|
seraphimzzzz
| 2025-08-30T00:06:02Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-30T00:05:57Z |
[View on Civ Archive](https://civarchive.com/models/513019?modelVersionId=635295)
|
shoot32323/Qwen2.5-0.5B-Instruct-Gensyn-Swarm-sly_fluffy_eagle
|
shoot32323
| 2025-08-30T00:01:53Z | 109 | 0 |
transformers
|
[
"transformers",
"safetensors",
"qwen2",
"text-generation",
"rl-swarm",
"genrl-swarm",
"grpo",
"gensyn",
"I am sly_fluffy_eagle",
"arxiv:1910.09700",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"4-bit",
"bitsandbytes",
"region:us"
] |
text-generation
| 2025-08-26T13:52:11Z |
---
library_name: transformers
tags:
- rl-swarm
- genrl-swarm
- grpo
- gensyn
- I am sly_fluffy_eagle
---
# Model Card for Model ID
<!-- Provide a quick summary of what the model is/does. -->
## Model Details
### Model Description
<!-- Provide a longer summary of what this model is. -->
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [More Information Needed]
- **Shared by [optional]:** [More Information Needed]
- **Model type:** [More Information Needed]
- **Language(s) (NLP):** [More Information Needed]
- **License:** [More Information Needed]
- **Finetuned from model [optional]:** [More Information Needed]
### Model Sources [optional]
<!-- Provide the basic links for the model. -->
- **Repository:** [More Information Needed]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed]
## Uses
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
### Direct Use
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
[More Information Needed]
### Downstream Use [optional]
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
[More Information Needed]
### Out-of-Scope Use
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
[More Information Needed]
## Bias, Risks, and Limitations
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
[More Information Needed]
### Recommendations
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
## How to Get Started with the Model
Use the code below to get started with the model.
[More Information Needed]
## Training Details
### Training Data
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
[More Information Needed]
### Training Procedure
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
#### Preprocessing [optional]
[More Information Needed]
#### Training Hyperparameters
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
#### Speeds, Sizes, Times [optional]
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
[More Information Needed]
## Evaluation
<!-- This section describes the evaluation protocols and provides the results. -->
### Testing Data, Factors & Metrics
#### Testing Data
<!-- This should link to a Dataset Card if possible. -->
[More Information Needed]
#### Factors
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
[More Information Needed]
#### Metrics
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
[More Information Needed]
### Results
[More Information Needed]
#### Summary
## Model Examination [optional]
<!-- Relevant interpretability work for the model goes here -->
[More Information Needed]
## Environmental Impact
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
- **Hardware Type:** [More Information Needed]
- **Hours used:** [More Information Needed]
- **Cloud Provider:** [More Information Needed]
- **Compute Region:** [More Information Needed]
- **Carbon Emitted:** [More Information Needed]
## Technical Specifications [optional]
### Model Architecture and Objective
[More Information Needed]
### Compute Infrastructure
[More Information Needed]
#### Hardware
[More Information Needed]
#### Software
[More Information Needed]
## Citation [optional]
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
**BibTeX:**
[More Information Needed]
**APA:**
[More Information Needed]
## Glossary [optional]
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
[More Information Needed]
## More Information [optional]
[More Information Needed]
## Model Card Authors [optional]
[More Information Needed]
## Model Card Contact
[More Information Needed]
|
AnonymousCS/populism_classifier_209
|
AnonymousCS
| 2025-08-29T23:58:57Z | 1 | 0 |
transformers
|
[
"transformers",
"safetensors",
"xlm-roberta",
"text-classification",
"generated_from_trainer",
"base_model:AnonymousCS/populism_xlmr_large",
"base_model:finetune:AnonymousCS/populism_xlmr_large",
"license:mit",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2025-08-26T06:23:59Z |
---
library_name: transformers
license: mit
base_model: AnonymousCS/populism_xlmr_large
tags:
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: populism_classifier_209
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. -->
# populism_classifier_209
This model is a fine-tuned version of [AnonymousCS/populism_xlmr_large](https://huggingface.co/AnonymousCS/populism_xlmr_large) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.2016
- Accuracy: 0.9523
- 1-f1: 0.0
- 1-recall: 0.0
- 1-precision: 0.0
- Balanced Acc: 0.5
## 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: 3e-05
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- num_epochs: 20
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | 1-f1 | 1-recall | 1-precision | Balanced Acc |
|:-------------:|:-----:|:-----:|:---------------:|:--------:|:----:|:--------:|:-----------:|:------------:|
| 0.0183 | 1.0 | 3484 | 0.2156 | 0.9523 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.2729 | 2.0 | 6968 | 0.1968 | 0.9523 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.1379 | 3.0 | 10452 | 0.2050 | 0.9523 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.0329 | 4.0 | 13936 | 0.2010 | 0.9523 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.1365 | 5.0 | 17420 | 0.2016 | 0.9523 | 0.0 | 0.0 | 0.0 | 0.5 |
### Framework versions
- Transformers 4.46.3
- Pytorch 2.4.1+cu121
- Datasets 3.1.0
- Tokenizers 0.20.3
|
amethyst9/766261
|
amethyst9
| 2025-08-29T23:54:26Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T23:54:20Z |
[View on Civ Archive](https://civarchive.com/models/518266?modelVersionId=857241)
|
rvipitkirubbe/blockassist-bc-mottled_foraging_ape_1756509695
|
rvipitkirubbe
| 2025-08-29T23:47:59Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"mottled foraging ape",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T23:47:54Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- mottled foraging ape
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
liukevin666/blockassist-bc-yawning_striped_cassowary_1756511104
|
liukevin666
| 2025-08-29T23:46:56Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"yawning striped cassowary",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T23:46:03Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- yawning striped cassowary
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
MartialTerran/HART-SURYA_model
|
MartialTerran
| 2025-08-29T23:45:13Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-23T18:30:10Z |
### [Fictional] Public Expert Break-Out Session: Evaluating the HART-SURYA Proposal
**Location:** The AI Conference 2025, Pier 48, San Francisco
**Track:** AI Frontiers
**Time:** 3:30 PM, Wednesday, September 17th
The room is packed, with standing room only. The screen behind the panelists displays the title: "Existential AI: Can a Smarter Model Save Our Electrical Grid from the Sun?"
**Moderator:** "Welcome, everyone. We have a special, unscripted session today to discuss a fascinating proposal that emerged from the open-source community, aimed at improving a critical NASA AI model called Surya. The goal of Surya is to understand our sun, but the stakes couldn't be higher. A Carrington-level solar event today could collapse our global power grid, sending us back to the dark ages. The question on the table is a proposal by a 'Martial Terran' called HART, or Heliocentric Adaptive-Rotation Tokenization. Is it a game-changer for predicting these events, or a complex distraction?
"Let's start with the big picture. DJ, as the former US Chief Data Scientist, frame this problem for us."
**NK Palik:** "Gladly. People need to understand this isn't just an academic exercise. We are, at this moment, flying blind. A massive Coronal Mass Ejection, or CME, could hit us with only hours of warning, if that. The result would be trillions in damages and a breakdown of society. It's not *if*, it's *when*. We have the data streaming from the sun, but we're not extracting the maximum intelligence from it. The current Surya model is a great step. But the core question this HART proposal raises is: can we make it fundamentally better? For a problem of this magnitude, we have a national security obligation to chase down every credible performance improvement."
**Moderator:** "Tris, you work on applying AI to grand scientific challenges at Deepmind. What's your take on the HART proposal's scientific merit?"
**Tris Wtiarkenn:** "From a first-principles perspective, it's incredibly elegant. What this 'Martial Terran' correctly identifies is that the current model is forced to waste a huge amount of its capacity learning a basic, predictable kinematic motion: the sun's differential rotation. It's like asking a genius to predict the stock market, but first forcing them to re-derive the laws of gravity every single time they look at the data. HART essentially says: let's handle the predictable physics in the data-processing step. Let's de-rotate the sun in the input data so the transformer can dedicate its *entire* intelligence to the much harder problem—the *intrinsic evolution* of solar features that actually lead to an eruption. It's a classic, beautiful example of physics-informed AI."
**Ion Satoic:** "Elegance is one thing, but petabytes of data are another." All eyes turn to the Berkeley professor and Databricks co-founder. "I read the proposal, and the engineer in me immediately got nervous. This 'Stage 2: Dynamic, Per-Band Image Warping' is computationally non-trivial. For every time-sequence of images, you are calculating a complex, non-linear flow field and resampling the image. You're shifting the computational burden from the model's inference stage to the data-ingestion pipeline. So, while you might get a more efficient *model*, your total pipeline cost and complexity could skyrocket. At NASA's scale, that's a massive engineering challenge. Is the trade-off worth it?"
**Lin Qoia:** "I'm with Ion on this. The proposal itself actually offers a much more practical first step. Why are we even debating the full, complex warping pipeline when 'Optimization 1: Masked Tokenization' is sitting right there?" she asks, leaning into her microphone. "The author points out that 21.5% of the input tokens are just black space. By simply masking out these tokens, we could get a 20% reduction in compute and memory usage *right now* with very low implementation risk. From a production AI standpoint, you always go for the low-hanging fruit first. Let's bank the 20% win, see how the model improves, and then use that as the baseline to evaluate whether the far more complex HART approach provides enough marginal benefit."
**Jure Lekovsec:** "I think we need to be careful about the potential downsides of the HART warping itself," the Stanford professor cautions. "This resampling operation, `grid_sample`, is an interpolation. Interpolation can introduce subtle artifacts or smooth over the very faint, high-frequency signals that might be the critical precursors to a solar flare. You could, in theory, 'de-rotate' the sun so well that you accidentally erase the very signal you're looking for. It's a clever feature engineering step, but it's not without risk. A more robust approach might be to use something like a graph neural network on a spherical projection of the sun, which is more native to the data's geometry and doesn't require resampling the source pixels."
**Christopher Krihoffoch:** "This technical debate is fantastic, but let's bring it back to the ground. Or, rather, to the grid," he says, cutting through the academic back-and-forth. "At the Pentagon's innovation unit, we had a mantra: 'Test it.' Right now, this is a proposal in a GitHub issue. We need a bake-off. It should be a three-way competition. Model 1 is the current Surya baseline. Model 2 is Martial's suggestion, which Lin endorses: Surya with the simple masked tokenization. Model 3 is Martial's full HART implementation. We then run historical data for the 100 biggest solar flares on record through all three models. The winner is the one that gives us the longest, most reliable warning time. Does one model give us 12 hours of warning when another gives us 4? That's the only metric that matters when civilization is on the line. This is a solvable, empirical question."
**NK Palik:** "Chris is exactly right. We need to operationalize this. We can't let the perfect be the enemy of the good. Lin's point is sharp: a 20% efficiency gain is not trivial. That could mean a faster, larger, or more frequently updated model *today*. But Tris's point about the elegance of the HART approach is the long-term goal. By encoding known physics, we could unlock a new level of predictive power. So, the path forward seems clear: implement the mask now. Benchmark the full HART proposal rigorously, paying close attention to Jure's concern about artifacts. And frame the entire effort around Christopher's metric: actionable warning time. We have a clear and present danger, and this proposal lays out a tangible path to improving our defenses."
**Moderator:** "So, the consensus is a pragmatic, two-track approach. An immediate, low-risk optimization and a higher-risk, higher-reward research track, all benchmarked against the single metric of saving the world. It seems even in the world of advanced AI, the simplest solution is often the best place to start. Thank you all for a truly spirited discussion."
|
bah63843/blockassist-bc-plump_fast_antelope_1756510690
|
bah63843
| 2025-08-29T23:39:01Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"plump fast antelope",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T23:38:52Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- plump fast antelope
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
Emanon14/LoRA
|
Emanon14
| 2025-08-29T23:38:18Z | 0 | 45 | null |
[
"text-to-image",
"stable-diffusion",
"stable-diffusion-xl",
"en",
"license:other",
"region:us"
] |
text-to-image
| 2025-02-01T00:26:10Z |
---
license: other
license_name: faipl-1.0-sd
license_link: https://freedevproject.org/faipl-1.0-sd/
language:
- en
pipeline_tag: text-to-image
tags:
- text-to-image
- stable-diffusion
- stable-diffusion-xl
---
# Slider LoRA
## What is this?
- Here is some my LoRA for illustrious.
- You can adjust the character's appearance like a sliders in 3D games.
- You don't need to include specific words in your prompts.
- Just use the LoRA and adjust the weights.
<details>
<summary>Body</summary>
## AreolaeSize_XL_Ilst
![You won't find a sample image here. Some things are simply too fabulous for public display... or maybe I just didn't want to get the README flagged.]()
Adjusts the size of areolae to be smaller/larger.
## AssSize_XL_Ilst
Adjusts the size of ass to be smaller/larger.
## BreastsSize_XL_Ilst
Adjusts the size of breasts to be smaller/larger.
## Height_XL_Ilst
Adjusts the height to be shorter/taller.
## LegLength_XL_Ilst
Adjusts the length of legs to be shorter/taller.
## Muscle_XL_Ilst
Smooths/defines abdominal muscles and ribs.
## Neck_XL_Ilst
Adjusts the length of the neck to be shorter/longer.
## ShoulderSize_XL_Ilst
Adjusts the width of the shoulders to be narrower/wider.
## Stumpy_XL_Ilst
Adjusts the waistline to be thinner/thicker.
## ThighSize_XL_Ilst
Adjusts the size of the thighs to be thinner/thicker.
## WaistSize_XL_Ilst
Adjusts the waist circumference to be thinner/thicker.
</details>
<details>
<summary>Face</summary>
## Chin_XL_Ilst
Adjusts the length of chin to be shorter/taller.
## EyeDistance_XL_Ilst
Adjusts the distance between the eyes to be narrower/wider.
## EyeHeight_XL_Ilst
Adjusts the vertical position of the eyes to be lower/higher.
## EyeSize_XL_Ilst
Adjusts the size of the eyes to be smaller/larger.
## Faceline_XL_Ilst
Adjusts the width of the face to be narrower/wider.
## HeadSize_XL_Ilst
Adjusts the size of the head to be smaller/larger.
## UpperHead_XL_Ilst
Adjusts the length of the head(upper) to be shorter/longer.
</details>
<details>
<summary>Others</summary>
## BreastsMove_XL_Ilst
Moving breasts to down/up.
<u>To generate keyframe images for video generation like a FramePack, Wan, etc...</u>
## HandSize_XL_Ilst
Adjusts the size of the hands to be smaller/larger.
<u>This LoRA may cause a bad anatomy</u>
## PupilWidth_XL_Ilst
Adjusts the width of the Pupils to be narrower/wider.
<u>This LoRA made by ADDifT</u>
</details>
|
sekirr/blockassist-bc-masked_tenacious_whale_1756510449
|
sekirr
| 2025-08-29T23:34:49Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"masked tenacious whale",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T23:34:46Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- masked tenacious whale
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
crystalline7/874980
|
crystalline7
| 2025-08-29T23:32:15Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T23:32:10Z |
[View on Civ Archive](https://civarchive.com/models/842942?modelVersionId=968477)
|
qualcomm/MobileNet-v3-Large
|
qualcomm
| 2025-08-29T23:31:30Z | 84 | 1 |
pytorch
|
[
"pytorch",
"tflite",
"backbone",
"real_time",
"android",
"image-classification",
"arxiv:1905.02244",
"license:other",
"region:us"
] |
image-classification
| 2024-02-25T23:05:55Z |
---
library_name: pytorch
license: other
tags:
- backbone
- real_time
- android
pipeline_tag: image-classification
---
# MobileNet-v3-Large: Optimized for Mobile Deployment
## Imagenet classifier and general purpose backbone
MobileNet-v3-Large is a machine learning model that can classify images from the Imagenet dataset. It can also be used as a backbone in building more complex models for specific use cases.
This model is an implementation of MobileNet-v3-Large found [here](https://github.com/pytorch/vision/blob/main/torchvision/models/mobilenetv3.py).
This repository provides scripts to run MobileNet-v3-Large on Qualcomm® devices.
More details on model performance across various devices, can be found
[here](https://aihub.qualcomm.com/models/mobilenet_v3_large).
### Model Details
- **Model Type:** Model_use_case.image_classification
- **Model Stats:**
- Model checkpoint: Imagenet
- Input resolution: 224x224
- Number of parameters: 5.47M
- Model size (float): 20.9 MB
- Model size (w8a16): 6.35 MB
| Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
|---|---|---|---|---|---|---|---|---|
| MobileNet-v3-Large | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 3.022 ms | 0 - 28 MB | NPU | [MobileNet-v3-Large.tflite](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.tflite) |
| MobileNet-v3-Large | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 2.847 ms | 1 - 24 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.dlc) |
| MobileNet-v3-Large | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 1.344 ms | 0 - 42 MB | NPU | [MobileNet-v3-Large.tflite](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.tflite) |
| MobileNet-v3-Large | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 1.77 ms | 0 - 40 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.dlc) |
| MobileNet-v3-Large | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 0.969 ms | 0 - 93 MB | NPU | [MobileNet-v3-Large.tflite](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.tflite) |
| MobileNet-v3-Large | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 0.936 ms | 0 - 71 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.dlc) |
| MobileNet-v3-Large | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 1.365 ms | 0 - 29 MB | NPU | [MobileNet-v3-Large.tflite](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.tflite) |
| MobileNet-v3-Large | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 1.301 ms | 1 - 25 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.dlc) |
| MobileNet-v3-Large | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 3.022 ms | 0 - 28 MB | NPU | [MobileNet-v3-Large.tflite](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.tflite) |
| MobileNet-v3-Large | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 2.847 ms | 1 - 24 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.dlc) |
| MobileNet-v3-Large | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 0.972 ms | 0 - 93 MB | NPU | [MobileNet-v3-Large.tflite](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.tflite) |
| MobileNet-v3-Large | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 0.936 ms | 0 - 73 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.dlc) |
| MobileNet-v3-Large | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 1.791 ms | 0 - 35 MB | NPU | [MobileNet-v3-Large.tflite](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.tflite) |
| MobileNet-v3-Large | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 1.735 ms | 1 - 31 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.dlc) |
| MobileNet-v3-Large | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 0.971 ms | 0 - 93 MB | NPU | [MobileNet-v3-Large.tflite](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.tflite) |
| MobileNet-v3-Large | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 0.935 ms | 1 - 8 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.dlc) |
| MobileNet-v3-Large | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 1.365 ms | 0 - 29 MB | NPU | [MobileNet-v3-Large.tflite](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.tflite) |
| MobileNet-v3-Large | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 1.301 ms | 1 - 25 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.dlc) |
| MobileNet-v3-Large | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 0.97 ms | 0 - 93 MB | NPU | [MobileNet-v3-Large.tflite](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.tflite) |
| MobileNet-v3-Large | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 0.935 ms | 1 - 7 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.dlc) |
| MobileNet-v3-Large | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 0.814 ms | 0 - 72 MB | NPU | [MobileNet-v3-Large.onnx.zip](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.onnx.zip) |
| MobileNet-v3-Large | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 0.641 ms | 0 - 40 MB | NPU | [MobileNet-v3-Large.tflite](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.tflite) |
| MobileNet-v3-Large | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 0.632 ms | 0 - 34 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.dlc) |
| MobileNet-v3-Large | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 0.552 ms | 0 - 32 MB | NPU | [MobileNet-v3-Large.onnx.zip](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.onnx.zip) |
| MobileNet-v3-Large | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 0.613 ms | 0 - 35 MB | NPU | [MobileNet-v3-Large.tflite](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.tflite) |
| MobileNet-v3-Large | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 0.58 ms | 0 - 26 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.dlc) |
| MobileNet-v3-Large | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 0.561 ms | 1 - 32 MB | NPU | [MobileNet-v3-Large.onnx.zip](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.onnx.zip) |
| MobileNet-v3-Large | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 1.097 ms | 59 - 59 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.dlc) |
| MobileNet-v3-Large | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 0.904 ms | 13 - 13 MB | NPU | [MobileNet-v3-Large.onnx.zip](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large.onnx.zip) |
| MobileNet-v3-Large | w8a16 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 2.053 ms | 0 - 22 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large_w8a16.dlc) |
| MobileNet-v3-Large | w8a16 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 1.227 ms | 0 - 40 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large_w8a16.dlc) |
| MobileNet-v3-Large | w8a16 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 0.933 ms | 0 - 36 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large_w8a16.dlc) |
| MobileNet-v3-Large | w8a16 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 1.141 ms | 0 - 22 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large_w8a16.dlc) |
| MobileNet-v3-Large | w8a16 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | QNN_DLC | 3.402 ms | 0 - 28 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large_w8a16.dlc) |
| MobileNet-v3-Large | w8a16 | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 2.053 ms | 0 - 22 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large_w8a16.dlc) |
| MobileNet-v3-Large | w8a16 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 0.932 ms | 0 - 37 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large_w8a16.dlc) |
| MobileNet-v3-Large | w8a16 | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 1.513 ms | 0 - 30 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large_w8a16.dlc) |
| MobileNet-v3-Large | w8a16 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 0.935 ms | 0 - 36 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large_w8a16.dlc) |
| MobileNet-v3-Large | w8a16 | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 1.141 ms | 0 - 22 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large_w8a16.dlc) |
| MobileNet-v3-Large | w8a16 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 0.93 ms | 0 - 36 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large_w8a16.dlc) |
| MobileNet-v3-Large | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 0.648 ms | 0 - 36 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large_w8a16.dlc) |
| MobileNet-v3-Large | w8a16 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 0.562 ms | 0 - 27 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large_w8a16.dlc) |
| MobileNet-v3-Large | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 1.081 ms | 32 - 32 MB | NPU | [MobileNet-v3-Large.dlc](https://huggingface.co/qualcomm/MobileNet-v3-Large/blob/main/MobileNet-v3-Large_w8a16.dlc) |
## Installation
Install the package via pip:
```bash
pip install qai-hub-models
```
## Configure Qualcomm® AI Hub to run this model on a cloud-hosted device
Sign-in to [Qualcomm® AI Hub](https://app.aihub.qualcomm.com/) with your
Qualcomm® ID. Once signed in navigate to `Account -> Settings -> API Token`.
With this API token, you can configure your client to run models on the cloud
hosted devices.
```bash
qai-hub configure --api_token API_TOKEN
```
Navigate to [docs](https://app.aihub.qualcomm.com/docs/) for more information.
## Demo off target
The package contains a simple end-to-end demo that downloads pre-trained
weights and runs this model on a sample input.
```bash
python -m qai_hub_models.models.mobilenet_v3_large.demo
```
The above demo runs a reference implementation of pre-processing, model
inference, and post processing.
**NOTE**: If you want running in a Jupyter Notebook or Google Colab like
environment, please add the following to your cell (instead of the above).
```
%run -m qai_hub_models.models.mobilenet_v3_large.demo
```
### Run model on a cloud-hosted device
In addition to the demo, you can also run the model on a cloud-hosted Qualcomm®
device. This script does the following:
* Performance check on-device on a cloud-hosted device
* Downloads compiled assets that can be deployed on-device for Android.
* Accuracy check between PyTorch and on-device outputs.
```bash
python -m qai_hub_models.models.mobilenet_v3_large.export
```
## How does this work?
This [export script](https://aihub.qualcomm.com/models/mobilenet_v3_large/qai_hub_models/models/MobileNet-v3-Large/export.py)
leverages [Qualcomm® AI Hub](https://aihub.qualcomm.com/) to optimize, validate, and deploy this model
on-device. Lets go through each step below in detail:
Step 1: **Compile model for on-device deployment**
To compile a PyTorch model for on-device deployment, we first trace the model
in memory using the `jit.trace` and then call the `submit_compile_job` API.
```python
import torch
import qai_hub as hub
from qai_hub_models.models.mobilenet_v3_large import Model
# Load the model
torch_model = Model.from_pretrained()
# Device
device = hub.Device("Samsung Galaxy S24")
# Trace model
input_shape = torch_model.get_input_spec()
sample_inputs = torch_model.sample_inputs()
pt_model = torch.jit.trace(torch_model, [torch.tensor(data[0]) for _, data in sample_inputs.items()])
# Compile model on a specific device
compile_job = hub.submit_compile_job(
model=pt_model,
device=device,
input_specs=torch_model.get_input_spec(),
)
# Get target model to run on-device
target_model = compile_job.get_target_model()
```
Step 2: **Performance profiling on cloud-hosted device**
After compiling models from step 1. Models can be profiled model on-device using the
`target_model`. Note that this scripts runs the model on a device automatically
provisioned in the cloud. Once the job is submitted, you can navigate to a
provided job URL to view a variety of on-device performance metrics.
```python
profile_job = hub.submit_profile_job(
model=target_model,
device=device,
)
```
Step 3: **Verify on-device accuracy**
To verify the accuracy of the model on-device, you can run on-device inference
on sample input data on the same cloud hosted device.
```python
input_data = torch_model.sample_inputs()
inference_job = hub.submit_inference_job(
model=target_model,
device=device,
inputs=input_data,
)
on_device_output = inference_job.download_output_data()
```
With the output of the model, you can compute like PSNR, relative errors or
spot check the output with expected output.
**Note**: This on-device profiling and inference requires access to Qualcomm®
AI Hub. [Sign up for access](https://myaccount.qualcomm.com/signup).
## Run demo on a cloud-hosted device
You can also run the demo on-device.
```bash
python -m qai_hub_models.models.mobilenet_v3_large.demo --eval-mode on-device
```
**NOTE**: If you want running in a Jupyter Notebook or Google Colab like
environment, please add the following to your cell (instead of the above).
```
%run -m qai_hub_models.models.mobilenet_v3_large.demo -- --eval-mode on-device
```
## Deploying compiled model to Android
The models can be deployed using multiple runtimes:
- TensorFlow Lite (`.tflite` export): [This
tutorial](https://www.tensorflow.org/lite/android/quickstart) provides a
guide to deploy the .tflite model in an Android application.
- QNN (`.so` export ): This [sample
app](https://docs.qualcomm.com/bundle/publicresource/topics/80-63442-50/sample_app.html)
provides instructions on how to use the `.so` shared library in an Android application.
## View on Qualcomm® AI Hub
Get more details on MobileNet-v3-Large's performance across various devices [here](https://aihub.qualcomm.com/models/mobilenet_v3_large).
Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
## License
* The license for the original implementation of MobileNet-v3-Large can be found
[here](https://github.com/pytorch/vision/blob/main/LICENSE).
* The license for the compiled assets for on-device deployment can be found [here](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/Qualcomm+AI+Hub+Proprietary+License.pdf)
## References
* [Searching for MobileNetV3](https://arxiv.org/abs/1905.02244)
* [Source Model Implementation](https://github.com/pytorch/vision/blob/main/torchvision/models/mobilenetv3.py)
## Community
* Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
* For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).
|
crystalline7/553055
|
crystalline7
| 2025-08-29T23:30:42Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T23:30:36Z |
[View on Civ Archive](https://civarchive.com/models/572643?modelVersionId=638314)
|
qualcomm/Mistral-3B
|
qualcomm
| 2025-08-29T23:30:20Z | 0 | 5 |
pytorch
|
[
"pytorch",
"llm",
"generative_ai",
"android",
"text-generation",
"license:unknown",
"region:us"
] |
text-generation
| 2024-10-21T23:25:14Z |
---
library_name: pytorch
license: unknown
tags:
- llm
- generative_ai
- android
pipeline_tag: text-generation
---
# Mistral-3B: Optimized for Mobile Deployment
## State-of-the-art large language model useful on a variety of language understanding and generation tasks
Mistral 3B model is Mistral AI's first generation edge model, optimized for optimal performance on Snapdragon platforms.
This model is an implementation of Mistral-3B found [here](https://github.com/mistralai/mistral-inference).
Please contact us to purchase this model. More details on model performance across various devices, can be found [here](https://aihub.qualcomm.com/models/mistral_3b).
**WARNING**: The model assets are not readily available for download due to licensing restrictions.
### Model Details
- **Model Type:** Model_use_case.text_generation
- **Model Stats:**
- Input sequence length for Prompt Processor: 128
- Max context length: 4096
- Num of key-value heads: 8
- Number of parameters: 3B
- Precision: w4a16 + w8a16 (few layers)
- Use: Initiate conversation with prompt-processor and then token generator for subsequent iterations.
- Minimum QNN SDK version required: 2.27.7
- Supported languages: English.
- TTFT: Time To First Token is the time it takes to generate the first response token. This is expressed as a range because it varies based on the length of the prompt. The lower bound is for a short prompt (up to 128 tokens, i.e., one iteration of the prompt processor) and the upper bound is for a prompt using the full context length (4096 tokens).
- Response Rate: Rate of response generation after the first response token.
| Model | Precision | Device | Chipset | Target Runtime | Response Rate (tokens per second) | Time To First Token (range, seconds)
|---|---|---|---|---|---|
| Mistral-3B | w4a16 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_CONTEXT_BINARY | 21.05 | 0.092289 - 2.9532736 | -- | -- |
## Deploying Mistral 3B on-device
Please follow the [LLM on-device deployment](https://github.com/quic/ai-hub-apps/tree/main/tutorials/llm_on_genie) tutorial.
## References
* [Source Model Implementation](https://github.com/mistralai/mistral-inference)
## Community
* Join [our AI Hub Slack community](https://qualcomm-ai-hub.slack.com/join/shared_invite/zt-2d5zsmas3-Sj0Q9TzslueCjS31eXG2UA#/shared-invite/email) to collaborate, post questions and learn more about on-device AI.
* For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).
## Usage and Limitations
Model may not be used for or in connection with any of the following applications:
- Accessing essential private and public services and benefits;
- Administration of justice and democratic processes;
- Assessing or recognizing the emotional state of a person;
- Biometric and biometrics-based systems, including categorization of persons based on sensitive characteristics;
- Education and vocational training;
- Employment and workers management;
- Exploitation of the vulnerabilities of persons resulting in harmful behavior;
- General purpose social scoring;
- Law enforcement;
- Management and operation of critical infrastructure;
- Migration, asylum and border control management;
- Predictive policing;
- Real-time remote biometric identification in public spaces;
- Recommender systems of social media platforms;
- Scraping of facial images (from the internet or otherwise); and/or
- Subliminal manipulation
|
luckeciano/Qwen-2.5-7B-GRPO-NoBaseline-FisherFull-0.1-1e-4-1e-7-v2_2775
|
luckeciano
| 2025-08-29T23:27:50Z | 0 | 0 |
transformers
|
[
"transformers",
"safetensors",
"qwen2",
"text-generation",
"generated_from_trainer",
"open-r1",
"trl",
"grpo",
"conversational",
"dataset:DigitalLearningGmbH/MATH-lighteval",
"arxiv:2402.03300",
"base_model:Qwen/Qwen2.5-Math-7B",
"base_model:finetune:Qwen/Qwen2.5-Math-7B",
"autotrain_compatible",
"text-generation-inference",
"endpoints_compatible",
"region:us"
] |
text-generation
| 2025-08-29T19:17:35Z |
---
base_model: Qwen/Qwen2.5-Math-7B
datasets: DigitalLearningGmbH/MATH-lighteval
library_name: transformers
model_name: Qwen-2.5-7B-GRPO-NoBaseline-FisherFull-0.1-1e-4-1e-7-v2_2775
tags:
- generated_from_trainer
- open-r1
- trl
- grpo
licence: license
---
# Model Card for Qwen-2.5-7B-GRPO-NoBaseline-FisherFull-0.1-1e-4-1e-7-v2_2775
This model is a fine-tuned version of [Qwen/Qwen2.5-Math-7B](https://huggingface.co/Qwen/Qwen2.5-Math-7B) on the [DigitalLearningGmbH/MATH-lighteval](https://huggingface.co/datasets/DigitalLearningGmbH/MATH-lighteval) dataset.
It has been trained using [TRL](https://github.com/huggingface/trl).
## Quick start
```python
from transformers import pipeline
question = "If you had a time machine, but could only go to the past or the future once and never return, which would you choose and why?"
generator = pipeline("text-generation", model="luckeciano/Qwen-2.5-7B-GRPO-NoBaseline-FisherFull-0.1-1e-4-1e-7-v2_2775", device="cuda")
output = generator([{"role": "user", "content": question}], max_new_tokens=128, return_full_text=False)[0]
print(output["generated_text"])
```
## Training procedure
[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="150" height="24"/>](https://wandb.ai/max-ent-llms/PolicyGradientStability/runs/zn0exbaw)
This model was trained with GRPO, a method introduced in [DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models](https://huggingface.co/papers/2402.03300).
### Framework versions
- TRL: 0.16.0.dev0
- Transformers: 4.49.0
- Pytorch: 2.5.1
- Datasets: 3.4.1
- Tokenizers: 0.21.2
## Citations
Cite GRPO as:
```bibtex
@article{zhihong2024deepseekmath,
title = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
year = 2024,
eprint = {arXiv:2402.03300},
}
```
Cite TRL as:
```bibtex
@misc{vonwerra2022trl,
title = {{TRL: Transformer Reinforcement Learning}},
author = {Leandro von Werra and Younes Belkada and Lewis Tunstall and Edward Beeching and Tristan Thrush and Nathan Lambert and Shengyi Huang and Kashif Rasul and Quentin Gallouédec},
year = 2020,
journal = {GitHub repository},
publisher = {GitHub},
howpublished = {\url{https://github.com/huggingface/trl}}
}
```
|
seraphimzzzz/502065
|
seraphimzzzz
| 2025-08-29T23:27:20Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T23:27:14Z |
[View on Civ Archive](https://civarchive.com/models/528094?modelVersionId=586768)
|
qualcomm/LeViT
|
qualcomm
| 2025-08-29T23:27:16Z | 36 | 0 |
pytorch
|
[
"pytorch",
"tflite",
"android",
"image-classification",
"arxiv:2104.01136",
"license:other",
"region:us"
] |
image-classification
| 2025-01-23T01:29:46Z |
---
library_name: pytorch
license: other
tags:
- android
pipeline_tag: image-classification
---
# LeViT: Optimized for Mobile Deployment
## Imagenet classifier and general purpose backbone
LeViT is a vision transformer model that can classify images from the Imagenet dataset.
This model is an implementation of LeViT found [here](https://github.com/facebookresearch/LeViT).
This repository provides scripts to run LeViT on Qualcomm® devices.
More details on model performance across various devices, can be found
[here](https://aihub.qualcomm.com/models/levit).
### Model Details
- **Model Type:** Model_use_case.image_classification
- **Model Stats:**
- Model checkpoint: LeViT-128S
- Input resolution: 224x224
- Number of parameters: 7.82M
- Model size (float): 29.9 MB
- Model size (w8a16): 8.83 MB
| Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
|---|---|---|---|---|---|---|---|---|
| LeViT | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 4.109 ms | 0 - 42 MB | NPU | [LeViT.tflite](https://huggingface.co/qualcomm/LeViT/blob/main/LeViT.tflite) |
| LeViT | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 1.861 ms | 0 - 50 MB | NPU | [LeViT.tflite](https://huggingface.co/qualcomm/LeViT/blob/main/LeViT.tflite) |
| LeViT | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 1.569 ms | 0 - 85 MB | NPU | [LeViT.tflite](https://huggingface.co/qualcomm/LeViT/blob/main/LeViT.tflite) |
| LeViT | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 2.091 ms | 0 - 43 MB | NPU | [LeViT.tflite](https://huggingface.co/qualcomm/LeViT/blob/main/LeViT.tflite) |
| LeViT | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 1.571 ms | 0 - 88 MB | NPU | [LeViT.tflite](https://huggingface.co/qualcomm/LeViT/blob/main/LeViT.tflite) |
| LeViT | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 1.56 ms | 1 - 51 MB | NPU | [LeViT.onnx.zip](https://huggingface.co/qualcomm/LeViT/blob/main/LeViT.onnx.zip) |
| LeViT | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 1.056 ms | 0 - 53 MB | NPU | [LeViT.tflite](https://huggingface.co/qualcomm/LeViT/blob/main/LeViT.tflite) |
| LeViT | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 1.04 ms | 0 - 47 MB | NPU | [LeViT.onnx.zip](https://huggingface.co/qualcomm/LeViT/blob/main/LeViT.onnx.zip) |
| LeViT | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 0.839 ms | 0 - 48 MB | NPU | [LeViT.tflite](https://huggingface.co/qualcomm/LeViT/blob/main/LeViT.tflite) |
| LeViT | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 1.118 ms | 1 - 45 MB | NPU | [LeViT.onnx.zip](https://huggingface.co/qualcomm/LeViT/blob/main/LeViT.onnx.zip) |
| LeViT | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 1.63 ms | 16 - 16 MB | NPU | [LeViT.onnx.zip](https://huggingface.co/qualcomm/LeViT/blob/main/LeViT.onnx.zip) |
| LeViT | w8a16 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 2.756 ms | 0 - 26 MB | NPU | [LeViT.dlc](https://huggingface.co/qualcomm/LeViT/blob/main/LeViT_w8a16.dlc) |
| LeViT | w8a16 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 1.575 ms | 0 - 35 MB | NPU | [LeViT.dlc](https://huggingface.co/qualcomm/LeViT/blob/main/LeViT_w8a16.dlc) |
| LeViT | w8a16 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 1.354 ms | 0 - 10 MB | NPU | [LeViT.dlc](https://huggingface.co/qualcomm/LeViT/blob/main/LeViT_w8a16.dlc) |
| LeViT | w8a16 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 1.675 ms | 0 - 25 MB | NPU | [LeViT.dlc](https://huggingface.co/qualcomm/LeViT/blob/main/LeViT_w8a16.dlc) |
| LeViT | w8a16 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | QNN_DLC | 4.133 ms | 0 - 34 MB | NPU | [LeViT.dlc](https://huggingface.co/qualcomm/LeViT/blob/main/LeViT_w8a16.dlc) |
| LeViT | w8a16 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 1.371 ms | 0 - 11 MB | NPU | [LeViT.dlc](https://huggingface.co/qualcomm/LeViT/blob/main/LeViT_w8a16.dlc) |
| LeViT | w8a16 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 3.983 ms | 0 - 41 MB | NPU | [LeViT.onnx.zip](https://huggingface.co/qualcomm/LeViT/blob/main/LeViT_w8a16.onnx.zip) |
| LeViT | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 0.917 ms | 0 - 35 MB | NPU | [LeViT.dlc](https://huggingface.co/qualcomm/LeViT/blob/main/LeViT_w8a16.dlc) |
| LeViT | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 2.881 ms | 0 - 68 MB | NPU | [LeViT.onnx.zip](https://huggingface.co/qualcomm/LeViT/blob/main/LeViT_w8a16.onnx.zip) |
| LeViT | w8a16 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 0.779 ms | 0 - 25 MB | NPU | [LeViT.dlc](https://huggingface.co/qualcomm/LeViT/blob/main/LeViT_w8a16.dlc) |
| LeViT | w8a16 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 3.275 ms | 0 - 56 MB | NPU | [LeViT.onnx.zip](https://huggingface.co/qualcomm/LeViT/blob/main/LeViT_w8a16.onnx.zip) |
| LeViT | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 1.565 ms | 12 - 12 MB | NPU | [LeViT.dlc](https://huggingface.co/qualcomm/LeViT/blob/main/LeViT_w8a16.dlc) |
| LeViT | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 4.249 ms | 14 - 14 MB | NPU | [LeViT.onnx.zip](https://huggingface.co/qualcomm/LeViT/blob/main/LeViT_w8a16.onnx.zip) |
## Installation
Install the package via pip:
```bash
pip install "qai-hub-models[levit]"
```
## Configure Qualcomm® AI Hub to run this model on a cloud-hosted device
Sign-in to [Qualcomm® AI Hub](https://app.aihub.qualcomm.com/) with your
Qualcomm® ID. Once signed in navigate to `Account -> Settings -> API Token`.
With this API token, you can configure your client to run models on the cloud
hosted devices.
```bash
qai-hub configure --api_token API_TOKEN
```
Navigate to [docs](https://app.aihub.qualcomm.com/docs/) for more information.
## Demo off target
The package contains a simple end-to-end demo that downloads pre-trained
weights and runs this model on a sample input.
```bash
python -m qai_hub_models.models.levit.demo
```
The above demo runs a reference implementation of pre-processing, model
inference, and post processing.
**NOTE**: If you want running in a Jupyter Notebook or Google Colab like
environment, please add the following to your cell (instead of the above).
```
%run -m qai_hub_models.models.levit.demo
```
### Run model on a cloud-hosted device
In addition to the demo, you can also run the model on a cloud-hosted Qualcomm®
device. This script does the following:
* Performance check on-device on a cloud-hosted device
* Downloads compiled assets that can be deployed on-device for Android.
* Accuracy check between PyTorch and on-device outputs.
```bash
python -m qai_hub_models.models.levit.export
```
## How does this work?
This [export script](https://aihub.qualcomm.com/models/levit/qai_hub_models/models/LeViT/export.py)
leverages [Qualcomm® AI Hub](https://aihub.qualcomm.com/) to optimize, validate, and deploy this model
on-device. Lets go through each step below in detail:
Step 1: **Compile model for on-device deployment**
To compile a PyTorch model for on-device deployment, we first trace the model
in memory using the `jit.trace` and then call the `submit_compile_job` API.
```python
import torch
import qai_hub as hub
from qai_hub_models.models.levit import Model
# Load the model
torch_model = Model.from_pretrained()
# Device
device = hub.Device("Samsung Galaxy S24")
# Trace model
input_shape = torch_model.get_input_spec()
sample_inputs = torch_model.sample_inputs()
pt_model = torch.jit.trace(torch_model, [torch.tensor(data[0]) for _, data in sample_inputs.items()])
# Compile model on a specific device
compile_job = hub.submit_compile_job(
model=pt_model,
device=device,
input_specs=torch_model.get_input_spec(),
)
# Get target model to run on-device
target_model = compile_job.get_target_model()
```
Step 2: **Performance profiling on cloud-hosted device**
After compiling models from step 1. Models can be profiled model on-device using the
`target_model`. Note that this scripts runs the model on a device automatically
provisioned in the cloud. Once the job is submitted, you can navigate to a
provided job URL to view a variety of on-device performance metrics.
```python
profile_job = hub.submit_profile_job(
model=target_model,
device=device,
)
```
Step 3: **Verify on-device accuracy**
To verify the accuracy of the model on-device, you can run on-device inference
on sample input data on the same cloud hosted device.
```python
input_data = torch_model.sample_inputs()
inference_job = hub.submit_inference_job(
model=target_model,
device=device,
inputs=input_data,
)
on_device_output = inference_job.download_output_data()
```
With the output of the model, you can compute like PSNR, relative errors or
spot check the output with expected output.
**Note**: This on-device profiling and inference requires access to Qualcomm®
AI Hub. [Sign up for access](https://myaccount.qualcomm.com/signup).
## Run demo on a cloud-hosted device
You can also run the demo on-device.
```bash
python -m qai_hub_models.models.levit.demo --eval-mode on-device
```
**NOTE**: If you want running in a Jupyter Notebook or Google Colab like
environment, please add the following to your cell (instead of the above).
```
%run -m qai_hub_models.models.levit.demo -- --eval-mode on-device
```
## Deploying compiled model to Android
The models can be deployed using multiple runtimes:
- TensorFlow Lite (`.tflite` export): [This
tutorial](https://www.tensorflow.org/lite/android/quickstart) provides a
guide to deploy the .tflite model in an Android application.
- QNN (`.so` export ): This [sample
app](https://docs.qualcomm.com/bundle/publicresource/topics/80-63442-50/sample_app.html)
provides instructions on how to use the `.so` shared library in an Android application.
## View on Qualcomm® AI Hub
Get more details on LeViT's performance across various devices [here](https://aihub.qualcomm.com/models/levit).
Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
## License
* The license for the original implementation of LeViT can be found
[here](https://github.com/facebookresearch/LeViT?tab=Apache-2.0-1-ov-file).
* The license for the compiled assets for on-device deployment can be found [here](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/Qualcomm+AI+Hub+Proprietary+License.pdf)
## References
* [LeViT: a Vision Transformer in ConvNet's Clothing for Faster Inference](https://arxiv.org/abs/2104.01136)
* [Source Model Implementation](https://github.com/facebookresearch/LeViT)
## Community
* Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
* For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).
|
katanyasekolah/blockassist-bc-silky_sprightly_cassowary_1756508397
|
katanyasekolah
| 2025-08-29T23:26:13Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"silky sprightly cassowary",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T23:26:10Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- silky sprightly cassowary
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
qualcomm/HuggingFace-WavLM-Base-Plus
|
qualcomm
| 2025-08-29T23:25:54Z | 32 | 5 |
pytorch
|
[
"pytorch",
"tflite",
"backbone",
"android",
"automatic-speech-recognition",
"arxiv:2110.13900",
"license:other",
"region:us"
] |
automatic-speech-recognition
| 2024-02-25T22:46:56Z |
---
library_name: pytorch
license: other
tags:
- backbone
- android
pipeline_tag: automatic-speech-recognition
---
# HuggingFace-WavLM-Base-Plus: Optimized for Mobile Deployment
## Real-time Speech processing
HuggingFaceWavLMBasePlus is a real time speech processing backbone based on Microsoft's WavLM model.
This model is an implementation of HuggingFace-WavLM-Base-Plus found [here](https://huggingface.co/patrickvonplaten/wavlm-libri-clean-100h-base-plus/tree/main).
This repository provides scripts to run HuggingFace-WavLM-Base-Plus on Qualcomm® devices.
More details on model performance across various devices, can be found
[here](https://aihub.qualcomm.com/models/huggingface_wavlm_base_plus).
### Model Details
- **Model Type:** Model_use_case.speech_recognition
- **Model Stats:**
- Model checkpoint: wavlm-libri-clean-100h-base-plus
- Input resolution: 1x320000
- Number of parameters: 95.1M
- Model size (float): 363 MB
| Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
|---|---|---|---|---|---|---|---|---|
| HuggingFace-WavLM-Base-Plus | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 799.175 ms | 0 - 769 MB | NPU | [HuggingFace-WavLM-Base-Plus.tflite](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.tflite) |
| HuggingFace-WavLM-Base-Plus | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 851.696 ms | 1 - 766 MB | NPU | [HuggingFace-WavLM-Base-Plus.dlc](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.dlc) |
| HuggingFace-WavLM-Base-Plus | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 579.676 ms | 1 - 1092 MB | NPU | [HuggingFace-WavLM-Base-Plus.dlc](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.dlc) |
| HuggingFace-WavLM-Base-Plus | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 246.813 ms | 0 - 54 MB | NPU | [HuggingFace-WavLM-Base-Plus.tflite](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.tflite) |
| HuggingFace-WavLM-Base-Plus | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 286.854 ms | 1 - 40 MB | NPU | [HuggingFace-WavLM-Base-Plus.dlc](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.dlc) |
| HuggingFace-WavLM-Base-Plus | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 297.214 ms | 0 - 769 MB | NPU | [HuggingFace-WavLM-Base-Plus.tflite](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.tflite) |
| HuggingFace-WavLM-Base-Plus | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 799.175 ms | 0 - 769 MB | NPU | [HuggingFace-WavLM-Base-Plus.tflite](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.tflite) |
| HuggingFace-WavLM-Base-Plus | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 851.696 ms | 1 - 766 MB | NPU | [HuggingFace-WavLM-Base-Plus.dlc](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.dlc) |
| HuggingFace-WavLM-Base-Plus | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 252.816 ms | 0 - 55 MB | NPU | [HuggingFace-WavLM-Base-Plus.tflite](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.tflite) |
| HuggingFace-WavLM-Base-Plus | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 289.601 ms | 0 - 45 MB | NPU | [HuggingFace-WavLM-Base-Plus.dlc](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.dlc) |
| HuggingFace-WavLM-Base-Plus | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 574.211 ms | 0 - 989 MB | NPU | [HuggingFace-WavLM-Base-Plus.tflite](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.tflite) |
| HuggingFace-WavLM-Base-Plus | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 438.607 ms | 1 - 919 MB | NPU | [HuggingFace-WavLM-Base-Plus.dlc](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.dlc) |
| HuggingFace-WavLM-Base-Plus | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 251.051 ms | 0 - 55 MB | NPU | [HuggingFace-WavLM-Base-Plus.tflite](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.tflite) |
| HuggingFace-WavLM-Base-Plus | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 289.072 ms | 0 - 47 MB | NPU | [HuggingFace-WavLM-Base-Plus.dlc](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.dlc) |
| HuggingFace-WavLM-Base-Plus | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 297.214 ms | 0 - 769 MB | NPU | [HuggingFace-WavLM-Base-Plus.tflite](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.tflite) |
| HuggingFace-WavLM-Base-Plus | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 247.296 ms | 0 - 54 MB | NPU | [HuggingFace-WavLM-Base-Plus.tflite](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.tflite) |
| HuggingFace-WavLM-Base-Plus | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 288.266 ms | 0 - 46 MB | NPU | [HuggingFace-WavLM-Base-Plus.dlc](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.dlc) |
| HuggingFace-WavLM-Base-Plus | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 474.061 ms | 1 - 49 MB | NPU | [HuggingFace-WavLM-Base-Plus.onnx.zip](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.onnx.zip) |
| HuggingFace-WavLM-Base-Plus | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 176.905 ms | 0 - 824 MB | NPU | [HuggingFace-WavLM-Base-Plus.tflite](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.tflite) |
| HuggingFace-WavLM-Base-Plus | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 212.337 ms | 0 - 846 MB | NPU | [HuggingFace-WavLM-Base-Plus.dlc](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.dlc) |
| HuggingFace-WavLM-Base-Plus | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 345.939 ms | 0 - 892 MB | NPU | [HuggingFace-WavLM-Base-Plus.onnx.zip](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.onnx.zip) |
| HuggingFace-WavLM-Base-Plus | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 164.613 ms | 0 - 766 MB | NPU | [HuggingFace-WavLM-Base-Plus.tflite](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.tflite) |
| HuggingFace-WavLM-Base-Plus | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 218.802 ms | 0 - 703 MB | NPU | [HuggingFace-WavLM-Base-Plus.dlc](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.dlc) |
| HuggingFace-WavLM-Base-Plus | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 329.417 ms | 1 - 810 MB | NPU | [HuggingFace-WavLM-Base-Plus.onnx.zip](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.onnx.zip) |
| HuggingFace-WavLM-Base-Plus | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 291.485 ms | 332 - 332 MB | NPU | [HuggingFace-WavLM-Base-Plus.dlc](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.dlc) |
| HuggingFace-WavLM-Base-Plus | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 497.079 ms | 205 - 205 MB | NPU | [HuggingFace-WavLM-Base-Plus.onnx.zip](https://huggingface.co/qualcomm/HuggingFace-WavLM-Base-Plus/blob/main/HuggingFace-WavLM-Base-Plus.onnx.zip) |
## Installation
Install the package via pip:
```bash
pip install "qai-hub-models[huggingface-wavlm-base-plus]"
```
## Configure Qualcomm® AI Hub to run this model on a cloud-hosted device
Sign-in to [Qualcomm® AI Hub](https://app.aihub.qualcomm.com/) with your
Qualcomm® ID. Once signed in navigate to `Account -> Settings -> API Token`.
With this API token, you can configure your client to run models on the cloud
hosted devices.
```bash
qai-hub configure --api_token API_TOKEN
```
Navigate to [docs](https://app.aihub.qualcomm.com/docs/) for more information.
## Demo off target
The package contains a simple end-to-end demo that downloads pre-trained
weights and runs this model on a sample input.
```bash
python -m qai_hub_models.models.huggingface_wavlm_base_plus.demo
```
The above demo runs a reference implementation of pre-processing, model
inference, and post processing.
**NOTE**: If you want running in a Jupyter Notebook or Google Colab like
environment, please add the following to your cell (instead of the above).
```
%run -m qai_hub_models.models.huggingface_wavlm_base_plus.demo
```
### Run model on a cloud-hosted device
In addition to the demo, you can also run the model on a cloud-hosted Qualcomm®
device. This script does the following:
* Performance check on-device on a cloud-hosted device
* Downloads compiled assets that can be deployed on-device for Android.
* Accuracy check between PyTorch and on-device outputs.
```bash
python -m qai_hub_models.models.huggingface_wavlm_base_plus.export
```
## How does this work?
This [export script](https://aihub.qualcomm.com/models/huggingface_wavlm_base_plus/qai_hub_models/models/HuggingFace-WavLM-Base-Plus/export.py)
leverages [Qualcomm® AI Hub](https://aihub.qualcomm.com/) to optimize, validate, and deploy this model
on-device. Lets go through each step below in detail:
Step 1: **Compile model for on-device deployment**
To compile a PyTorch model for on-device deployment, we first trace the model
in memory using the `jit.trace` and then call the `submit_compile_job` API.
```python
import torch
import qai_hub as hub
from qai_hub_models.models.huggingface_wavlm_base_plus import Model
# Load the model
torch_model = Model.from_pretrained()
# Device
device = hub.Device("Samsung Galaxy S24")
# Trace model
input_shape = torch_model.get_input_spec()
sample_inputs = torch_model.sample_inputs()
pt_model = torch.jit.trace(torch_model, [torch.tensor(data[0]) for _, data in sample_inputs.items()])
# Compile model on a specific device
compile_job = hub.submit_compile_job(
model=pt_model,
device=device,
input_specs=torch_model.get_input_spec(),
)
# Get target model to run on-device
target_model = compile_job.get_target_model()
```
Step 2: **Performance profiling on cloud-hosted device**
After compiling models from step 1. Models can be profiled model on-device using the
`target_model`. Note that this scripts runs the model on a device automatically
provisioned in the cloud. Once the job is submitted, you can navigate to a
provided job URL to view a variety of on-device performance metrics.
```python
profile_job = hub.submit_profile_job(
model=target_model,
device=device,
)
```
Step 3: **Verify on-device accuracy**
To verify the accuracy of the model on-device, you can run on-device inference
on sample input data on the same cloud hosted device.
```python
input_data = torch_model.sample_inputs()
inference_job = hub.submit_inference_job(
model=target_model,
device=device,
inputs=input_data,
)
on_device_output = inference_job.download_output_data()
```
With the output of the model, you can compute like PSNR, relative errors or
spot check the output with expected output.
**Note**: This on-device profiling and inference requires access to Qualcomm®
AI Hub. [Sign up for access](https://myaccount.qualcomm.com/signup).
## Run demo on a cloud-hosted device
You can also run the demo on-device.
```bash
python -m qai_hub_models.models.huggingface_wavlm_base_plus.demo --eval-mode on-device
```
**NOTE**: If you want running in a Jupyter Notebook or Google Colab like
environment, please add the following to your cell (instead of the above).
```
%run -m qai_hub_models.models.huggingface_wavlm_base_plus.demo -- --eval-mode on-device
```
## Deploying compiled model to Android
The models can be deployed using multiple runtimes:
- TensorFlow Lite (`.tflite` export): [This
tutorial](https://www.tensorflow.org/lite/android/quickstart) provides a
guide to deploy the .tflite model in an Android application.
- QNN (`.so` export ): This [sample
app](https://docs.qualcomm.com/bundle/publicresource/topics/80-63442-50/sample_app.html)
provides instructions on how to use the `.so` shared library in an Android application.
## View on Qualcomm® AI Hub
Get more details on HuggingFace-WavLM-Base-Plus's performance across various devices [here](https://aihub.qualcomm.com/models/huggingface_wavlm_base_plus).
Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
## License
* The license for the original implementation of HuggingFace-WavLM-Base-Plus can be found
[here](https://github.com/microsoft/unilm/blob/master/LICENSE).
* The license for the compiled assets for on-device deployment can be found [here](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/Qualcomm+AI+Hub+Proprietary+License.pdf)
## References
* [WavLM: Large-Scale Self-Supervised Pre-Training for Full Stack Speech Processing](https://arxiv.org/abs/2110.13900)
* [Source Model Implementation](https://huggingface.co/patrickvonplaten/wavlm-libri-clean-100h-base-plus/tree/main)
## Community
* Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
* For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).
|
ultratopaz/802801
|
ultratopaz
| 2025-08-29T23:25:27Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T23:25:27Z |
[View on Civ Archive](https://civarchive.com/models/798412?modelVersionId=894328)
|
onurulu17/Qwen2.5-VL-7B-CXR
|
onurulu17
| 2025-08-29T23:23:47Z | 0 | 0 | null |
[
"safetensors",
"region:us"
] | null | 2025-08-29T21:50:08Z |
# Qwen2.5-VL-7B-CXR
## Model Description
This is a **Vision-Language model** (VLM) based on Qwen2.5-VL-7B-Instruct, fine-tuned on **MIMIC-CXR** chest X-ray dataset using **QLoRA PEFT**. The model can analyze chest X-ray images and provide **findings and impressions** in a clear, concise, and medically relevant manner.
**Base model:** Qwen/Qwen2.5-VL-7B-Instruct
**Fine-tuned with:** LoRA PEFT on MIMIC-CXR dataset
**License:** Apache 2.0
---
## Intended Use
- Automatic chest X-ray report generation
- Radiology VQA and clinical decision support
- Research purposes for medical image-language tasks
**Caution:** Use under supervision of medical professionals; outputs are for informational purposes and not a substitute for professional medical advice.
---
## How to Use
```python
from transformers import Qwen2_5_VLForConditionalGeneration, AutoProcessor
from peft import PeftModel
from qwen_vl_utils import process_vision_info
import torch
base_model_id = "Qwen/Qwen2.5-VL-7B-Instruct"
adapter_id = "onurulu17/Qwen2.5-VL-7B-CXR"
# Load base model
model = Qwen2_5_VLForConditionalGeneration.from_pretrained(
base_model_id,
device_map="auto",
torch_dtype=torch.bfloat16,
)
# Load LoRA adapter
model = PeftModel.from_pretrained(model, adapter_id)
# Processor
processor = AutoProcessor.from_pretrained(base_model_id)
# Example inference
def generate_text_from_sample(model, processor, sample, max_new_tokens=1024, device="cuda"):
text_input = processor.apply_chat_template(sample[:1], tokenize=False, add_generation_prompt=True)
image_inputs, _ = process_vision_info(sample)
model_inputs = processor(text=[text_input], images=image_inputs, return_tensors="pt").to(device)
generated_ids = model.generate(**model_inputs, max_new_tokens=max_new_tokens)
trimmed_generated_ids = [out_ids[len(in_ids):] for in_ids, out_ids in zip(model_inputs.input_ids, generated_ids)]
output_text = processor.batch_decode(trimmed_generated_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)
return output_text[0]
# Example usage
sample = [
{'role': 'user',
'content': [{'type': 'image', 'image': "./chest_xray_image.jpg"},
{'type': 'text', 'text': 'Please analyze this chest X-ray and provide the findings and impression.'}]}
]
output = generate_text_from_sample(model, processor, sample)
print(output)
|
qualcomm/FFNet-78S-LowRes
|
qualcomm
| 2025-08-29T23:22:58Z | 66 | 1 |
pytorch
|
[
"pytorch",
"tflite",
"real_time",
"android",
"image-segmentation",
"arxiv:2206.08236",
"license:other",
"region:us"
] |
image-segmentation
| 2024-02-25T22:45:20Z |
---
library_name: pytorch
license: other
tags:
- real_time
- android
pipeline_tag: image-segmentation
---
# FFNet-78S-LowRes: Optimized for Mobile Deployment
## Semantic segmentation for automotive street scenes
FFNet-78S-LowRes is a "fuss-free network" that segments street scene images with per-pixel classes like road, sidewalk, and pedestrian. Trained on the Cityscapes dataset.
This model is an implementation of FFNet-78S-LowRes found [here](https://github.com/Qualcomm-AI-research/FFNet).
This repository provides scripts to run FFNet-78S-LowRes on Qualcomm® devices.
More details on model performance across various devices, can be found
[here](https://aihub.qualcomm.com/models/ffnet_78s_lowres).
### Model Details
- **Model Type:** Model_use_case.semantic_segmentation
- **Model Stats:**
- Model checkpoint: ffnet78S_BCC_cityscapes_state_dict_quarts_pre_down
- Input resolution: 1024x512
- Number of output classes: 19
- Number of parameters: 26.8M
- Model size (float): 102 MB
- Model size (w8a8): 26.0 MB
| Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
|---|---|---|---|---|---|---|---|---|
| FFNet-78S-LowRes | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 48.871 ms | 1 - 61 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.tflite) |
| FFNet-78S-LowRes | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 49.204 ms | 6 - 33 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.dlc) |
| FFNet-78S-LowRes | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 16.461 ms | 1 - 116 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.tflite) |
| FFNet-78S-LowRes | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 23.974 ms | 4 - 40 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.dlc) |
| FFNet-78S-LowRes | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 12.602 ms | 0 - 357 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.tflite) |
| FFNet-78S-LowRes | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 13.574 ms | 3 - 32 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.dlc) |
| FFNet-78S-LowRes | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 17.121 ms | 1 - 62 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.tflite) |
| FFNet-78S-LowRes | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 17.931 ms | 3 - 30 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.dlc) |
| FFNet-78S-LowRes | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 48.871 ms | 1 - 61 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.tflite) |
| FFNet-78S-LowRes | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 49.204 ms | 6 - 33 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.dlc) |
| FFNet-78S-LowRes | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 12.553 ms | 0 - 362 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.tflite) |
| FFNet-78S-LowRes | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 13.579 ms | 0 - 30 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.dlc) |
| FFNet-78S-LowRes | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 18.829 ms | 1 - 62 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.tflite) |
| FFNet-78S-LowRes | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 19.654 ms | 5 - 33 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.dlc) |
| FFNet-78S-LowRes | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 12.604 ms | 1 - 355 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.tflite) |
| FFNet-78S-LowRes | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 13.657 ms | 3 - 34 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.dlc) |
| FFNet-78S-LowRes | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 17.121 ms | 1 - 62 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.tflite) |
| FFNet-78S-LowRes | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 17.931 ms | 3 - 30 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.dlc) |
| FFNet-78S-LowRes | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 12.593 ms | 1 - 18 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.tflite) |
| FFNet-78S-LowRes | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 13.631 ms | 6 - 22 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.dlc) |
| FFNet-78S-LowRes | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 9.674 ms | 0 - 116 MB | NPU | [FFNet-78S-LowRes.onnx.zip](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.onnx.zip) |
| FFNet-78S-LowRes | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 8.578 ms | 1 - 118 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.tflite) |
| FFNet-78S-LowRes | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 8.981 ms | 6 - 43 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.dlc) |
| FFNet-78S-LowRes | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 7.133 ms | 7 - 52 MB | NPU | [FFNet-78S-LowRes.onnx.zip](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.onnx.zip) |
| FFNet-78S-LowRes | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 7.25 ms | 1 - 66 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.tflite) |
| FFNet-78S-LowRes | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 9.157 ms | 6 - 39 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.dlc) |
| FFNet-78S-LowRes | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 7.142 ms | 7 - 47 MB | NPU | [FFNet-78S-LowRes.onnx.zip](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.onnx.zip) |
| FFNet-78S-LowRes | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 14.332 ms | 109 - 109 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.dlc) |
| FFNet-78S-LowRes | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 9.278 ms | 47 - 47 MB | NPU | [FFNet-78S-LowRes.onnx.zip](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes.onnx.zip) |
| FFNet-78S-LowRes | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 7.796 ms | 0 - 42 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.tflite) |
| FFNet-78S-LowRes | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 10.54 ms | 2 - 44 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.dlc) |
| FFNet-78S-LowRes | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 3.414 ms | 0 - 78 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.tflite) |
| FFNet-78S-LowRes | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 7.185 ms | 2 - 78 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.dlc) |
| FFNet-78S-LowRes | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 3.014 ms | 0 - 9 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.tflite) |
| FFNet-78S-LowRes | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 4.663 ms | 1 - 176 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.dlc) |
| FFNet-78S-LowRes | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 3.538 ms | 0 - 41 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.tflite) |
| FFNet-78S-LowRes | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 5.125 ms | 2 - 44 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.dlc) |
| FFNet-78S-LowRes | w8a8 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | TFLITE | 10.505 ms | 0 - 64 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.tflite) |
| FFNet-78S-LowRes | w8a8 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | QNN_DLC | 22.713 ms | 2 - 66 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.dlc) |
| FFNet-78S-LowRes | w8a8 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | TFLITE | 73.263 ms | 12 - 24 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.tflite) |
| FFNet-78S-LowRes | w8a8 | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 7.796 ms | 0 - 42 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.tflite) |
| FFNet-78S-LowRes | w8a8 | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 10.54 ms | 2 - 44 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.dlc) |
| FFNet-78S-LowRes | w8a8 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 3.047 ms | 0 - 199 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.tflite) |
| FFNet-78S-LowRes | w8a8 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 4.646 ms | 2 - 12 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.dlc) |
| FFNet-78S-LowRes | w8a8 | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 4.737 ms | 0 - 47 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.tflite) |
| FFNet-78S-LowRes | w8a8 | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 6.549 ms | 2 - 50 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.dlc) |
| FFNet-78S-LowRes | w8a8 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 3.029 ms | 0 - 195 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.tflite) |
| FFNet-78S-LowRes | w8a8 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 4.655 ms | 2 - 12 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.dlc) |
| FFNet-78S-LowRes | w8a8 | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 3.538 ms | 0 - 41 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.tflite) |
| FFNet-78S-LowRes | w8a8 | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 5.125 ms | 2 - 44 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.dlc) |
| FFNet-78S-LowRes | w8a8 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 3.004 ms | 0 - 197 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.tflite) |
| FFNet-78S-LowRes | w8a8 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 4.651 ms | 0 - 146 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.dlc) |
| FFNet-78S-LowRes | w8a8 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 91.158 ms | 106 - 380 MB | NPU | [FFNet-78S-LowRes.onnx.zip](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.onnx.zip) |
| FFNet-78S-LowRes | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 2.186 ms | 0 - 73 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.tflite) |
| FFNet-78S-LowRes | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 3.306 ms | 2 - 73 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.dlc) |
| FFNet-78S-LowRes | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 77.596 ms | 147 - 3462 MB | NPU | [FFNet-78S-LowRes.onnx.zip](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.onnx.zip) |
| FFNet-78S-LowRes | w8a8 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 2.154 ms | 0 - 46 MB | NPU | [FFNet-78S-LowRes.tflite](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.tflite) |
| FFNet-78S-LowRes | w8a8 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 2.239 ms | 2 - 46 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.dlc) |
| FFNet-78S-LowRes | w8a8 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 75.622 ms | 146 - 2958 MB | NPU | [FFNet-78S-LowRes.onnx.zip](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.onnx.zip) |
| FFNet-78S-LowRes | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 5.069 ms | 175 - 175 MB | NPU | [FFNet-78S-LowRes.dlc](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.dlc) |
| FFNet-78S-LowRes | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 108.467 ms | 116 - 116 MB | NPU | [FFNet-78S-LowRes.onnx.zip](https://huggingface.co/qualcomm/FFNet-78S-LowRes/blob/main/FFNet-78S-LowRes_w8a8.onnx.zip) |
## Installation
Install the package via pip:
```bash
pip install "qai-hub-models[ffnet-78s-lowres]"
```
## Configure Qualcomm® AI Hub to run this model on a cloud-hosted device
Sign-in to [Qualcomm® AI Hub](https://app.aihub.qualcomm.com/) with your
Qualcomm® ID. Once signed in navigate to `Account -> Settings -> API Token`.
With this API token, you can configure your client to run models on the cloud
hosted devices.
```bash
qai-hub configure --api_token API_TOKEN
```
Navigate to [docs](https://app.aihub.qualcomm.com/docs/) for more information.
## Demo off target
The package contains a simple end-to-end demo that downloads pre-trained
weights and runs this model on a sample input.
```bash
python -m qai_hub_models.models.ffnet_78s_lowres.demo
```
The above demo runs a reference implementation of pre-processing, model
inference, and post processing.
**NOTE**: If you want running in a Jupyter Notebook or Google Colab like
environment, please add the following to your cell (instead of the above).
```
%run -m qai_hub_models.models.ffnet_78s_lowres.demo
```
### Run model on a cloud-hosted device
In addition to the demo, you can also run the model on a cloud-hosted Qualcomm®
device. This script does the following:
* Performance check on-device on a cloud-hosted device
* Downloads compiled assets that can be deployed on-device for Android.
* Accuracy check between PyTorch and on-device outputs.
```bash
python -m qai_hub_models.models.ffnet_78s_lowres.export
```
## How does this work?
This [export script](https://aihub.qualcomm.com/models/ffnet_78s_lowres/qai_hub_models/models/FFNet-78S-LowRes/export.py)
leverages [Qualcomm® AI Hub](https://aihub.qualcomm.com/) to optimize, validate, and deploy this model
on-device. Lets go through each step below in detail:
Step 1: **Compile model for on-device deployment**
To compile a PyTorch model for on-device deployment, we first trace the model
in memory using the `jit.trace` and then call the `submit_compile_job` API.
```python
import torch
import qai_hub as hub
from qai_hub_models.models.ffnet_78s_lowres import Model
# Load the model
torch_model = Model.from_pretrained()
# Device
device = hub.Device("Samsung Galaxy S24")
# Trace model
input_shape = torch_model.get_input_spec()
sample_inputs = torch_model.sample_inputs()
pt_model = torch.jit.trace(torch_model, [torch.tensor(data[0]) for _, data in sample_inputs.items()])
# Compile model on a specific device
compile_job = hub.submit_compile_job(
model=pt_model,
device=device,
input_specs=torch_model.get_input_spec(),
)
# Get target model to run on-device
target_model = compile_job.get_target_model()
```
Step 2: **Performance profiling on cloud-hosted device**
After compiling models from step 1. Models can be profiled model on-device using the
`target_model`. Note that this scripts runs the model on a device automatically
provisioned in the cloud. Once the job is submitted, you can navigate to a
provided job URL to view a variety of on-device performance metrics.
```python
profile_job = hub.submit_profile_job(
model=target_model,
device=device,
)
```
Step 3: **Verify on-device accuracy**
To verify the accuracy of the model on-device, you can run on-device inference
on sample input data on the same cloud hosted device.
```python
input_data = torch_model.sample_inputs()
inference_job = hub.submit_inference_job(
model=target_model,
device=device,
inputs=input_data,
)
on_device_output = inference_job.download_output_data()
```
With the output of the model, you can compute like PSNR, relative errors or
spot check the output with expected output.
**Note**: This on-device profiling and inference requires access to Qualcomm®
AI Hub. [Sign up for access](https://myaccount.qualcomm.com/signup).
## Run demo on a cloud-hosted device
You can also run the demo on-device.
```bash
python -m qai_hub_models.models.ffnet_78s_lowres.demo --eval-mode on-device
```
**NOTE**: If you want running in a Jupyter Notebook or Google Colab like
environment, please add the following to your cell (instead of the above).
```
%run -m qai_hub_models.models.ffnet_78s_lowres.demo -- --eval-mode on-device
```
## Deploying compiled model to Android
The models can be deployed using multiple runtimes:
- TensorFlow Lite (`.tflite` export): [This
tutorial](https://www.tensorflow.org/lite/android/quickstart) provides a
guide to deploy the .tflite model in an Android application.
- QNN (`.so` export ): This [sample
app](https://docs.qualcomm.com/bundle/publicresource/topics/80-63442-50/sample_app.html)
provides instructions on how to use the `.so` shared library in an Android application.
## View on Qualcomm® AI Hub
Get more details on FFNet-78S-LowRes's performance across various devices [here](https://aihub.qualcomm.com/models/ffnet_78s_lowres).
Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
## License
* The license for the original implementation of FFNet-78S-LowRes can be found
[here](https://github.com/Qualcomm-AI-research/FFNet/blob/master/LICENSE).
* The license for the compiled assets for on-device deployment can be found [here](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/Qualcomm+AI+Hub+Proprietary+License.pdf)
## References
* [Simple and Efficient Architectures for Semantic Segmentation](https://arxiv.org/abs/2206.08236)
* [Source Model Implementation](https://github.com/Qualcomm-AI-research/FFNet)
## Community
* Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
* For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).
|
qualcomm/FFNet-54S
|
qualcomm
| 2025-08-29T23:21:59Z | 35 | 1 |
pytorch
|
[
"pytorch",
"tflite",
"android",
"image-segmentation",
"arxiv:2206.08236",
"license:other",
"region:us"
] |
image-segmentation
| 2024-02-25T23:04:07Z |
---
library_name: pytorch
license: other
tags:
- android
pipeline_tag: image-segmentation
---
# FFNet-54S: Optimized for Mobile Deployment
## Semantic segmentation for automotive street scenes
FFNet-54S is a "fuss-free network" that segments street scene images with per-pixel classes like road, sidewalk, and pedestrian. Trained on the Cityscapes dataset.
This model is an implementation of FFNet-54S found [here](https://github.com/Qualcomm-AI-research/FFNet).
This repository provides scripts to run FFNet-54S on Qualcomm® devices.
More details on model performance across various devices, can be found
[here](https://aihub.qualcomm.com/models/ffnet_54s).
### Model Details
- **Model Type:** Model_use_case.semantic_segmentation
- **Model Stats:**
- Model checkpoint: ffnet54S_dBBB_cityscapes_state_dict_quarts
- Input resolution: 2048x1024
- Number of output classes: 19
- Number of parameters: 18.0M
- Model size (float): 68.8 MB
- Model size (w8a8): 17.5 MB
| Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
|---|---|---|---|---|---|---|---|---|
| FFNet-54S | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 169.646 ms | 2 - 66 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
| FFNet-54S | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 152.465 ms | 24 - 85 MB | NPU | [FFNet-54S.dlc](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.dlc) |
| FFNet-54S | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 60.571 ms | 2 - 100 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
| FFNet-54S | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 69.909 ms | 25 - 84 MB | NPU | [FFNet-54S.dlc](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.dlc) |
| FFNet-54S | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 47.062 ms | 2 - 24 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
| FFNet-54S | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 36.937 ms | 25 - 49 MB | NPU | [FFNet-54S.dlc](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.dlc) |
| FFNet-54S | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 62.95 ms | 0 - 64 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
| FFNet-54S | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 51.947 ms | 24 - 85 MB | NPU | [FFNet-54S.dlc](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.dlc) |
| FFNet-54S | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 169.646 ms | 2 - 66 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
| FFNet-54S | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 152.465 ms | 24 - 85 MB | NPU | [FFNet-54S.dlc](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.dlc) |
| FFNet-54S | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 46.957 ms | 2 - 17 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
| FFNet-54S | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 36.483 ms | 24 - 53 MB | NPU | [FFNet-54S.dlc](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.dlc) |
| FFNet-54S | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 68.961 ms | 2 - 67 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
| FFNet-54S | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 56.131 ms | 0 - 61 MB | NPU | [FFNet-54S.dlc](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.dlc) |
| FFNet-54S | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 47.206 ms | 2 - 25 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
| FFNet-54S | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 36.921 ms | 24 - 47 MB | NPU | [FFNet-54S.dlc](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.dlc) |
| FFNet-54S | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 62.95 ms | 0 - 64 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
| FFNet-54S | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 51.947 ms | 24 - 85 MB | NPU | [FFNet-54S.dlc](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.dlc) |
| FFNet-54S | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 47.581 ms | 2 - 18 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
| FFNet-54S | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 36.902 ms | 24 - 47 MB | NPU | [FFNet-54S.dlc](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.dlc) |
| FFNet-54S | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 31.817 ms | 24 - 49 MB | NPU | [FFNet-54S.onnx.zip](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.onnx.zip) |
| FFNet-54S | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 31.054 ms | 0 - 96 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
| FFNet-54S | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 25.432 ms | 23 - 91 MB | NPU | [FFNet-54S.dlc](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.dlc) |
| FFNet-54S | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 22.314 ms | 30 - 114 MB | NPU | [FFNet-54S.onnx.zip](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.onnx.zip) |
| FFNet-54S | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 29.931 ms | 2 - 69 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.tflite) |
| FFNet-54S | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 23.314 ms | 13 - 85 MB | NPU | [FFNet-54S.dlc](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.dlc) |
| FFNet-54S | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 19.746 ms | 30 - 86 MB | NPU | [FFNet-54S.onnx.zip](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.onnx.zip) |
| FFNet-54S | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 37.949 ms | 24 - 24 MB | NPU | [FFNet-54S.dlc](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.dlc) |
| FFNet-54S | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 33.218 ms | 24 - 24 MB | NPU | [FFNet-54S.onnx.zip](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S.onnx.zip) |
| FFNet-54S | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 25.325 ms | 1 - 44 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S_w8a8.tflite) |
| FFNet-54S | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 10.677 ms | 1 - 69 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S_w8a8.tflite) |
| FFNet-54S | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 9.245 ms | 1 - 16 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S_w8a8.tflite) |
| FFNet-54S | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 10.014 ms | 1 - 44 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S_w8a8.tflite) |
| FFNet-54S | w8a8 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | TFLITE | 61.723 ms | 0 - 142 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S_w8a8.tflite) |
| FFNet-54S | w8a8 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | TFLITE | 318.739 ms | 1 - 4 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S_w8a8.tflite) |
| FFNet-54S | w8a8 | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 25.325 ms | 1 - 44 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S_w8a8.tflite) |
| FFNet-54S | w8a8 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 9.237 ms | 1 - 10 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S_w8a8.tflite) |
| FFNet-54S | w8a8 | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 14.727 ms | 1 - 51 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S_w8a8.tflite) |
| FFNet-54S | w8a8 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 9.248 ms | 1 - 13 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S_w8a8.tflite) |
| FFNet-54S | w8a8 | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 10.014 ms | 1 - 44 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S_w8a8.tflite) |
| FFNet-54S | w8a8 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 9.248 ms | 1 - 12 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S_w8a8.tflite) |
| FFNet-54S | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 6.705 ms | 1 - 63 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S_w8a8.tflite) |
| FFNet-54S | w8a8 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 6.509 ms | 1 - 49 MB | NPU | [FFNet-54S.tflite](https://huggingface.co/qualcomm/FFNet-54S/blob/main/FFNet-54S_w8a8.tflite) |
## Installation
Install the package via pip:
```bash
pip install "qai-hub-models[ffnet-54s]"
```
## Configure Qualcomm® AI Hub to run this model on a cloud-hosted device
Sign-in to [Qualcomm® AI Hub](https://app.aihub.qualcomm.com/) with your
Qualcomm® ID. Once signed in navigate to `Account -> Settings -> API Token`.
With this API token, you can configure your client to run models on the cloud
hosted devices.
```bash
qai-hub configure --api_token API_TOKEN
```
Navigate to [docs](https://app.aihub.qualcomm.com/docs/) for more information.
## Demo off target
The package contains a simple end-to-end demo that downloads pre-trained
weights and runs this model on a sample input.
```bash
python -m qai_hub_models.models.ffnet_54s.demo
```
The above demo runs a reference implementation of pre-processing, model
inference, and post processing.
**NOTE**: If you want running in a Jupyter Notebook or Google Colab like
environment, please add the following to your cell (instead of the above).
```
%run -m qai_hub_models.models.ffnet_54s.demo
```
### Run model on a cloud-hosted device
In addition to the demo, you can also run the model on a cloud-hosted Qualcomm®
device. This script does the following:
* Performance check on-device on a cloud-hosted device
* Downloads compiled assets that can be deployed on-device for Android.
* Accuracy check between PyTorch and on-device outputs.
```bash
python -m qai_hub_models.models.ffnet_54s.export
```
## How does this work?
This [export script](https://aihub.qualcomm.com/models/ffnet_54s/qai_hub_models/models/FFNet-54S/export.py)
leverages [Qualcomm® AI Hub](https://aihub.qualcomm.com/) to optimize, validate, and deploy this model
on-device. Lets go through each step below in detail:
Step 1: **Compile model for on-device deployment**
To compile a PyTorch model for on-device deployment, we first trace the model
in memory using the `jit.trace` and then call the `submit_compile_job` API.
```python
import torch
import qai_hub as hub
from qai_hub_models.models.ffnet_54s import Model
# Load the model
torch_model = Model.from_pretrained()
# Device
device = hub.Device("Samsung Galaxy S24")
# Trace model
input_shape = torch_model.get_input_spec()
sample_inputs = torch_model.sample_inputs()
pt_model = torch.jit.trace(torch_model, [torch.tensor(data[0]) for _, data in sample_inputs.items()])
# Compile model on a specific device
compile_job = hub.submit_compile_job(
model=pt_model,
device=device,
input_specs=torch_model.get_input_spec(),
)
# Get target model to run on-device
target_model = compile_job.get_target_model()
```
Step 2: **Performance profiling on cloud-hosted device**
After compiling models from step 1. Models can be profiled model on-device using the
`target_model`. Note that this scripts runs the model on a device automatically
provisioned in the cloud. Once the job is submitted, you can navigate to a
provided job URL to view a variety of on-device performance metrics.
```python
profile_job = hub.submit_profile_job(
model=target_model,
device=device,
)
```
Step 3: **Verify on-device accuracy**
To verify the accuracy of the model on-device, you can run on-device inference
on sample input data on the same cloud hosted device.
```python
input_data = torch_model.sample_inputs()
inference_job = hub.submit_inference_job(
model=target_model,
device=device,
inputs=input_data,
)
on_device_output = inference_job.download_output_data()
```
With the output of the model, you can compute like PSNR, relative errors or
spot check the output with expected output.
**Note**: This on-device profiling and inference requires access to Qualcomm®
AI Hub. [Sign up for access](https://myaccount.qualcomm.com/signup).
## Run demo on a cloud-hosted device
You can also run the demo on-device.
```bash
python -m qai_hub_models.models.ffnet_54s.demo --eval-mode on-device
```
**NOTE**: If you want running in a Jupyter Notebook or Google Colab like
environment, please add the following to your cell (instead of the above).
```
%run -m qai_hub_models.models.ffnet_54s.demo -- --eval-mode on-device
```
## Deploying compiled model to Android
The models can be deployed using multiple runtimes:
- TensorFlow Lite (`.tflite` export): [This
tutorial](https://www.tensorflow.org/lite/android/quickstart) provides a
guide to deploy the .tflite model in an Android application.
- QNN (`.so` export ): This [sample
app](https://docs.qualcomm.com/bundle/publicresource/topics/80-63442-50/sample_app.html)
provides instructions on how to use the `.so` shared library in an Android application.
## View on Qualcomm® AI Hub
Get more details on FFNet-54S's performance across various devices [here](https://aihub.qualcomm.com/models/ffnet_54s).
Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
## License
* The license for the original implementation of FFNet-54S can be found
[here](https://github.com/Qualcomm-AI-research/FFNet/blob/master/LICENSE).
* The license for the compiled assets for on-device deployment can be found [here](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/Qualcomm+AI+Hub+Proprietary+License.pdf)
## References
* [Simple and Efficient Architectures for Semantic Segmentation](https://arxiv.org/abs/2206.08236)
* [Source Model Implementation](https://github.com/Qualcomm-AI-research/FFNet)
## Community
* Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
* For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).
|
ultratopaz/545631
|
ultratopaz
| 2025-08-29T23:20:06Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T23:19:59Z |
[View on Civ Archive](https://civarchive.com/models/490763?modelVersionId=630926)
|
ultratopaz/462057
|
ultratopaz
| 2025-08-29T23:18:55Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T23:18:52Z |
[View on Civ Archive](https://civarchive.com/models/490863?modelVersionId=545827)
|
amethyst9/536349
|
amethyst9
| 2025-08-29T23:17:47Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T23:17:41Z |
[View on Civ Archive](https://civarchive.com/models/526371?modelVersionId=621339)
|
qualcomm/EfficientViT-b2-cls
|
qualcomm
| 2025-08-29T23:17:29Z | 28 | 0 |
pytorch
|
[
"pytorch",
"tflite",
"backbone",
"real_time",
"android",
"image-classification",
"arxiv:2205.14756",
"license:other",
"region:us"
] |
image-classification
| 2024-11-13T02:09:12Z |
---
library_name: pytorch
license: other
tags:
- backbone
- real_time
- android
pipeline_tag: image-classification
---
# EfficientViT-b2-cls: Optimized for Mobile Deployment
## Imagenet classifier and general purpose backbone
EfficientViT is a machine learning model that can classify images from the Imagenet dataset. It can also be used as a backbone in building more complex models for specific use cases.
This model is an implementation of EfficientViT-b2-cls found [here](https://github.com/CVHub520/efficientvit).
This repository provides scripts to run EfficientViT-b2-cls on Qualcomm® devices.
More details on model performance across various devices, can be found
[here](https://aihub.qualcomm.com/models/efficientvit_b2_cls).
### Model Details
- **Model Type:** Model_use_case.image_classification
- **Model Stats:**
- Model checkpoint: Imagenet
- Input resolution: 224x224
- Number of parameters: 24.3M
- Model size (float): 92.9 MB
| Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
|---|---|---|---|---|---|---|---|---|
| EfficientViT-b2-cls | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 12.155 ms | 0 - 102 MB | NPU | [EfficientViT-b2-cls.tflite](https://huggingface.co/qualcomm/EfficientViT-b2-cls/blob/main/EfficientViT-b2-cls.tflite) |
| EfficientViT-b2-cls | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 12.855 ms | 1 - 62 MB | NPU | [EfficientViT-b2-cls.dlc](https://huggingface.co/qualcomm/EfficientViT-b2-cls/blob/main/EfficientViT-b2-cls.dlc) |
| EfficientViT-b2-cls | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 6.399 ms | 0 - 113 MB | NPU | [EfficientViT-b2-cls.tflite](https://huggingface.co/qualcomm/EfficientViT-b2-cls/blob/main/EfficientViT-b2-cls.tflite) |
| EfficientViT-b2-cls | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 6.872 ms | 0 - 71 MB | NPU | [EfficientViT-b2-cls.dlc](https://huggingface.co/qualcomm/EfficientViT-b2-cls/blob/main/EfficientViT-b2-cls.dlc) |
| EfficientViT-b2-cls | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 4.981 ms | 0 - 345 MB | NPU | [EfficientViT-b2-cls.tflite](https://huggingface.co/qualcomm/EfficientViT-b2-cls/blob/main/EfficientViT-b2-cls.tflite) |
| EfficientViT-b2-cls | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 5.364 ms | 0 - 16 MB | NPU | [EfficientViT-b2-cls.dlc](https://huggingface.co/qualcomm/EfficientViT-b2-cls/blob/main/EfficientViT-b2-cls.dlc) |
| EfficientViT-b2-cls | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 6.096 ms | 0 - 103 MB | NPU | [EfficientViT-b2-cls.tflite](https://huggingface.co/qualcomm/EfficientViT-b2-cls/blob/main/EfficientViT-b2-cls.tflite) |
| EfficientViT-b2-cls | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 6.607 ms | 1 - 62 MB | NPU | [EfficientViT-b2-cls.dlc](https://huggingface.co/qualcomm/EfficientViT-b2-cls/blob/main/EfficientViT-b2-cls.dlc) |
| EfficientViT-b2-cls | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 5.025 ms | 0 - 343 MB | NPU | [EfficientViT-b2-cls.tflite](https://huggingface.co/qualcomm/EfficientViT-b2-cls/blob/main/EfficientViT-b2-cls.tflite) |
| EfficientViT-b2-cls | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 5.367 ms | 0 - 18 MB | NPU | [EfficientViT-b2-cls.dlc](https://huggingface.co/qualcomm/EfficientViT-b2-cls/blob/main/EfficientViT-b2-cls.dlc) |
| EfficientViT-b2-cls | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 5.399 ms | 0 - 125 MB | NPU | [EfficientViT-b2-cls.onnx.zip](https://huggingface.co/qualcomm/EfficientViT-b2-cls/blob/main/EfficientViT-b2-cls.onnx.zip) |
| EfficientViT-b2-cls | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 3.425 ms | 0 - 116 MB | NPU | [EfficientViT-b2-cls.tflite](https://huggingface.co/qualcomm/EfficientViT-b2-cls/blob/main/EfficientViT-b2-cls.tflite) |
| EfficientViT-b2-cls | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 3.752 ms | 1 - 77 MB | NPU | [EfficientViT-b2-cls.dlc](https://huggingface.co/qualcomm/EfficientViT-b2-cls/blob/main/EfficientViT-b2-cls.dlc) |
| EfficientViT-b2-cls | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 3.709 ms | 0 - 73 MB | NPU | [EfficientViT-b2-cls.onnx.zip](https://huggingface.co/qualcomm/EfficientViT-b2-cls/blob/main/EfficientViT-b2-cls.onnx.zip) |
| EfficientViT-b2-cls | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 3.332 ms | 0 - 104 MB | NPU | [EfficientViT-b2-cls.tflite](https://huggingface.co/qualcomm/EfficientViT-b2-cls/blob/main/EfficientViT-b2-cls.tflite) |
| EfficientViT-b2-cls | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 3.289 ms | 1 - 66 MB | NPU | [EfficientViT-b2-cls.dlc](https://huggingface.co/qualcomm/EfficientViT-b2-cls/blob/main/EfficientViT-b2-cls.dlc) |
| EfficientViT-b2-cls | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 3.31 ms | 1 - 63 MB | NPU | [EfficientViT-b2-cls.onnx.zip](https://huggingface.co/qualcomm/EfficientViT-b2-cls/blob/main/EfficientViT-b2-cls.onnx.zip) |
| EfficientViT-b2-cls | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 6.092 ms | 274 - 274 MB | NPU | [EfficientViT-b2-cls.dlc](https://huggingface.co/qualcomm/EfficientViT-b2-cls/blob/main/EfficientViT-b2-cls.dlc) |
| EfficientViT-b2-cls | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 6.139 ms | 49 - 49 MB | NPU | [EfficientViT-b2-cls.onnx.zip](https://huggingface.co/qualcomm/EfficientViT-b2-cls/blob/main/EfficientViT-b2-cls.onnx.zip) |
## Installation
Install the package via pip:
```bash
pip install "qai-hub-models[efficientvit-b2-cls]"
```
## Configure Qualcomm® AI Hub to run this model on a cloud-hosted device
Sign-in to [Qualcomm® AI Hub](https://app.aihub.qualcomm.com/) with your
Qualcomm® ID. Once signed in navigate to `Account -> Settings -> API Token`.
With this API token, you can configure your client to run models on the cloud
hosted devices.
```bash
qai-hub configure --api_token API_TOKEN
```
Navigate to [docs](https://app.aihub.qualcomm.com/docs/) for more information.
## Demo off target
The package contains a simple end-to-end demo that downloads pre-trained
weights and runs this model on a sample input.
```bash
python -m qai_hub_models.models.efficientvit_b2_cls.demo
```
The above demo runs a reference implementation of pre-processing, model
inference, and post processing.
**NOTE**: If you want running in a Jupyter Notebook or Google Colab like
environment, please add the following to your cell (instead of the above).
```
%run -m qai_hub_models.models.efficientvit_b2_cls.demo
```
### Run model on a cloud-hosted device
In addition to the demo, you can also run the model on a cloud-hosted Qualcomm®
device. This script does the following:
* Performance check on-device on a cloud-hosted device
* Downloads compiled assets that can be deployed on-device for Android.
* Accuracy check between PyTorch and on-device outputs.
```bash
python -m qai_hub_models.models.efficientvit_b2_cls.export
```
## How does this work?
This [export script](https://aihub.qualcomm.com/models/efficientvit_b2_cls/qai_hub_models/models/EfficientViT-b2-cls/export.py)
leverages [Qualcomm® AI Hub](https://aihub.qualcomm.com/) to optimize, validate, and deploy this model
on-device. Lets go through each step below in detail:
Step 1: **Compile model for on-device deployment**
To compile a PyTorch model for on-device deployment, we first trace the model
in memory using the `jit.trace` and then call the `submit_compile_job` API.
```python
import torch
import qai_hub as hub
from qai_hub_models.models.efficientvit_b2_cls import Model
# Load the model
torch_model = Model.from_pretrained()
# Device
device = hub.Device("Samsung Galaxy S24")
# Trace model
input_shape = torch_model.get_input_spec()
sample_inputs = torch_model.sample_inputs()
pt_model = torch.jit.trace(torch_model, [torch.tensor(data[0]) for _, data in sample_inputs.items()])
# Compile model on a specific device
compile_job = hub.submit_compile_job(
model=pt_model,
device=device,
input_specs=torch_model.get_input_spec(),
)
# Get target model to run on-device
target_model = compile_job.get_target_model()
```
Step 2: **Performance profiling on cloud-hosted device**
After compiling models from step 1. Models can be profiled model on-device using the
`target_model`. Note that this scripts runs the model on a device automatically
provisioned in the cloud. Once the job is submitted, you can navigate to a
provided job URL to view a variety of on-device performance metrics.
```python
profile_job = hub.submit_profile_job(
model=target_model,
device=device,
)
```
Step 3: **Verify on-device accuracy**
To verify the accuracy of the model on-device, you can run on-device inference
on sample input data on the same cloud hosted device.
```python
input_data = torch_model.sample_inputs()
inference_job = hub.submit_inference_job(
model=target_model,
device=device,
inputs=input_data,
)
on_device_output = inference_job.download_output_data()
```
With the output of the model, you can compute like PSNR, relative errors or
spot check the output with expected output.
**Note**: This on-device profiling and inference requires access to Qualcomm®
AI Hub. [Sign up for access](https://myaccount.qualcomm.com/signup).
## Run demo on a cloud-hosted device
You can also run the demo on-device.
```bash
python -m qai_hub_models.models.efficientvit_b2_cls.demo --eval-mode on-device
```
**NOTE**: If you want running in a Jupyter Notebook or Google Colab like
environment, please add the following to your cell (instead of the above).
```
%run -m qai_hub_models.models.efficientvit_b2_cls.demo -- --eval-mode on-device
```
## Deploying compiled model to Android
The models can be deployed using multiple runtimes:
- TensorFlow Lite (`.tflite` export): [This
tutorial](https://www.tensorflow.org/lite/android/quickstart) provides a
guide to deploy the .tflite model in an Android application.
- QNN (`.so` export ): This [sample
app](https://docs.qualcomm.com/bundle/publicresource/topics/80-63442-50/sample_app.html)
provides instructions on how to use the `.so` shared library in an Android application.
## View on Qualcomm® AI Hub
Get more details on EfficientViT-b2-cls's performance across various devices [here](https://aihub.qualcomm.com/models/efficientvit_b2_cls).
Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
## License
* The license for the original implementation of EfficientViT-b2-cls can be found
[here](https://github.com/CVHub520/efficientvit/blob/main/LICENSE).
* The license for the compiled assets for on-device deployment can be found [here](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/Qualcomm+AI+Hub+Proprietary+License.pdf)
## References
* [EfficientViT: Multi-Scale Linear Attention for High-Resolution Dense Prediction](https://arxiv.org/abs/2205.14756)
* [Source Model Implementation](https://github.com/CVHub520/efficientvit)
## Community
* Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
* For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).
|
crystalline7/633421
|
crystalline7
| 2025-08-29T23:16:48Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T23:16:42Z |
[View on Civ Archive](https://civarchive.com/models/642444?modelVersionId=718584)
|
qualcomm/EfficientFormer
|
qualcomm
| 2025-08-29T23:15:45Z | 0 | 0 |
pytorch
|
[
"pytorch",
"tflite",
"android",
"image-classification",
"arxiv:2212.08059",
"license:other",
"region:us"
] |
image-classification
| 2025-08-29T21:48:16Z |
---
library_name: pytorch
license: other
tags:
- android
pipeline_tag: image-classification
---
# EfficientFormer: Optimized for Mobile Deployment
## Imagenet classifier and general purpose backbone
EfficientFormer is a vision transformer model that can classify images from the Imagenet dataset.
This model is an implementation of EfficientFormer found [here](https://github.com/snap-research/EfficientFormer).
This repository provides scripts to run EfficientFormer on Qualcomm® devices.
More details on model performance across various devices, can be found
[here](https://aihub.qualcomm.com/models/efficientformer).
### Model Details
- **Model Type:** Model_use_case.image_classification
- **Model Stats:**
- Model checkpoint: efficientformer_l1_300d
- Input resolution: 224x224
- Number of parameters: 12.3M
- Model size (float): 46.9 MB
- Model size (w8a16): 12.2 MB
| Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
|---|---|---|---|---|---|---|---|---|
| EfficientFormer | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 4.93 ms | 0 - 48 MB | NPU | [EfficientFormer.tflite](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer.tflite) |
| EfficientFormer | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 4.994 ms | 1 - 34 MB | NPU | [EfficientFormer.dlc](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer.dlc) |
| EfficientFormer | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 3.825 ms | 0 - 55 MB | NPU | [EfficientFormer.tflite](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer.tflite) |
| EfficientFormer | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 5.702 ms | 0 - 42 MB | NPU | [EfficientFormer.dlc](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer.dlc) |
| EfficientFormer | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 1.519 ms | 0 - 157 MB | NPU | [EfficientFormer.tflite](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer.tflite) |
| EfficientFormer | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 1.617 ms | 1 - 11 MB | NPU | [EfficientFormer.dlc](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer.dlc) |
| EfficientFormer | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 2.096 ms | 0 - 48 MB | NPU | [EfficientFormer.tflite](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer.tflite) |
| EfficientFormer | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 2.298 ms | 1 - 34 MB | NPU | [EfficientFormer.dlc](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer.dlc) |
| EfficientFormer | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 1.523 ms | 0 - 156 MB | NPU | [EfficientFormer.tflite](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer.tflite) |
| EfficientFormer | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 1.651 ms | 1 - 10 MB | NPU | [EfficientFormer.dlc](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer.dlc) |
| EfficientFormer | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 5.967 ms | 0 - 39 MB | NPU | [EfficientFormer.onnx.zip](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer.onnx.zip) |
| EfficientFormer | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 1.057 ms | 0 - 59 MB | NPU | [EfficientFormer.tflite](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer.tflite) |
| EfficientFormer | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 1.125 ms | 1 - 44 MB | NPU | [EfficientFormer.dlc](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer.dlc) |
| EfficientFormer | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 4.096 ms | 0 - 45 MB | NPU | [EfficientFormer.onnx.zip](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer.onnx.zip) |
| EfficientFormer | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 0.941 ms | 0 - 52 MB | NPU | [EfficientFormer.tflite](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer.tflite) |
| EfficientFormer | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 0.825 ms | 1 - 38 MB | NPU | [EfficientFormer.dlc](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer.dlc) |
| EfficientFormer | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 3.739 ms | 1 - 42 MB | NPU | [EfficientFormer.onnx.zip](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer.onnx.zip) |
| EfficientFormer | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 1.904 ms | 102 - 102 MB | NPU | [EfficientFormer.dlc](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer.dlc) |
| EfficientFormer | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 6.192 ms | 24 - 24 MB | NPU | [EfficientFormer.onnx.zip](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer.onnx.zip) |
| EfficientFormer | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 1.695 ms | 0 - 27 MB | NPU | [EfficientFormer.tflite](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer_w8a8.tflite) |
| EfficientFormer | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 0.852 ms | 0 - 46 MB | NPU | [EfficientFormer.tflite](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer_w8a8.tflite) |
| EfficientFormer | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 0.744 ms | 0 - 59 MB | NPU | [EfficientFormer.tflite](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer_w8a8.tflite) |
| EfficientFormer | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 0.997 ms | 0 - 27 MB | NPU | [EfficientFormer.tflite](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer_w8a8.tflite) |
| EfficientFormer | w8a8 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | TFLITE | 4.185 ms | 0 - 39 MB | NPU | [EfficientFormer.tflite](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer_w8a8.tflite) |
| EfficientFormer | w8a8 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | TFLITE | 44.509 ms | 1 - 80 MB | GPU | [EfficientFormer.tflite](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer_w8a8.tflite) |
| EfficientFormer | w8a8 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 0.743 ms | 0 - 59 MB | NPU | [EfficientFormer.tflite](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer_w8a8.tflite) |
| EfficientFormer | w8a8 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 9.494 ms | 25 - 49 MB | NPU | [EfficientFormer.onnx.zip](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer_w8a8.onnx.zip) |
| EfficientFormer | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 0.53 ms | 0 - 45 MB | NPU | [EfficientFormer.tflite](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer_w8a8.tflite) |
| EfficientFormer | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 6.246 ms | 23 - 82 MB | NPU | [EfficientFormer.onnx.zip](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer_w8a8.onnx.zip) |
| EfficientFormer | w8a8 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 0.41 ms | 1 - 34 MB | NPU | [EfficientFormer.tflite](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer_w8a8.tflite) |
| EfficientFormer | w8a8 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 5.503 ms | 28 - 71 MB | NPU | [EfficientFormer.onnx.zip](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer_w8a8.onnx.zip) |
| EfficientFormer | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 8.31 ms | 25 - 25 MB | NPU | [EfficientFormer.onnx.zip](https://huggingface.co/qualcomm/EfficientFormer/blob/main/EfficientFormer_w8a8.onnx.zip) |
## Installation
Install the package via pip:
```bash
pip install "qai-hub-models[efficientformer]"
```
## Configure Qualcomm® AI Hub to run this model on a cloud-hosted device
Sign-in to [Qualcomm® AI Hub](https://app.aihub.qualcomm.com/) with your
Qualcomm® ID. Once signed in navigate to `Account -> Settings -> API Token`.
With this API token, you can configure your client to run models on the cloud
hosted devices.
```bash
qai-hub configure --api_token API_TOKEN
```
Navigate to [docs](https://app.aihub.qualcomm.com/docs/) for more information.
## Demo off target
The package contains a simple end-to-end demo that downloads pre-trained
weights and runs this model on a sample input.
```bash
python -m qai_hub_models.models.efficientformer.demo
```
The above demo runs a reference implementation of pre-processing, model
inference, and post processing.
**NOTE**: If you want running in a Jupyter Notebook or Google Colab like
environment, please add the following to your cell (instead of the above).
```
%run -m qai_hub_models.models.efficientformer.demo
```
### Run model on a cloud-hosted device
In addition to the demo, you can also run the model on a cloud-hosted Qualcomm®
device. This script does the following:
* Performance check on-device on a cloud-hosted device
* Downloads compiled assets that can be deployed on-device for Android.
* Accuracy check between PyTorch and on-device outputs.
```bash
python -m qai_hub_models.models.efficientformer.export
```
## How does this work?
This [export script](https://aihub.qualcomm.com/models/efficientformer/qai_hub_models/models/EfficientFormer/export.py)
leverages [Qualcomm® AI Hub](https://aihub.qualcomm.com/) to optimize, validate, and deploy this model
on-device. Lets go through each step below in detail:
Step 1: **Compile model for on-device deployment**
To compile a PyTorch model for on-device deployment, we first trace the model
in memory using the `jit.trace` and then call the `submit_compile_job` API.
```python
import torch
import qai_hub as hub
from qai_hub_models.models.efficientformer import Model
# Load the model
torch_model = Model.from_pretrained()
# Device
device = hub.Device("Samsung Galaxy S24")
# Trace model
input_shape = torch_model.get_input_spec()
sample_inputs = torch_model.sample_inputs()
pt_model = torch.jit.trace(torch_model, [torch.tensor(data[0]) for _, data in sample_inputs.items()])
# Compile model on a specific device
compile_job = hub.submit_compile_job(
model=pt_model,
device=device,
input_specs=torch_model.get_input_spec(),
)
# Get target model to run on-device
target_model = compile_job.get_target_model()
```
Step 2: **Performance profiling on cloud-hosted device**
After compiling models from step 1. Models can be profiled model on-device using the
`target_model`. Note that this scripts runs the model on a device automatically
provisioned in the cloud. Once the job is submitted, you can navigate to a
provided job URL to view a variety of on-device performance metrics.
```python
profile_job = hub.submit_profile_job(
model=target_model,
device=device,
)
```
Step 3: **Verify on-device accuracy**
To verify the accuracy of the model on-device, you can run on-device inference
on sample input data on the same cloud hosted device.
```python
input_data = torch_model.sample_inputs()
inference_job = hub.submit_inference_job(
model=target_model,
device=device,
inputs=input_data,
)
on_device_output = inference_job.download_output_data()
```
With the output of the model, you can compute like PSNR, relative errors or
spot check the output with expected output.
**Note**: This on-device profiling and inference requires access to Qualcomm®
AI Hub. [Sign up for access](https://myaccount.qualcomm.com/signup).
## Run demo on a cloud-hosted device
You can also run the demo on-device.
```bash
python -m qai_hub_models.models.efficientformer.demo --eval-mode on-device
```
**NOTE**: If you want running in a Jupyter Notebook or Google Colab like
environment, please add the following to your cell (instead of the above).
```
%run -m qai_hub_models.models.efficientformer.demo -- --eval-mode on-device
```
## Deploying compiled model to Android
The models can be deployed using multiple runtimes:
- TensorFlow Lite (`.tflite` export): [This
tutorial](https://www.tensorflow.org/lite/android/quickstart) provides a
guide to deploy the .tflite model in an Android application.
- QNN (`.so` export ): This [sample
app](https://docs.qualcomm.com/bundle/publicresource/topics/80-63442-50/sample_app.html)
provides instructions on how to use the `.so` shared library in an Android application.
## View on Qualcomm® AI Hub
Get more details on EfficientFormer's performance across various devices [here](https://aihub.qualcomm.com/models/efficientformer).
Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
## License
* The license for the original implementation of EfficientFormer can be found
[here](https://github.com/snap-research/EfficientFormer?tab=License-1-ov-file#readme).
* The license for the compiled assets for on-device deployment can be found [here](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/Qualcomm+AI+Hub+Proprietary+License.pdf)
## References
* [Rethinking Vision Transformers for MobileNet Size and Speed](https://arxiv.org/abs/2212.08059)
* [Source Model Implementation](https://github.com/snap-research/EfficientFormer)
## Community
* Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
* For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).
|
ultratopaz/750003
|
ultratopaz
| 2025-08-29T23:14:44Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T23:14:38Z |
[View on Civ Archive](https://civarchive.com/models/747271?modelVersionId=835664)
|
qualcomm/DeepLabV3-Plus-MobileNet
|
qualcomm
| 2025-08-29T23:11:56Z | 242 | 0 |
pytorch
|
[
"pytorch",
"tflite",
"android",
"image-segmentation",
"arxiv:1706.05587",
"license:other",
"region:us"
] |
image-segmentation
| 2024-04-30T21:49:49Z |
---
library_name: pytorch
license: other
tags:
- android
pipeline_tag: image-segmentation
---
# DeepLabV3-Plus-MobileNet: Optimized for Mobile Deployment
## Deep Convolutional Neural Network model for semantic segmentation
DeepLabV3 is designed for semantic segmentation at multiple scales, trained on the various datasets. It uses MobileNet as a backbone.
This model is an implementation of DeepLabV3-Plus-MobileNet found [here](https://github.com/jfzhang95/pytorch-deeplab-xception).
This repository provides scripts to run DeepLabV3-Plus-MobileNet on Qualcomm® devices.
More details on model performance across various devices, can be found
[here](https://aihub.qualcomm.com/models/deeplabv3_plus_mobilenet).
### Model Details
- **Model Type:** Model_use_case.semantic_segmentation
- **Model Stats:**
- Model checkpoint: VOC2012
- Input resolution: 513x513
- Number of output classes: 21
- Number of parameters: 5.80M
- Model size (float): 22.2 MB
- Model size (w8a16): 6.67 MB
| Model | Precision | Device | Chipset | Target Runtime | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit | Target Model
|---|---|---|---|---|---|---|---|---|
| DeepLabV3-Plus-MobileNet | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 62.712 ms | 0 - 30 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.tflite) |
| DeepLabV3-Plus-MobileNet | float | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 58.624 ms | 0 - 33 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) |
| DeepLabV3-Plus-MobileNet | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 19.289 ms | 0 - 43 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.tflite) |
| DeepLabV3-Plus-MobileNet | float | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 18.649 ms | 3 - 58 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) |
| DeepLabV3-Plus-MobileNet | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 13.043 ms | 0 - 12 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.tflite) |
| DeepLabV3-Plus-MobileNet | float | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 11.073 ms | 3 - 16 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) |
| DeepLabV3-Plus-MobileNet | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 19.562 ms | 0 - 31 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.tflite) |
| DeepLabV3-Plus-MobileNet | float | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 17.427 ms | 2 - 35 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) |
| DeepLabV3-Plus-MobileNet | float | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 62.712 ms | 0 - 30 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.tflite) |
| DeepLabV3-Plus-MobileNet | float | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 58.624 ms | 0 - 33 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) |
| DeepLabV3-Plus-MobileNet | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 13.094 ms | 0 - 12 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.tflite) |
| DeepLabV3-Plus-MobileNet | float | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 11.088 ms | 3 - 20 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) |
| DeepLabV3-Plus-MobileNet | float | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 22.2 ms | 0 - 36 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.tflite) |
| DeepLabV3-Plus-MobileNet | float | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 19.014 ms | 2 - 53 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) |
| DeepLabV3-Plus-MobileNet | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 13.062 ms | 0 - 13 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.tflite) |
| DeepLabV3-Plus-MobileNet | float | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 11.071 ms | 4 - 21 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) |
| DeepLabV3-Plus-MobileNet | float | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 19.562 ms | 0 - 31 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.tflite) |
| DeepLabV3-Plus-MobileNet | float | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 17.427 ms | 2 - 35 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) |
| DeepLabV3-Plus-MobileNet | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 13.092 ms | 0 - 14 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.tflite) |
| DeepLabV3-Plus-MobileNet | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 11.056 ms | 3 - 16 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) |
| DeepLabV3-Plus-MobileNet | float | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 10.731 ms | 0 - 35 MB | NPU | [DeepLabV3-Plus-MobileNet.onnx.zip](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.onnx.zip) |
| DeepLabV3-Plus-MobileNet | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 9.14 ms | 0 - 49 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.tflite) |
| DeepLabV3-Plus-MobileNet | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 7.99 ms | 3 - 48 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) |
| DeepLabV3-Plus-MobileNet | float | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 7.561 ms | 0 - 44 MB | NPU | [DeepLabV3-Plus-MobileNet.onnx.zip](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.onnx.zip) |
| DeepLabV3-Plus-MobileNet | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 8.606 ms | 0 - 35 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.tflite) |
| DeepLabV3-Plus-MobileNet | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 7.418 ms | 3 - 67 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) |
| DeepLabV3-Plus-MobileNet | float | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 7.498 ms | 2 - 39 MB | NPU | [DeepLabV3-Plus-MobileNet.onnx.zip](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.onnx.zip) |
| DeepLabV3-Plus-MobileNet | float | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 11.89 ms | 19 - 19 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.dlc) |
| DeepLabV3-Plus-MobileNet | float | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 11.937 ms | 10 - 10 MB | NPU | [DeepLabV3-Plus-MobileNet.onnx.zip](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet.onnx.zip) |
| DeepLabV3-Plus-MobileNet | w8a16 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 22.273 ms | 2 - 45 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) |
| DeepLabV3-Plus-MobileNet | w8a16 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 11.199 ms | 2 - 57 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) |
| DeepLabV3-Plus-MobileNet | w8a16 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 8.832 ms | 2 - 18 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) |
| DeepLabV3-Plus-MobileNet | w8a16 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 9.369 ms | 2 - 45 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) |
| DeepLabV3-Plus-MobileNet | w8a16 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | QNN_DLC | 43.995 ms | 2 - 112 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) |
| DeepLabV3-Plus-MobileNet | w8a16 | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 22.273 ms | 2 - 45 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) |
| DeepLabV3-Plus-MobileNet | w8a16 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 8.834 ms | 2 - 17 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) |
| DeepLabV3-Plus-MobileNet | w8a16 | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 13.391 ms | 2 - 52 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) |
| DeepLabV3-Plus-MobileNet | w8a16 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 8.825 ms | 2 - 17 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) |
| DeepLabV3-Plus-MobileNet | w8a16 | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 9.369 ms | 2 - 45 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) |
| DeepLabV3-Plus-MobileNet | w8a16 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 8.841 ms | 2 - 18 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) |
| DeepLabV3-Plus-MobileNet | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 6.394 ms | 2 - 58 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) |
| DeepLabV3-Plus-MobileNet | w8a16 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 133.467 ms | 53 - 2953 MB | NPU | [DeepLabV3-Plus-MobileNet.onnx.zip](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.onnx.zip) |
| DeepLabV3-Plus-MobileNet | w8a16 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 5.209 ms | 2 - 50 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) |
| DeepLabV3-Plus-MobileNet | w8a16 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 134.935 ms | 100 - 1747 MB | NPU | [DeepLabV3-Plus-MobileNet.onnx.zip](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.onnx.zip) |
| DeepLabV3-Plus-MobileNet | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 9.552 ms | 19 - 19 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.dlc) |
| DeepLabV3-Plus-MobileNet | w8a16 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 179.765 ms | 133 - 133 MB | NPU | [DeepLabV3-Plus-MobileNet.onnx.zip](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a16.onnx.zip) |
| DeepLabV3-Plus-MobileNet | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | TFLITE | 11.936 ms | 0 - 35 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) |
| DeepLabV3-Plus-MobileNet | w8a8 | QCS8275 (Proxy) | Qualcomm® QCS8275 (Proxy) | QNN_DLC | 10.623 ms | 1 - 38 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) |
| DeepLabV3-Plus-MobileNet | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | TFLITE | 5.387 ms | 0 - 52 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) |
| DeepLabV3-Plus-MobileNet | w8a8 | QCS8450 (Proxy) | Qualcomm® QCS8450 (Proxy) | QNN_DLC | 6.462 ms | 1 - 52 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) |
| DeepLabV3-Plus-MobileNet | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | TFLITE | 4.907 ms | 0 - 17 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) |
| DeepLabV3-Plus-MobileNet | w8a8 | QCS8550 (Proxy) | Qualcomm® QCS8550 (Proxy) | QNN_DLC | 4.123 ms | 1 - 11 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) |
| DeepLabV3-Plus-MobileNet | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | TFLITE | 5.486 ms | 0 - 35 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) |
| DeepLabV3-Plus-MobileNet | w8a8 | QCS9075 (Proxy) | Qualcomm® QCS9075 (Proxy) | QNN_DLC | 4.944 ms | 1 - 38 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) |
| DeepLabV3-Plus-MobileNet | w8a8 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | TFLITE | 18.03 ms | 0 - 42 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) |
| DeepLabV3-Plus-MobileNet | w8a8 | RB3 Gen 2 (Proxy) | Qualcomm® QCS6490 (Proxy) | QNN_DLC | 19.849 ms | 1 - 54 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) |
| DeepLabV3-Plus-MobileNet | w8a8 | RB5 (Proxy) | Qualcomm® QCS8250 (Proxy) | TFLITE | 172.017 ms | 3 - 6 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) |
| DeepLabV3-Plus-MobileNet | w8a8 | SA7255P ADP | Qualcomm® SA7255P | TFLITE | 11.936 ms | 0 - 35 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) |
| DeepLabV3-Plus-MobileNet | w8a8 | SA7255P ADP | Qualcomm® SA7255P | QNN_DLC | 10.623 ms | 1 - 38 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) |
| DeepLabV3-Plus-MobileNet | w8a8 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | TFLITE | 4.91 ms | 0 - 13 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) |
| DeepLabV3-Plus-MobileNet | w8a8 | SA8255 (Proxy) | Qualcomm® SA8255P (Proxy) | QNN_DLC | 4.135 ms | 1 - 11 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) |
| DeepLabV3-Plus-MobileNet | w8a8 | SA8295P ADP | Qualcomm® SA8295P | TFLITE | 7.203 ms | 0 - 41 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) |
| DeepLabV3-Plus-MobileNet | w8a8 | SA8295P ADP | Qualcomm® SA8295P | QNN_DLC | 6.382 ms | 1 - 45 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) |
| DeepLabV3-Plus-MobileNet | w8a8 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | TFLITE | 4.894 ms | 0 - 18 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) |
| DeepLabV3-Plus-MobileNet | w8a8 | SA8650 (Proxy) | Qualcomm® SA8650P (Proxy) | QNN_DLC | 4.124 ms | 1 - 12 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) |
| DeepLabV3-Plus-MobileNet | w8a8 | SA8775P ADP | Qualcomm® SA8775P | TFLITE | 5.486 ms | 0 - 35 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) |
| DeepLabV3-Plus-MobileNet | w8a8 | SA8775P ADP | Qualcomm® SA8775P | QNN_DLC | 4.944 ms | 1 - 38 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) |
| DeepLabV3-Plus-MobileNet | w8a8 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | TFLITE | 4.899 ms | 0 - 16 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) |
| DeepLabV3-Plus-MobileNet | w8a8 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | QNN_DLC | 4.131 ms | 1 - 12 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) |
| DeepLabV3-Plus-MobileNet | w8a8 | Samsung Galaxy S23 | Snapdragon® 8 Gen 2 Mobile | ONNX | 124.508 ms | 92 - 207 MB | NPU | [DeepLabV3-Plus-MobileNet.onnx.zip](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.onnx.zip) |
| DeepLabV3-Plus-MobileNet | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | TFLITE | 3.479 ms | 0 - 49 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) |
| DeepLabV3-Plus-MobileNet | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | QNN_DLC | 2.857 ms | 1 - 56 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) |
| DeepLabV3-Plus-MobileNet | w8a8 | Samsung Galaxy S24 | Snapdragon® 8 Gen 3 Mobile | ONNX | 96.628 ms | 85 - 1634 MB | NPU | [DeepLabV3-Plus-MobileNet.onnx.zip](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.onnx.zip) |
| DeepLabV3-Plus-MobileNet | w8a8 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | TFLITE | 3.164 ms | 0 - 41 MB | NPU | [DeepLabV3-Plus-MobileNet.tflite](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.tflite) |
| DeepLabV3-Plus-MobileNet | w8a8 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | QNN_DLC | 2.569 ms | 1 - 45 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) |
| DeepLabV3-Plus-MobileNet | w8a8 | Snapdragon 8 Elite QRD | Snapdragon® 8 Elite Mobile | ONNX | 97.314 ms | 90 - 1259 MB | NPU | [DeepLabV3-Plus-MobileNet.onnx.zip](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.onnx.zip) |
| DeepLabV3-Plus-MobileNet | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | QNN_DLC | 4.57 ms | 13 - 13 MB | NPU | [DeepLabV3-Plus-MobileNet.dlc](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.dlc) |
| DeepLabV3-Plus-MobileNet | w8a8 | Snapdragon X Elite CRD | Snapdragon® X Elite | ONNX | 118.39 ms | 133 - 133 MB | NPU | [DeepLabV3-Plus-MobileNet.onnx.zip](https://huggingface.co/qualcomm/DeepLabV3-Plus-MobileNet/blob/main/DeepLabV3-Plus-MobileNet_w8a8.onnx.zip) |
## Installation
Install the package via pip:
```bash
pip install qai-hub-models
```
## Configure Qualcomm® AI Hub to run this model on a cloud-hosted device
Sign-in to [Qualcomm® AI Hub](https://app.aihub.qualcomm.com/) with your
Qualcomm® ID. Once signed in navigate to `Account -> Settings -> API Token`.
With this API token, you can configure your client to run models on the cloud
hosted devices.
```bash
qai-hub configure --api_token API_TOKEN
```
Navigate to [docs](https://app.aihub.qualcomm.com/docs/) for more information.
## Demo off target
The package contains a simple end-to-end demo that downloads pre-trained
weights and runs this model on a sample input.
```bash
python -m qai_hub_models.models.deeplabv3_plus_mobilenet.demo
```
The above demo runs a reference implementation of pre-processing, model
inference, and post processing.
**NOTE**: If you want running in a Jupyter Notebook or Google Colab like
environment, please add the following to your cell (instead of the above).
```
%run -m qai_hub_models.models.deeplabv3_plus_mobilenet.demo
```
### Run model on a cloud-hosted device
In addition to the demo, you can also run the model on a cloud-hosted Qualcomm®
device. This script does the following:
* Performance check on-device on a cloud-hosted device
* Downloads compiled assets that can be deployed on-device for Android.
* Accuracy check between PyTorch and on-device outputs.
```bash
python -m qai_hub_models.models.deeplabv3_plus_mobilenet.export
```
## How does this work?
This [export script](https://aihub.qualcomm.com/models/deeplabv3_plus_mobilenet/qai_hub_models/models/DeepLabV3-Plus-MobileNet/export.py)
leverages [Qualcomm® AI Hub](https://aihub.qualcomm.com/) to optimize, validate, and deploy this model
on-device. Lets go through each step below in detail:
Step 1: **Compile model for on-device deployment**
To compile a PyTorch model for on-device deployment, we first trace the model
in memory using the `jit.trace` and then call the `submit_compile_job` API.
```python
import torch
import qai_hub as hub
from qai_hub_models.models.deeplabv3_plus_mobilenet import Model
# Load the model
torch_model = Model.from_pretrained()
# Device
device = hub.Device("Samsung Galaxy S24")
# Trace model
input_shape = torch_model.get_input_spec()
sample_inputs = torch_model.sample_inputs()
pt_model = torch.jit.trace(torch_model, [torch.tensor(data[0]) for _, data in sample_inputs.items()])
# Compile model on a specific device
compile_job = hub.submit_compile_job(
model=pt_model,
device=device,
input_specs=torch_model.get_input_spec(),
)
# Get target model to run on-device
target_model = compile_job.get_target_model()
```
Step 2: **Performance profiling on cloud-hosted device**
After compiling models from step 1. Models can be profiled model on-device using the
`target_model`. Note that this scripts runs the model on a device automatically
provisioned in the cloud. Once the job is submitted, you can navigate to a
provided job URL to view a variety of on-device performance metrics.
```python
profile_job = hub.submit_profile_job(
model=target_model,
device=device,
)
```
Step 3: **Verify on-device accuracy**
To verify the accuracy of the model on-device, you can run on-device inference
on sample input data on the same cloud hosted device.
```python
input_data = torch_model.sample_inputs()
inference_job = hub.submit_inference_job(
model=target_model,
device=device,
inputs=input_data,
)
on_device_output = inference_job.download_output_data()
```
With the output of the model, you can compute like PSNR, relative errors or
spot check the output with expected output.
**Note**: This on-device profiling and inference requires access to Qualcomm®
AI Hub. [Sign up for access](https://myaccount.qualcomm.com/signup).
## Run demo on a cloud-hosted device
You can also run the demo on-device.
```bash
python -m qai_hub_models.models.deeplabv3_plus_mobilenet.demo --eval-mode on-device
```
**NOTE**: If you want running in a Jupyter Notebook or Google Colab like
environment, please add the following to your cell (instead of the above).
```
%run -m qai_hub_models.models.deeplabv3_plus_mobilenet.demo -- --eval-mode on-device
```
## Deploying compiled model to Android
The models can be deployed using multiple runtimes:
- TensorFlow Lite (`.tflite` export): [This
tutorial](https://www.tensorflow.org/lite/android/quickstart) provides a
guide to deploy the .tflite model in an Android application.
- QNN (`.so` export ): This [sample
app](https://docs.qualcomm.com/bundle/publicresource/topics/80-63442-50/sample_app.html)
provides instructions on how to use the `.so` shared library in an Android application.
## View on Qualcomm® AI Hub
Get more details on DeepLabV3-Plus-MobileNet's performance across various devices [here](https://aihub.qualcomm.com/models/deeplabv3_plus_mobilenet).
Explore all available models on [Qualcomm® AI Hub](https://aihub.qualcomm.com/)
## License
* The license for the original implementation of DeepLabV3-Plus-MobileNet can be found
[here](https://github.com/jfzhang95/pytorch-deeplab-xception/blob/master/LICENSE).
* The license for the compiled assets for on-device deployment can be found [here](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/Qualcomm+AI+Hub+Proprietary+License.pdf)
## References
* [Rethinking Atrous Convolution for Semantic Image Segmentation](https://arxiv.org/abs/1706.05587)
* [Source Model Implementation](https://github.com/jfzhang95/pytorch-deeplab-xception)
## Community
* Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
* For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).
|
amethyst9/609395
|
amethyst9
| 2025-08-29T23:09:23Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T23:09:18Z |
[View on Civ Archive](https://civarchive.com/models/621225?modelVersionId=694521)
|
motza0025/blockassist-bc-hairy_fierce_mosquito_1756507482
|
motza0025
| 2025-08-29T23:08:52Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"hairy fierce mosquito",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T23:08:34Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- hairy fierce mosquito
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
crystalline7/764486
|
crystalline7
| 2025-08-29T23:06:02Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T23:05:53Z |
[View on Civ Archive](https://civarchive.com/models/760113?modelVersionId=849953)
|
amethyst9/717603
|
amethyst9
| 2025-08-29T23:04:18Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T23:04:12Z |
[View on Civ Archive](https://civarchive.com/models/719062?modelVersionId=804045)
|
ultratopaz/930325
|
ultratopaz
| 2025-08-29T23:01:41Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T23:01:36Z |
[View on Civ Archive](https://civarchive.com/models/915095?modelVersionId=1024201)
|
ultratopaz/728450
|
ultratopaz
| 2025-08-29T22:59:53Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T22:59:48Z |
[View on Civ Archive](https://civarchive.com/models/728322?modelVersionId=814419)
|
ultratopaz/565554
|
ultratopaz
| 2025-08-29T22:56:34Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T22:56:28Z |
[View on Civ Archive](https://civarchive.com/models/583071?modelVersionId=650482)
|
crystalline7/446156
|
crystalline7
| 2025-08-29T22:55:51Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T22:55:45Z |
[View on Civ Archive](https://civarchive.com/models/475846?modelVersionId=529270)
|
ypszn/blockassist-bc-yapping_pawing_worm_1756508093
|
ypszn
| 2025-08-29T22:55:40Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"yapping pawing worm",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T22:55:33Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- yapping pawing worm
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
crystalline7/398723
|
crystalline7
| 2025-08-29T22:55:11Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T22:55:08Z |
[View on Civ Archive](https://civarchive.com/models/431131?modelVersionId=480313)
|
vendi11/blockassist-bc-placid_placid_llama_1756507881
|
vendi11
| 2025-08-29T22:52:03Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"placid placid llama",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T22:52:00Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- placid placid llama
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
keras/mistral_0.3_instruct_7b_en
|
keras
| 2025-08-29T22:47:42Z | 0 | 0 |
keras-hub
|
[
"keras-hub",
"text-generation",
"region:us"
] |
text-generation
| 2025-08-29T22:43:57Z |
---
library_name: keras-hub
pipeline_tag: text-generation
---
This is a [`Mistral` model](https://keras.io/api/keras_hub/models/mistral) uploaded using the KerasHub library and can be used with JAX, TensorFlow, and PyTorch backends.
This model is related to a `CausalLM` task.
Model config:
* **name:** mistral_backbone_1
* **trainable:** True
* **vocabulary_size:** 32768
* **num_layers:** 32
* **num_query_heads:** 32
* **hidden_dim:** 4096
* **intermediate_dim:** 14336
* **rope_max_wavelength:** 1000000.0
* **rope_scaling_factor:** 1.0
* **num_key_value_heads:** 8
* **sliding_window:** None
* **layer_norm_epsilon:** 1e-05
* **dropout:** 0
This model card has been generated automatically and should be completed by the model author. See [Model Cards documentation](https://huggingface.co/docs/hub/model-cards) for more information.
|
rettertop/blockassist-bc-fishy_hunting_sparrow_1756507304
|
rettertop
| 2025-08-29T22:42:29Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"fishy hunting sparrow",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T22:41:45Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- fishy hunting sparrow
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
liukevin666/blockassist-bc-yawning_striped_cassowary_1756507129
|
liukevin666
| 2025-08-29T22:40:06Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"yawning striped cassowary",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T22:39:44Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- yawning striped cassowary
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
Sonic-man/blockassist-bc-poisonous_graceful_cow_1756504864
|
Sonic-man
| 2025-08-29T22:38:56Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"poisonous graceful cow",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T22:38:50Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- poisonous graceful cow
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
bah63843/blockassist-bc-plump_fast_antelope_1756506980
|
bah63843
| 2025-08-29T22:37:11Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"plump fast antelope",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T22:37:02Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- plump fast antelope
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
mradermacher/Foundation-Sec-8B-Chinese-Chat-GGUF
|
mradermacher
| 2025-08-29T22:36:14Z | 0 | 0 |
transformers
|
[
"transformers",
"gguf",
"mergekit",
"merge",
"security",
"cybersecurity",
"network-security",
"llama",
"en",
"base_model:sds-ai/Foundation-Sec-8B-Chinese-Chat",
"base_model:quantized:sds-ai/Foundation-Sec-8B-Chinese-Chat",
"license:llama3.1",
"endpoints_compatible",
"region:us",
"conversational"
] | null | 2025-08-29T22:00:30Z |
---
base_model: sds-ai/Foundation-Sec-8B-Chinese-Chat
language:
- en
library_name: transformers
license: llama3.1
mradermacher:
readme_rev: 1
quantized_by: mradermacher
tags:
- mergekit
- merge
- security
- cybersecurity
- network-security
- llama
---
## About
<!-- ### quantize_version: 2 -->
<!-- ### output_tensor_quantised: 1 -->
<!-- ### convert_type: hf -->
<!-- ### vocab_type: -->
<!-- ### tags: -->
<!-- ### quants: x-f16 Q4_K_S Q2_K Q8_0 Q6_K Q3_K_M Q3_K_S Q3_K_L Q4_K_M Q5_K_S Q5_K_M IQ4_XS -->
<!-- ### quants_skip: -->
<!-- ### skip_mmproj: -->
static quants of https://huggingface.co/sds-ai/Foundation-Sec-8B-Chinese-Chat
<!-- provided-files -->
***For a convenient overview and download list, visit our [model page for this model](https://hf.tst.eu/model#Foundation-Sec-8B-Chinese-Chat-GGUF).***
weighted/imatrix quants seem not to be available (by me) at this time. If they do not show up a week or so after the static ones, I have probably not planned for them. Feel free to request them by opening a Community Discussion.
## Usage
If you are unsure how to use GGUF files, refer to one of [TheBloke's
READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for
more details, including on how to concatenate multi-part files.
## Provided Quants
(sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
| Link | Type | Size/GB | Notes |
|:-----|:-----|--------:|:------|
| [GGUF](https://huggingface.co/mradermacher/Foundation-Sec-8B-Chinese-Chat-GGUF/resolve/main/Foundation-Sec-8B-Chinese-Chat.Q2_K.gguf) | Q2_K | 3.3 | |
| [GGUF](https://huggingface.co/mradermacher/Foundation-Sec-8B-Chinese-Chat-GGUF/resolve/main/Foundation-Sec-8B-Chinese-Chat.Q3_K_S.gguf) | Q3_K_S | 3.8 | |
| [GGUF](https://huggingface.co/mradermacher/Foundation-Sec-8B-Chinese-Chat-GGUF/resolve/main/Foundation-Sec-8B-Chinese-Chat.Q3_K_M.gguf) | Q3_K_M | 4.1 | lower quality |
| [GGUF](https://huggingface.co/mradermacher/Foundation-Sec-8B-Chinese-Chat-GGUF/resolve/main/Foundation-Sec-8B-Chinese-Chat.Q3_K_L.gguf) | Q3_K_L | 4.4 | |
| [GGUF](https://huggingface.co/mradermacher/Foundation-Sec-8B-Chinese-Chat-GGUF/resolve/main/Foundation-Sec-8B-Chinese-Chat.Q4_K_S.gguf) | Q4_K_S | 4.8 | fast, recommended |
| [GGUF](https://huggingface.co/mradermacher/Foundation-Sec-8B-Chinese-Chat-GGUF/resolve/main/Foundation-Sec-8B-Chinese-Chat.Q6_K.gguf) | Q6_K | 6.7 | very good quality |
| [GGUF](https://huggingface.co/mradermacher/Foundation-Sec-8B-Chinese-Chat-GGUF/resolve/main/Foundation-Sec-8B-Chinese-Chat.Q8_0.gguf) | Q8_0 | 8.6 | fast, best quality |
Here is a handy graph by ikawrakow comparing some lower-quality quant
types (lower is better):
And here are Artefact2's thoughts on the matter:
https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9
## FAQ / Model Request
See https://huggingface.co/mradermacher/model_requests for some answers to
questions you might have and/or if you want some other model quantized.
## Thanks
I thank my company, [nethype GmbH](https://www.nethype.de/), for letting
me use its servers and providing upgrades to my workstation to enable
this work in my free time.
<!-- end -->
|
seraphimzzzz/398686
|
seraphimzzzz
| 2025-08-29T22:34:31Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T22:34:31Z |
[View on Civ Archive](https://civarchive.com/models/431088?modelVersionId=480270)
|
crystalline7/833702
|
crystalline7
| 2025-08-29T22:32:57Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T22:32:51Z |
[View on Civ Archive](https://civarchive.com/models/585030?modelVersionId=926215)
|
crystalline7/965306
|
crystalline7
| 2025-08-29T22:28:48Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T22:28:41Z |
[View on Civ Archive](https://civarchive.com/models/585614?modelVersionId=1059766)
|
vendi11/blockassist-bc-placid_placid_llama_1756506339
|
vendi11
| 2025-08-29T22:26:20Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"placid placid llama",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T22:26:17Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- placid placid llama
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
AnonymousCS/populism_classifier_bsample_231
|
AnonymousCS
| 2025-08-29T22:25:13Z | 0 | 0 |
transformers
|
[
"transformers",
"safetensors",
"xlm-roberta",
"text-classification",
"generated_from_trainer",
"base_model:AnonymousCS/populism_xlmr_large",
"base_model:finetune:AnonymousCS/populism_xlmr_large",
"license:mit",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2025-08-29T22:21:49Z |
---
library_name: transformers
license: mit
base_model: AnonymousCS/populism_xlmr_large
tags:
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: populism_classifier_bsample_231
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. -->
# populism_classifier_bsample_231
This model is a fine-tuned version of [AnonymousCS/populism_xlmr_large](https://huggingface.co/AnonymousCS/populism_xlmr_large) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.7087
- Accuracy: 0.0765
- 1-f1: 0.1420
- 1-recall: 1.0
- 1-precision: 0.0765
- Balanced Acc: 0.5
## 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-06
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.06
- num_epochs: 15
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | 1-f1 | 1-recall | 1-precision | Balanced Acc |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:------:|:--------:|:-----------:|:------------:|
| 0.9082 | 1.0 | 11 | 0.4867 | 0.9235 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.7994 | 2.0 | 22 | 0.5335 | 0.9235 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.7519 | 3.0 | 33 | 0.5816 | 0.9235 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.8839 | 4.0 | 44 | 0.6264 | 0.9235 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.7397 | 5.0 | 55 | 0.6723 | 0.9235 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.8888 | 6.0 | 66 | 0.7087 | 0.0765 | 0.1420 | 1.0 | 0.0765 | 0.5 |
### Framework versions
- Transformers 4.46.3
- Pytorch 2.4.1+cu121
- Datasets 3.1.0
- Tokenizers 0.20.3
|
stegostegosaur/mms-1b-ngen
|
stegostegosaur
| 2025-08-29T22:24:40Z | 0 | 0 |
transformers
|
[
"transformers",
"safetensors",
"wav2vec2",
"automatic-speech-recognition",
"generated_from_trainer",
"base_model:facebook/mms-1b-all",
"base_model:finetune:facebook/mms-1b-all",
"license:cc-by-nc-4.0",
"endpoints_compatible",
"region:us"
] |
automatic-speech-recognition
| 2025-05-13T14:04:24Z |
---
library_name: transformers
license: cc-by-nc-4.0
base_model: facebook/mms-1b-all
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: mms-1b-ngen
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. -->
# mms-1b-ngen
This model is a fine-tuned version of [facebook/mms-1b-all](https://huggingface.co/facebook/mms-1b-all) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 4.9158
- Wer: 1.0696
## 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-06
- train_batch_size: 1
- eval_batch_size: 1
- seed: 42
- gradient_accumulation_steps: 16
- total_train_batch_size: 16
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 100
- num_epochs: 5
### Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|:-------------:|:------:|:----:|:---------------:|:------:|
| No log | 1.0 | 42 | 5.2245 | 1.0059 |
| 83.4246 | 2.0 | 84 | 5.1472 | 1.0207 |
| 79.1451 | 3.0 | 126 | 5.0228 | 1.0585 |
| 82.1059 | 4.0 | 168 | 4.9392 | 1.0681 |
| 75.786 | 4.8997 | 205 | 4.9158 | 1.0696 |
### Framework versions
- Transformers 4.48.0.dev0
- Pytorch 2.5.1+cu124
- Datasets 3.2.0
- Tokenizers 0.21.4
|
crystalline7/1038516
|
crystalline7
| 2025-08-29T22:24:16Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T22:24:06Z |
[View on Civ Archive](https://civarchive.com/models/1010752?modelVersionId=1132979)
|
ultratopaz/561183
|
ultratopaz
| 2025-08-29T22:23:41Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T22:23:35Z |
[View on Civ Archive](https://civarchive.com/models/579304?modelVersionId=646118)
|
bah63843/blockassist-bc-plump_fast_antelope_1756505995
|
bah63843
| 2025-08-29T22:20:49Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"plump fast antelope",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T22:20:41Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- plump fast antelope
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
mradermacher/codeqwen3-14b-i1-GGUF
|
mradermacher
| 2025-08-29T22:17:11Z | 0 | 0 |
transformers
|
[
"transformers",
"gguf",
"code",
"programming-tasks",
"algorithmic-reasoning",
"code-generation",
"non-commercial",
"python",
"synthetic",
"en",
"uk",
"dataset:anon-researcher-ua/ua-codeforces-cots-open-r1",
"dataset:anon-researcher-ua/ua-codeforces-cots-open-r1-for-training",
"dataset:open-r1/codeforces-cots",
"base_model:anon-researcher-ua/codeqwen3-14b",
"base_model:quantized:anon-researcher-ua/codeqwen3-14b",
"license:cc-by-4.0",
"endpoints_compatible",
"region:us",
"imatrix",
"conversational"
] | null | 2025-08-29T16:26:01Z |
---
base_model: anon-researcher-ua/codeqwen3-14b
datasets:
- anon-researcher-ua/ua-codeforces-cots-open-r1
- anon-researcher-ua/ua-codeforces-cots-open-r1-for-training
- open-r1/codeforces-cots
language:
- en
- uk
library_name: transformers
license: cc-by-4.0
mradermacher:
readme_rev: 1
quantized_by: mradermacher
tags:
- code
- programming-tasks
- algorithmic-reasoning
- code-generation
- non-commercial
- python
- synthetic
---
## About
<!-- ### quantize_version: 2 -->
<!-- ### output_tensor_quantised: 1 -->
<!-- ### convert_type: hf -->
<!-- ### vocab_type: -->
<!-- ### tags: nicoboss -->
<!-- ### quants: Q2_K IQ3_M Q4_K_S IQ3_XXS Q3_K_M small-IQ4_NL Q4_K_M IQ2_M Q6_K IQ4_XS Q2_K_S IQ1_M Q3_K_S IQ2_XXS Q3_K_L IQ2_XS Q5_K_S IQ2_S IQ1_S Q5_K_M Q4_0 IQ3_XS Q4_1 IQ3_S -->
<!-- ### quants_skip: -->
<!-- ### skip_mmproj: -->
weighted/imatrix quants of https://huggingface.co/anon-researcher-ua/codeqwen3-14b
<!-- provided-files -->
***For a convenient overview and download list, visit our [model page for this model](https://hf.tst.eu/model#codeqwen3-14b-i1-GGUF).***
static quants are available at https://huggingface.co/mradermacher/codeqwen3-14b-GGUF
## Usage
If you are unsure how to use GGUF files, refer to one of [TheBloke's
READMEs](https://huggingface.co/TheBloke/KafkaLM-70B-German-V0.1-GGUF) for
more details, including on how to concatenate multi-part files.
## Provided Quants
(sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)
| Link | Type | Size/GB | Notes |
|:-----|:-----|--------:|:------|
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.imatrix.gguf) | imatrix | 0.1 | imatrix file (for creating your own qwuants) |
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.i1-IQ1_S.gguf) | i1-IQ1_S | 3.7 | for the desperate |
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.i1-IQ1_M.gguf) | i1-IQ1_M | 3.9 | mostly desperate |
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.i1-IQ2_XXS.gguf) | i1-IQ2_XXS | 4.4 | |
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.i1-IQ2_XS.gguf) | i1-IQ2_XS | 4.8 | |
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.i1-IQ2_S.gguf) | i1-IQ2_S | 5.1 | |
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.i1-IQ2_M.gguf) | i1-IQ2_M | 5.4 | |
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.i1-Q2_K_S.gguf) | i1-Q2_K_S | 5.5 | very low quality |
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.i1-Q2_K.gguf) | i1-Q2_K | 5.9 | IQ3_XXS probably better |
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.i1-IQ3_XXS.gguf) | i1-IQ3_XXS | 6.0 | lower quality |
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.i1-IQ3_XS.gguf) | i1-IQ3_XS | 6.5 | |
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.i1-Q3_K_S.gguf) | i1-Q3_K_S | 6.8 | IQ3_XS probably better |
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.i1-IQ3_S.gguf) | i1-IQ3_S | 6.8 | beats Q3_K* |
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.i1-IQ3_M.gguf) | i1-IQ3_M | 7.0 | |
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.i1-Q3_K_M.gguf) | i1-Q3_K_M | 7.4 | IQ3_S probably better |
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.i1-Q3_K_L.gguf) | i1-Q3_K_L | 8.0 | IQ3_M probably better |
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.i1-IQ4_XS.gguf) | i1-IQ4_XS | 8.2 | |
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.i1-IQ4_NL.gguf) | i1-IQ4_NL | 8.6 | prefer IQ4_XS |
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.i1-Q4_0.gguf) | i1-Q4_0 | 8.6 | fast, low quality |
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.i1-Q4_K_S.gguf) | i1-Q4_K_S | 8.7 | optimal size/speed/quality |
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.i1-Q4_K_M.gguf) | i1-Q4_K_M | 9.1 | fast, recommended |
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.i1-Q4_1.gguf) | i1-Q4_1 | 9.5 | |
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.i1-Q5_K_S.gguf) | i1-Q5_K_S | 10.4 | |
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.i1-Q5_K_M.gguf) | i1-Q5_K_M | 10.6 | |
| [GGUF](https://huggingface.co/mradermacher/codeqwen3-14b-i1-GGUF/resolve/main/codeqwen3-14b.i1-Q6_K.gguf) | i1-Q6_K | 12.2 | practically like static Q6_K |
Here is a handy graph by ikawrakow comparing some lower-quality quant
types (lower is better):
And here are Artefact2's thoughts on the matter:
https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9
## FAQ / Model Request
See https://huggingface.co/mradermacher/model_requests for some answers to
questions you might have and/or if you want some other model quantized.
## Thanks
I thank my company, [nethype GmbH](https://www.nethype.de/), for letting
me use its servers and providing upgrades to my workstation to enable
this work in my free time. Additional thanks to [@nicoboss](https://huggingface.co/nicoboss) for giving me access to his private supercomputer, enabling me to provide many more imatrix quants, at much higher quality, than I would otherwise be able to.
<!-- end -->
|
crystalline7/1008796
|
crystalline7
| 2025-08-29T22:17:07Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T22:17:01Z |
[View on Civ Archive](https://civarchive.com/models/973733?modelVersionId=1090387)
|
ultratopaz/589644
|
ultratopaz
| 2025-08-29T22:15:40Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T22:15:33Z |
[View on Civ Archive](https://civarchive.com/models/601919?modelVersionId=674600)
|
bah63843/blockassist-bc-plump_fast_antelope_1756505620
|
bah63843
| 2025-08-29T22:15:12Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"plump fast antelope",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T22:15:04Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- plump fast antelope
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
vendi11/blockassist-bc-placid_placid_llama_1756505628
|
vendi11
| 2025-08-29T22:14:30Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"placid placid llama",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T22:14:27Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- placid placid llama
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
seraphimzzzz/513872
|
seraphimzzzz
| 2025-08-29T22:13:42Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T22:13:36Z |
[View on Civ Archive](https://civarchive.com/models/538655?modelVersionId=598820)
|
seraphimzzzz/530260
|
seraphimzzzz
| 2025-08-29T22:13:15Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T22:13:09Z |
[View on Civ Archive](https://civarchive.com/models/552705?modelVersionId=615074)
|
Samuell43/blockassist-bc-fast_gregarious_warthog_1756505240
|
Samuell43
| 2025-08-29T22:07:51Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"fast gregarious warthog",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T22:07:46Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- fast gregarious warthog
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
ultratopaz/517385
|
ultratopaz
| 2025-08-29T22:07:12Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T22:07:07Z |
[View on Civ Archive](https://civarchive.com/models/500299?modelVersionId=602400)
|
ultratopaz/490828
|
ultratopaz
| 2025-08-29T22:01:12Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T22:01:06Z |
[View on Civ Archive](https://civarchive.com/models/517826?modelVersionId=575415)
|
seraphimzzzz/1345862
|
seraphimzzzz
| 2025-08-29T21:59:24Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T21:59:07Z |
[View on Civ Archive](https://civarchive.com/models/584311?modelVersionId=1444694)
|
crystalline7/668268
|
crystalline7
| 2025-08-29T21:58:59Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T21:58:56Z |
[View on Civ Archive](https://civarchive.com/models/672814?modelVersionId=753170)
|
Babsie/Loki-Omega-70B-6QMerged-GGUF
|
Babsie
| 2025-08-29T21:56:14Z | 0 | 0 | null |
[
"gguf",
"endpoints_compatible",
"region:us",
"conversational"
] | null | 2025-08-29T21:40:17Z |
# Loki-Omega-70B-6QMerged-GGUF
This repository contains a **single merged GGUF file** for running ReadyArt/L3.3-The-Omega-Directive-70B-Unslop-v2.0 with llama.cpp (Q6_K quantization).
## File
* `Loki-Omega-70B-6QMerged.gguf` (57GB) - Full merged model, ready to use
## Quick Start (llama.cpp server, OpenAI-compatible)
```bash
pip install "llama-cpp-python[server]"
python -m llama_cpp.server \
--model /path/to/Loki-Omega-70B-6QMerged.gguf \
--host 0.0.0.0 --port 8000 \
--n_ctx 32000
## Quantization: Q6_K (maintains quality and nuance)
Note: This is the merged version. For split files, see the original repository.
## Warning!!
You can use it if you want, but he vomits on everything.
Loki loves being told: **"NO LOKI! NO!"**
|
klmdr22/blockassist-bc-wild_loud_newt_1756504500
|
klmdr22
| 2025-08-29T21:55:42Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"wild loud newt",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T21:55:39Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- wild loud newt
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
AnonymousCS/populism_classifier_bsample_224
|
AnonymousCS
| 2025-08-29T21:53:54Z | 0 | 0 |
transformers
|
[
"transformers",
"safetensors",
"xlm-roberta",
"text-classification",
"generated_from_trainer",
"base_model:AnonymousCS/populism_xlmr_large",
"base_model:finetune:AnonymousCS/populism_xlmr_large",
"license:mit",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2025-08-29T21:50:15Z |
---
library_name: transformers
license: mit
base_model: AnonymousCS/populism_xlmr_large
tags:
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: populism_classifier_bsample_224
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. -->
# populism_classifier_bsample_224
This model is a fine-tuned version of [AnonymousCS/populism_xlmr_large](https://huggingface.co/AnonymousCS/populism_xlmr_large) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.6397
- Accuracy: 0.9067
- 1-f1: 0.0
- 1-recall: 0.0
- 1-precision: 0.0
- Balanced Acc: 0.5
## 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-06
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.06
- num_epochs: 15
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | 1-f1 | 1-recall | 1-precision | Balanced Acc |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:----:|:--------:|:-----------:|:------------:|
| 0.8273 | 1.0 | 8 | 0.4888 | 0.9067 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.8127 | 2.0 | 16 | 0.5213 | 0.9067 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.8245 | 3.0 | 24 | 0.5523 | 0.9067 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.822 | 4.0 | 32 | 0.5841 | 0.9067 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.7491 | 5.0 | 40 | 0.6106 | 0.9067 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.8004 | 6.0 | 48 | 0.6397 | 0.9067 | 0.0 | 0.0 | 0.0 | 0.5 |
### Framework versions
- Transformers 4.46.3
- Pytorch 2.4.1+cu121
- Datasets 3.1.0
- Tokenizers 0.20.3
|
crystalline7/664567
|
crystalline7
| 2025-08-29T21:53:53Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T21:53:47Z |
[View on Civ Archive](https://civarchive.com/models/431088?modelVersionId=751160)
|
ultratopaz/816359
|
ultratopaz
| 2025-08-29T21:53:31Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T21:53:24Z |
[View on Civ Archive](https://civarchive.com/models/774205?modelVersionId=908666)
|
crystalline7/472144
|
crystalline7
| 2025-08-29T21:51:48Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T21:51:41Z |
[View on Civ Archive](https://civarchive.com/models/500288?modelVersionId=556107)
|
AnonymousCS/populism_classifier_bsample_222
|
AnonymousCS
| 2025-08-29T21:44:56Z | 0 | 0 |
transformers
|
[
"transformers",
"safetensors",
"xlm-roberta",
"text-classification",
"generated_from_trainer",
"base_model:AnonymousCS/populism_xlmr_large",
"base_model:finetune:AnonymousCS/populism_xlmr_large",
"license:mit",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2025-08-29T21:40:26Z |
---
library_name: transformers
license: mit
base_model: AnonymousCS/populism_xlmr_large
tags:
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: populism_classifier_bsample_222
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. -->
# populism_classifier_bsample_222
This model is a fine-tuned version of [AnonymousCS/populism_xlmr_large](https://huggingface.co/AnonymousCS/populism_xlmr_large) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.4628
- Accuracy: 0.9733
- 1-f1: 0.0
- 1-recall: 0.0
- 1-precision: 0.0
- Balanced Acc: 0.5
## 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-06
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.06
- num_epochs: 15
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | 1-f1 | 1-recall | 1-precision | Balanced Acc |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:----:|:--------:|:-----------:|:------------:|
| 0.7252 | 1.0 | 38 | 0.4432 | 0.9733 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.7347 | 2.0 | 76 | 0.4492 | 0.9733 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.6432 | 3.0 | 114 | 0.4571 | 0.9733 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.6242 | 4.0 | 152 | 0.4603 | 0.9733 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.6736 | 5.0 | 190 | 0.4551 | 0.9733 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.6253 | 6.0 | 228 | 0.4628 | 0.9733 | 0.0 | 0.0 | 0.0 | 0.5 |
### Framework versions
- Transformers 4.46.3
- Pytorch 2.4.1+cu121
- Datasets 3.1.0
- Tokenizers 0.20.3
|
canoplos112/blockassist-bc-yapping_sleek_squirrel_1756503578
|
canoplos112
| 2025-08-29T21:41:29Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"yapping sleek squirrel",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T21:40:13Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- yapping sleek squirrel
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
Alasil/DR_trial
|
Alasil
| 2025-08-29T21:37:08Z | 47 | 0 |
peft
|
[
"peft",
"safetensors",
"arxiv:1910.09700",
"base_model:google/medgemma-4b-it",
"base_model:adapter:google/medgemma-4b-it",
"region:us"
] | null | 2025-08-14T19:26:02Z |
---
base_model: google/medgemma-4b-it
library_name: peft
---
# Model Card for Model ID
<!-- Provide a quick summary of what the model is/does. -->
## Model Details
### Model Description
<!-- Provide a longer summary of what this model is. -->
- **Developed by:** [More Information Needed]
- **Funded by [optional]:** [More Information Needed]
- **Shared by [optional]:** [More Information Needed]
- **Model type:** [More Information Needed]
- **Language(s) (NLP):** [More Information Needed]
- **License:** [More Information Needed]
- **Finetuned from model [optional]:** [More Information Needed]
### Model Sources [optional]
<!-- Provide the basic links for the model. -->
- **Repository:** [More Information Needed]
- **Paper [optional]:** [More Information Needed]
- **Demo [optional]:** [More Information Needed]
## Uses
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
### Direct Use
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
[More Information Needed]
### Downstream Use [optional]
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
[More Information Needed]
### Out-of-Scope Use
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
[More Information Needed]
## Bias, Risks, and Limitations
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
[More Information Needed]
### Recommendations
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
## How to Get Started with the Model
Use the code below to get started with the model.
[More Information Needed]
## Training Details
### Training Data
<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->
[More Information Needed]
### Training Procedure
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
#### Preprocessing [optional]
[More Information Needed]
#### Training Hyperparameters
- **Training regime:** [More Information Needed] <!--fp32, fp16 mixed precision, bf16 mixed precision, bf16 non-mixed precision, fp16 non-mixed precision, fp8 mixed precision -->
#### Speeds, Sizes, Times [optional]
<!-- This section provides information about throughput, start/end time, checkpoint size if relevant, etc. -->
[More Information Needed]
## Evaluation
<!-- This section describes the evaluation protocols and provides the results. -->
### Testing Data, Factors & Metrics
#### Testing Data
<!-- This should link to a Dataset Card if possible. -->
[More Information Needed]
#### Factors
<!-- These are the things the evaluation is disaggregating by, e.g., subpopulations or domains. -->
[More Information Needed]
#### Metrics
<!-- These are the evaluation metrics being used, ideally with a description of why. -->
[More Information Needed]
### Results
[More Information Needed]
#### Summary
## Model Examination [optional]
<!-- Relevant interpretability work for the model goes here -->
[More Information Needed]
## Environmental Impact
<!-- Total emissions (in grams of CO2eq) and additional considerations, such as electricity usage, go here. Edit the suggested text below accordingly -->
Carbon emissions can be estimated using the [Machine Learning Impact calculator](https://mlco2.github.io/impact#compute) presented in [Lacoste et al. (2019)](https://arxiv.org/abs/1910.09700).
- **Hardware Type:** [More Information Needed]
- **Hours used:** [More Information Needed]
- **Cloud Provider:** [More Information Needed]
- **Compute Region:** [More Information Needed]
- **Carbon Emitted:** [More Information Needed]
## Technical Specifications [optional]
### Model Architecture and Objective
[More Information Needed]
### Compute Infrastructure
[More Information Needed]
#### Hardware
[More Information Needed]
#### Software
[More Information Needed]
## Citation [optional]
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
**BibTeX:**
[More Information Needed]
**APA:**
[More Information Needed]
## Glossary [optional]
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
[More Information Needed]
## More Information [optional]
[More Information Needed]
## Model Card Authors [optional]
[More Information Needed]
## Model Card Contact
[More Information Needed]
### Framework versions
- PEFT 0.15.2
|
ypszn/blockassist-bc-yapping_pawing_worm_1756502749
|
ypszn
| 2025-08-29T21:26:40Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"yapping pawing worm",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T21:26:37Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- yapping pawing worm
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
AnonymousCS/populism_classifier_bsample_218
|
AnonymousCS
| 2025-08-29T21:25:10Z | 0 | 0 |
transformers
|
[
"transformers",
"safetensors",
"xlm-roberta",
"text-classification",
"generated_from_trainer",
"base_model:AnonymousCS/populism_xlmr_large",
"base_model:finetune:AnonymousCS/populism_xlmr_large",
"license:mit",
"autotrain_compatible",
"endpoints_compatible",
"region:us"
] |
text-classification
| 2025-08-29T21:21:24Z |
---
library_name: transformers
license: mit
base_model: AnonymousCS/populism_xlmr_large
tags:
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: populism_classifier_bsample_218
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. -->
# populism_classifier_bsample_218
This model is a fine-tuned version of [AnonymousCS/populism_xlmr_large](https://huggingface.co/AnonymousCS/populism_xlmr_large) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.5674
- Accuracy: 0.9713
- 1-f1: 0.0
- 1-recall: 0.0
- 1-precision: 0.0
- Balanced Acc: 0.5
## 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-06
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.06
- num_epochs: 15
- mixed_precision_training: Native AMP
### Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | 1-f1 | 1-recall | 1-precision | Balanced Acc |
|:-------------:|:-----:|:----:|:---------------:|:--------:|:----:|:--------:|:-----------:|:------------:|
| 0.7654 | 1.0 | 17 | 0.4507 | 0.9713 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.769 | 2.0 | 34 | 0.4912 | 0.9713 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.627 | 3.0 | 51 | 0.5235 | 0.9713 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.6791 | 4.0 | 68 | 0.5460 | 0.9713 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.7041 | 5.0 | 85 | 0.5592 | 0.9713 | 0.0 | 0.0 | 0.0 | 0.5 |
| 0.6558 | 6.0 | 102 | 0.5674 | 0.9713 | 0.0 | 0.0 | 0.0 | 0.5 |
### Framework versions
- Transformers 4.46.3
- Pytorch 2.4.1+cu121
- Datasets 3.1.0
- Tokenizers 0.20.3
|
Samuell43/blockassist-bc-fast_gregarious_warthog_1756502171
|
Samuell43
| 2025-08-29T21:17:36Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"fast gregarious warthog",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T21:16:42Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- fast gregarious warthog
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
klmdr22/blockassist-bc-wild_loud_newt_1756502057
|
klmdr22
| 2025-08-29T21:15:00Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"wild loud newt",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T21:14:56Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- wild loud newt
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
duyntnet/Aethora-7b-v1-imatrix-GGUF
|
duyntnet
| 2025-08-29T21:08:24Z | 0 | 0 |
transformers
|
[
"transformers",
"gguf",
"imatrix",
"Aethora-7b-v1",
"text-generation",
"en",
"license:other",
"region:us",
"conversational"
] |
text-generation
| 2025-08-29T20:18:09Z |
---
license: other
language:
- en
pipeline_tag: text-generation
inference: false
tags:
- transformers
- gguf
- imatrix
- Aethora-7b-v1
---
Quantizations of https://huggingface.co/SteelStorage/Aethora-7b-v1
### Open source inference clients/UIs
* [llama.cpp](https://github.com/ggerganov/llama.cpp)
* [KoboldCPP](https://github.com/LostRuins/koboldcpp)
* [text-generation-webui](https://github.com/oobabooga/text-generation-webui)
* [jan](https://github.com/menloresearch/jan)
* [ik_llama.cpp](https://github.com/ikawrakow/ik_llama.cpp)
* [croco.cpp](https://github.com/Nexesenex/croco.cpp)
### Closed source inference clients/UIs
* [LM Studio](https://lmstudio.ai/)
* More will be added...
---
# From original readme
Aethora-7B-V1
=============
**Creator:** [SteelSkull](https://huggingface.co/Steelskull)
**About Aethora:** Trained on 2 Full Epochs of Aethora-7b-V1 using Aether-V1.9 Dataset, Aethora is a model trained specifically for general use with a focus in RP/Story based on the 2.5mil row (around 1 billion tokens) Aether dataset.
**Model Quants:** Quants provided by: \[N/A\] .
**Model Sources:**
* Developed & Funded by: [Steelskull](https://huggingface.co/Steelskull)
* Finetuned from model: [Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2)
* Finetuning Repository: [Aether Dataset](https://huggingface.co/datasets/TheSkullery/Aether-V1.9)
* Model type: BF16
* License: A2
**Finetune Information:**
* Hardware Type: H100 x1
* Hours Used: 60-Hrs
* Cloud Provider: [Runpod.io](https://www.runpod.io)
* Compute Region: US-IL
**Dataset Information:**
* Version v1.9: Fixed an error where 'system' and 'tools' records were not being carried over to the final dataframe. Added an 'origins' record for dataset sources.
* Version 1.8.5: Removed missing conversations or starting messages that are empty, and selectively omitted certain phrases for coherence and relevance.
**Datasets Used:**
* grimulkan/bluemoon\_Karen\_cleaned
* Doctor-Shotgun/no-robots-sharegpt
* Locutusque/Hercules-v3.0
* jondurbin/airoboros-3.2
* openerotica/freedom-rp
* teknium/OpenHermes-2.5
* Doctor-Shotgun/capybara-sharegpt
* KaraKaraWitch/PIPPA-ShareGPT-formatted
* Locutusque/bagel-clean-v0.3-shuffled
* Locutusque/hyperion-v3.0
**Dataset Summary (Processed / Removed):**
* Total Objects Removed: **209074**
* Deduplication Stats: Starting row count: 4738917, Final row count: 2673175, Rows removed: **2065742**
|
rvipitkirubbe/blockassist-bc-mottled_foraging_ape_1756500004
|
rvipitkirubbe
| 2025-08-29T21:05:47Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"mottled foraging ape",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T21:05:43Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- mottled foraging ape
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
GroomerG/blockassist-bc-vicious_pawing_badger_1756500009
|
GroomerG
| 2025-08-29T21:04:47Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"vicious pawing badger",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T21:04:42Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- vicious pawing badger
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
ORIGINAL-VAZADO-VIDEO-DO-SURFISTA/FULL.VAZADO.VIDEO.DO.SURFISTA.CLIP
|
ORIGINAL-VAZADO-VIDEO-DO-SURFISTA
| 2025-08-29T20:58:35Z | 0 | 0 | null |
[
"region:us"
] | null | 2025-08-29T20:58:06Z |
<animated-image data-catalyst=""><a href="https://tinyurl.com/5xr5mb3e?leaked-videos/" rel="nofollow" data-target="animated-image.originalLink"><img src="https://static.wixstatic.com/media/b249f9_adac8f70fb3f45b88691696c77de18f3~mv2.gif" alt="Foo" data-canonical-src="https://static.wixstatic.com/media/b249f9_adac8f70fb3f45b88691696c77de18f3~mv2.gif" style="max-width: 100%; display: inline-block;" data-target="animated-image.originalImage"></a>
|
gensynw/blockassist-bc-deadly_sharp_albatross_1756500980
|
gensynw
| 2025-08-29T20:57:15Z | 0 | 0 | null |
[
"gensyn",
"blockassist",
"gensyn-blockassist",
"minecraft",
"deadly sharp albatross",
"arxiv:2504.07091",
"region:us"
] | null | 2025-08-29T20:56:23Z |
---
tags:
- gensyn
- blockassist
- gensyn-blockassist
- minecraft
- deadly sharp albatross
---
# Gensyn BlockAssist
Gensyn's BlockAssist is a distributed extension of the paper [AssistanceZero: Scalably Solving Assistance Games](https://arxiv.org/abs/2504.07091).
|
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.