Model save
Browse files- README.md +74 -0
- all_results.json +9 -0
- generation_config.json +8 -0
- train_results.json +9 -0
- trainer_state.json +1423 -0
README.md
ADDED
|
@@ -0,0 +1,74 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
license: gemma
|
| 4 |
+
base_model: google/gemma-2-9b-it
|
| 5 |
+
tags:
|
| 6 |
+
- trl
|
| 7 |
+
- simpo
|
| 8 |
+
- generated_from_trainer
|
| 9 |
+
model-index:
|
| 10 |
+
- name: simpo-train-filtered-full
|
| 11 |
+
results: []
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 15 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 16 |
+
|
| 17 |
+
# simpo-train-filtered-full
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [google/gemma-2-9b-it](https://huggingface.co/google/gemma-2-9b-it) on an unknown dataset.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 2.7628
|
| 22 |
+
- Rewards/chosen: -19.4831
|
| 23 |
+
- Rewards/rejected: -24.7305
|
| 24 |
+
- Rewards/accuracies: 0.7807
|
| 25 |
+
- Rewards/margins: 5.2474
|
| 26 |
+
- Logps/rejected: -2.4731
|
| 27 |
+
- Logps/chosen: -1.9483
|
| 28 |
+
- Logits/rejected: -13.6040
|
| 29 |
+
- Logits/chosen: -13.5429
|
| 30 |
+
|
| 31 |
+
## Model description
|
| 32 |
+
|
| 33 |
+
More information needed
|
| 34 |
+
|
| 35 |
+
## Intended uses & limitations
|
| 36 |
+
|
| 37 |
+
More information needed
|
| 38 |
+
|
| 39 |
+
## Training and evaluation data
|
| 40 |
+
|
| 41 |
+
More information needed
|
| 42 |
+
|
| 43 |
+
## Training procedure
|
| 44 |
+
|
| 45 |
+
### Training hyperparameters
|
| 46 |
+
|
| 47 |
+
The following hyperparameters were used during training:
|
| 48 |
+
- learning_rate: 8e-07
|
| 49 |
+
- train_batch_size: 2
|
| 50 |
+
- eval_batch_size: 4
|
| 51 |
+
- seed: 42
|
| 52 |
+
- distributed_type: multi-GPU
|
| 53 |
+
- num_devices: 4
|
| 54 |
+
- gradient_accumulation_steps: 16
|
| 55 |
+
- total_train_batch_size: 128
|
| 56 |
+
- total_eval_batch_size: 16
|
| 57 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 58 |
+
- lr_scheduler_type: cosine
|
| 59 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 60 |
+
- num_epochs: 1
|
| 61 |
+
|
| 62 |
+
### Training results
|
| 63 |
+
|
| 64 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 65 |
+
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 66 |
+
| 2.7931 | 0.8724 | 400 | 2.7628 | -19.4831 | -24.7305 | 0.7807 | 5.2474 | -2.4731 | -1.9483 | -13.6040 | -13.5429 |
|
| 67 |
+
|
| 68 |
+
|
| 69 |
+
### Framework versions
|
| 70 |
+
|
| 71 |
+
- Transformers 4.44.2
|
| 72 |
+
- Pytorch 2.7.0+cu128
|
| 73 |
+
- Datasets 2.18.0
|
| 74 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9989094874591058,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 3.523275049492782,
|
| 5 |
+
"train_runtime": 6692.3835,
|
| 6 |
+
"train_samples": 58684,
|
| 7 |
+
"train_samples_per_second": 8.769,
|
| 8 |
+
"train_steps_per_second": 0.068
|
| 9 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 2,
|
| 4 |
+
"cache_implementation": "hybrid",
|
| 5 |
+
"eos_token_id": 1,
|
| 6 |
+
"pad_token_id": 0,
|
| 7 |
+
"transformers_version": "4.44.2"
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9989094874591058,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 3.523275049492782,
|
| 5 |
+
"train_runtime": 6692.3835,
|
| 6 |
+
"train_samples": 58684,
|
| 7 |
+
"train_samples_per_second": 8.769,
|
| 8 |
+
"train_steps_per_second": 0.068
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1423 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9989094874591058,
|
| 5 |
+
"eval_steps": 400,
|
| 6 |
+
"global_step": 458,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.010905125408942203,
|
| 13 |
+
"grad_norm": 64.1503528533801,
|
| 14 |
+
"learning_rate": 8.695652173913042e-08,
|
| 15 |
+
"logits/chosen": -10.94892692565918,
|
| 16 |
+
"logits/rejected": -10.934142112731934,
|
| 17 |
+
"logps/chosen": -0.8995465040206909,
|
| 18 |
+
"logps/rejected": -0.8591210246086121,
|
| 19 |
+
"loss": 5.7639,
|
| 20 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 21 |
+
"rewards/chosen": -8.995465278625488,
|
| 22 |
+
"rewards/margins": -0.40425485372543335,
|
| 23 |
+
"rewards/rejected": -8.59121036529541,
|
| 24 |
+
"step": 5
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.021810250817884406,
|
| 28 |
+
"grad_norm": 111.15906224445196,
|
| 29 |
+
"learning_rate": 1.7391304347826085e-07,
|
| 30 |
+
"logits/chosen": -9.873212814331055,
|
| 31 |
+
"logits/rejected": -9.899297714233398,
|
| 32 |
+
"logps/chosen": -1.1384799480438232,
|
| 33 |
+
"logps/rejected": -1.158919095993042,
|
| 34 |
+
"loss": 6.1082,
|
| 35 |
+
"rewards/accuracies": 0.518750011920929,
|
| 36 |
+
"rewards/chosen": -11.384800910949707,
|
| 37 |
+
"rewards/margins": 0.20439176261425018,
|
| 38 |
+
"rewards/rejected": -11.589191436767578,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.03271537622682661,
|
| 43 |
+
"grad_norm": 96.8211465004871,
|
| 44 |
+
"learning_rate": 2.608695652173913e-07,
|
| 45 |
+
"logits/chosen": -10.422977447509766,
|
| 46 |
+
"logits/rejected": -10.321990013122559,
|
| 47 |
+
"logps/chosen": -0.8700820803642273,
|
| 48 |
+
"logps/rejected": -0.840315043926239,
|
| 49 |
+
"loss": 5.5576,
|
| 50 |
+
"rewards/accuracies": 0.48124998807907104,
|
| 51 |
+
"rewards/chosen": -8.700819969177246,
|
| 52 |
+
"rewards/margins": -0.29766958951950073,
|
| 53 |
+
"rewards/rejected": -8.40315055847168,
|
| 54 |
+
"step": 15
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.04362050163576881,
|
| 58 |
+
"grad_norm": 59.01122752402013,
|
| 59 |
+
"learning_rate": 3.478260869565217e-07,
|
| 60 |
+
"logits/chosen": -9.897313117980957,
|
| 61 |
+
"logits/rejected": -9.771203994750977,
|
| 62 |
+
"logps/chosen": -0.9016021490097046,
|
| 63 |
+
"logps/rejected": -0.9813264608383179,
|
| 64 |
+
"loss": 5.4183,
|
| 65 |
+
"rewards/accuracies": 0.574999988079071,
|
| 66 |
+
"rewards/chosen": -9.016023635864258,
|
| 67 |
+
"rewards/margins": 0.797242283821106,
|
| 68 |
+
"rewards/rejected": -9.813264846801758,
|
| 69 |
+
"step": 20
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.05452562704471101,
|
| 73 |
+
"grad_norm": 77.6935326260568,
|
| 74 |
+
"learning_rate": 4.3478260869565214e-07,
|
| 75 |
+
"logits/chosen": -10.004715919494629,
|
| 76 |
+
"logits/rejected": -10.082869529724121,
|
| 77 |
+
"logps/chosen": -0.7554069757461548,
|
| 78 |
+
"logps/rejected": -0.7040703296661377,
|
| 79 |
+
"loss": 5.4042,
|
| 80 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 81 |
+
"rewards/chosen": -7.554069519042969,
|
| 82 |
+
"rewards/margins": -0.5133668780326843,
|
| 83 |
+
"rewards/rejected": -7.040703773498535,
|
| 84 |
+
"step": 25
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.06543075245365322,
|
| 88 |
+
"grad_norm": 53.25393883515416,
|
| 89 |
+
"learning_rate": 5.217391304347826e-07,
|
| 90 |
+
"logits/chosen": -9.147699356079102,
|
| 91 |
+
"logits/rejected": -9.047504425048828,
|
| 92 |
+
"logps/chosen": -0.6732769012451172,
|
| 93 |
+
"logps/rejected": -0.7553219199180603,
|
| 94 |
+
"loss": 5.079,
|
| 95 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 96 |
+
"rewards/chosen": -6.732769012451172,
|
| 97 |
+
"rewards/margins": 0.8204498291015625,
|
| 98 |
+
"rewards/rejected": -7.553219795227051,
|
| 99 |
+
"step": 30
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.07633587786259542,
|
| 103 |
+
"grad_norm": 49.376076232037846,
|
| 104 |
+
"learning_rate": 6.08695652173913e-07,
|
| 105 |
+
"logits/chosen": -9.227429389953613,
|
| 106 |
+
"logits/rejected": -9.0322847366333,
|
| 107 |
+
"logps/chosen": -0.6311928033828735,
|
| 108 |
+
"logps/rejected": -0.6472190022468567,
|
| 109 |
+
"loss": 4.9879,
|
| 110 |
+
"rewards/accuracies": 0.543749988079071,
|
| 111 |
+
"rewards/chosen": -6.311927795410156,
|
| 112 |
+
"rewards/margins": 0.160261332988739,
|
| 113 |
+
"rewards/rejected": -6.472189426422119,
|
| 114 |
+
"step": 35
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.08724100327153762,
|
| 118 |
+
"grad_norm": 71.82957946446516,
|
| 119 |
+
"learning_rate": 6.956521739130434e-07,
|
| 120 |
+
"logits/chosen": -8.839519500732422,
|
| 121 |
+
"logits/rejected": -8.598153114318848,
|
| 122 |
+
"logps/chosen": -0.5917301177978516,
|
| 123 |
+
"logps/rejected": -0.6095324158668518,
|
| 124 |
+
"loss": 4.9341,
|
| 125 |
+
"rewards/accuracies": 0.543749988079071,
|
| 126 |
+
"rewards/chosen": -5.917301177978516,
|
| 127 |
+
"rewards/margins": 0.1780225783586502,
|
| 128 |
+
"rewards/rejected": -6.09532356262207,
|
| 129 |
+
"step": 40
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.09814612868047982,
|
| 133 |
+
"grad_norm": 62.13252777111053,
|
| 134 |
+
"learning_rate": 7.826086956521739e-07,
|
| 135 |
+
"logits/chosen": -8.288870811462402,
|
| 136 |
+
"logits/rejected": -8.250112533569336,
|
| 137 |
+
"logps/chosen": -0.5544935464859009,
|
| 138 |
+
"logps/rejected": -0.5844836235046387,
|
| 139 |
+
"loss": 4.6875,
|
| 140 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 141 |
+
"rewards/chosen": -5.5449347496032715,
|
| 142 |
+
"rewards/margins": 0.29990124702453613,
|
| 143 |
+
"rewards/rejected": -5.844836235046387,
|
| 144 |
+
"step": 45
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.10905125408942203,
|
| 148 |
+
"grad_norm": 50.37809623047656,
|
| 149 |
+
"learning_rate": 7.998139534493406e-07,
|
| 150 |
+
"logits/chosen": -8.140437126159668,
|
| 151 |
+
"logits/rejected": -7.910015106201172,
|
| 152 |
+
"logps/chosen": -0.638617753982544,
|
| 153 |
+
"logps/rejected": -0.7139040231704712,
|
| 154 |
+
"loss": 4.5683,
|
| 155 |
+
"rewards/accuracies": 0.574999988079071,
|
| 156 |
+
"rewards/chosen": -6.386178493499756,
|
| 157 |
+
"rewards/margins": 0.7528623938560486,
|
| 158 |
+
"rewards/rejected": -7.139040470123291,
|
| 159 |
+
"step": 50
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.11995637949836423,
|
| 163 |
+
"grad_norm": 61.0124851348064,
|
| 164 |
+
"learning_rate": 7.990584359406726e-07,
|
| 165 |
+
"logits/chosen": -9.38193416595459,
|
| 166 |
+
"logits/rejected": -9.200593948364258,
|
| 167 |
+
"logps/chosen": -0.6021402478218079,
|
| 168 |
+
"logps/rejected": -0.7378085255622864,
|
| 169 |
+
"loss": 4.6702,
|
| 170 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 171 |
+
"rewards/chosen": -6.021402359008789,
|
| 172 |
+
"rewards/margins": 1.3566832542419434,
|
| 173 |
+
"rewards/rejected": -7.378085136413574,
|
| 174 |
+
"step": 55
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.13086150490730644,
|
| 178 |
+
"grad_norm": 58.34041763150138,
|
| 179 |
+
"learning_rate": 7.977229168076815e-07,
|
| 180 |
+
"logits/chosen": -8.994357109069824,
|
| 181 |
+
"logits/rejected": -9.095643043518066,
|
| 182 |
+
"logps/chosen": -0.6254938244819641,
|
| 183 |
+
"logps/rejected": -0.7731696963310242,
|
| 184 |
+
"loss": 4.4672,
|
| 185 |
+
"rewards/accuracies": 0.606249988079071,
|
| 186 |
+
"rewards/chosen": -6.254937648773193,
|
| 187 |
+
"rewards/margins": 1.4767591953277588,
|
| 188 |
+
"rewards/rejected": -7.731697082519531,
|
| 189 |
+
"step": 60
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.14176663031624864,
|
| 193 |
+
"grad_norm": 54.06832565592651,
|
| 194 |
+
"learning_rate": 7.95809337127655e-07,
|
| 195 |
+
"logits/chosen": -9.56706428527832,
|
| 196 |
+
"logits/rejected": -9.66978931427002,
|
| 197 |
+
"logps/chosen": -0.6539489030838013,
|
| 198 |
+
"logps/rejected": -0.7899436950683594,
|
| 199 |
+
"loss": 4.4539,
|
| 200 |
+
"rewards/accuracies": 0.625,
|
| 201 |
+
"rewards/chosen": -6.539488315582275,
|
| 202 |
+
"rewards/margins": 1.359948992729187,
|
| 203 |
+
"rewards/rejected": -7.899436950683594,
|
| 204 |
+
"step": 65
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.15267175572519084,
|
| 208 |
+
"grad_norm": 61.12821185450308,
|
| 209 |
+
"learning_rate": 7.933204781457008e-07,
|
| 210 |
+
"logits/chosen": -9.499263763427734,
|
| 211 |
+
"logits/rejected": -9.496051788330078,
|
| 212 |
+
"logps/chosen": -0.6623054146766663,
|
| 213 |
+
"logps/rejected": -0.7747452259063721,
|
| 214 |
+
"loss": 4.4992,
|
| 215 |
+
"rewards/accuracies": 0.65625,
|
| 216 |
+
"rewards/chosen": -6.623053550720215,
|
| 217 |
+
"rewards/margins": 1.1243983507156372,
|
| 218 |
+
"rewards/rejected": -7.747453212738037,
|
| 219 |
+
"step": 70
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.16357688113413305,
|
| 223 |
+
"grad_norm": 51.861021015777716,
|
| 224 |
+
"learning_rate": 7.902599572324151e-07,
|
| 225 |
+
"logits/chosen": -9.391631126403809,
|
| 226 |
+
"logits/rejected": -9.375069618225098,
|
| 227 |
+
"logps/chosen": -0.6535338163375854,
|
| 228 |
+
"logps/rejected": -0.8060415983200073,
|
| 229 |
+
"loss": 4.4979,
|
| 230 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 231 |
+
"rewards/chosen": -6.535338401794434,
|
| 232 |
+
"rewards/margins": 1.5250780582427979,
|
| 233 |
+
"rewards/rejected": -8.060416221618652,
|
| 234 |
+
"step": 75
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.17448200654307525,
|
| 238 |
+
"grad_norm": 70.13525940500534,
|
| 239 |
+
"learning_rate": 7.86632222626304e-07,
|
| 240 |
+
"logits/chosen": -9.956920623779297,
|
| 241 |
+
"logits/rejected": -9.78042984008789,
|
| 242 |
+
"logps/chosen": -0.6164920926094055,
|
| 243 |
+
"logps/rejected": -0.7590925097465515,
|
| 244 |
+
"loss": 4.429,
|
| 245 |
+
"rewards/accuracies": 0.65625,
|
| 246 |
+
"rewards/chosen": -6.164921760559082,
|
| 247 |
+
"rewards/margins": 1.4260046482086182,
|
| 248 |
+
"rewards/rejected": -7.590925693511963,
|
| 249 |
+
"step": 80
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.18538713195201745,
|
| 253 |
+
"grad_norm": 73.7268648843484,
|
| 254 |
+
"learning_rate": 7.824425469686014e-07,
|
| 255 |
+
"logits/chosen": -9.828500747680664,
|
| 256 |
+
"logits/rejected": -9.678921699523926,
|
| 257 |
+
"logps/chosen": -0.6361075043678284,
|
| 258 |
+
"logps/rejected": -0.7591807246208191,
|
| 259 |
+
"loss": 4.438,
|
| 260 |
+
"rewards/accuracies": 0.65625,
|
| 261 |
+
"rewards/chosen": -6.361075401306152,
|
| 262 |
+
"rewards/margins": 1.2307318449020386,
|
| 263 |
+
"rewards/rejected": -7.591805934906006,
|
| 264 |
+
"step": 85
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.19629225736095965,
|
| 268 |
+
"grad_norm": 69.62622029011087,
|
| 269 |
+
"learning_rate": 7.776970196398794e-07,
|
| 270 |
+
"logits/chosen": -9.843175888061523,
|
| 271 |
+
"logits/rejected": -9.792081832885742,
|
| 272 |
+
"logps/chosen": -0.6243542432785034,
|
| 273 |
+
"logps/rejected": -0.7286983132362366,
|
| 274 |
+
"loss": 4.3737,
|
| 275 |
+
"rewards/accuracies": 0.668749988079071,
|
| 276 |
+
"rewards/chosen": -6.2435431480407715,
|
| 277 |
+
"rewards/margins": 1.0434401035308838,
|
| 278 |
+
"rewards/rejected": -7.286982536315918,
|
| 279 |
+
"step": 90
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.20719738276990185,
|
| 283 |
+
"grad_norm": 66.25031731585119,
|
| 284 |
+
"learning_rate": 7.724025379095878e-07,
|
| 285 |
+
"logits/chosen": -10.194116592407227,
|
| 286 |
+
"logits/rejected": -10.006689071655273,
|
| 287 |
+
"logps/chosen": -0.6633736491203308,
|
| 288 |
+
"logps/rejected": -0.85277259349823,
|
| 289 |
+
"loss": 4.2355,
|
| 290 |
+
"rewards/accuracies": 0.637499988079071,
|
| 291 |
+
"rewards/chosen": -6.633737087249756,
|
| 292 |
+
"rewards/margins": 1.8939902782440186,
|
| 293 |
+
"rewards/rejected": -8.527726173400879,
|
| 294 |
+
"step": 95
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.21810250817884405,
|
| 298 |
+
"grad_norm": 75.08429330907302,
|
| 299 |
+
"learning_rate": 7.665667969113885e-07,
|
| 300 |
+
"logits/chosen": -9.845739364624023,
|
| 301 |
+
"logits/rejected": -9.587841033935547,
|
| 302 |
+
"logps/chosen": -0.6887632608413696,
|
| 303 |
+
"logps/rejected": -0.872734546661377,
|
| 304 |
+
"loss": 4.2625,
|
| 305 |
+
"rewards/accuracies": 0.668749988079071,
|
| 306 |
+
"rewards/chosen": -6.887633323669434,
|
| 307 |
+
"rewards/margins": 1.839712381362915,
|
| 308 |
+
"rewards/rejected": -8.72734546661377,
|
| 309 |
+
"step": 100
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.22900763358778625,
|
| 313 |
+
"grad_norm": 69.02863249400193,
|
| 314 |
+
"learning_rate": 7.601982784588541e-07,
|
| 315 |
+
"logits/chosen": -10.62452220916748,
|
| 316 |
+
"logits/rejected": -10.43299674987793,
|
| 317 |
+
"logps/chosen": -0.7019127011299133,
|
| 318 |
+
"logps/rejected": -0.7768973112106323,
|
| 319 |
+
"loss": 4.2785,
|
| 320 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 321 |
+
"rewards/chosen": -7.019126892089844,
|
| 322 |
+
"rewards/margins": 0.7498467564582825,
|
| 323 |
+
"rewards/rejected": -7.768974304199219,
|
| 324 |
+
"step": 105
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.23991275899672845,
|
| 328 |
+
"grad_norm": 71.91219250814,
|
| 329 |
+
"learning_rate": 7.533062387177842e-07,
|
| 330 |
+
"logits/chosen": -10.65683650970459,
|
| 331 |
+
"logits/rejected": -10.636886596679688,
|
| 332 |
+
"logps/chosen": -0.7213658094406128,
|
| 333 |
+
"logps/rejected": -0.8907195329666138,
|
| 334 |
+
"loss": 4.1753,
|
| 335 |
+
"rewards/accuracies": 0.6875,
|
| 336 |
+
"rewards/chosen": -7.213658332824707,
|
| 337 |
+
"rewards/margins": 1.6935373544692993,
|
| 338 |
+
"rewards/rejected": -8.907196044921875,
|
| 339 |
+
"step": 110
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.25081788440567065,
|
| 343 |
+
"grad_norm": 73.69796183664997,
|
| 344 |
+
"learning_rate": 7.459006947530617e-07,
|
| 345 |
+
"logits/chosen": -11.56556224822998,
|
| 346 |
+
"logits/rejected": -11.49560260772705,
|
| 347 |
+
"logps/chosen": -0.838672935962677,
|
| 348 |
+
"logps/rejected": -1.0182416439056396,
|
| 349 |
+
"loss": 4.0384,
|
| 350 |
+
"rewards/accuracies": 0.668749988079071,
|
| 351 |
+
"rewards/chosen": -8.38672924041748,
|
| 352 |
+
"rewards/margins": 1.7956879138946533,
|
| 353 |
+
"rewards/rejected": -10.182416915893555,
|
| 354 |
+
"step": 115
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.2617230098146129,
|
| 358 |
+
"grad_norm": 94.92610924652415,
|
| 359 |
+
"learning_rate": 7.379924099695959e-07,
|
| 360 |
+
"logits/chosen": -11.925847053527832,
|
| 361 |
+
"logits/rejected": -11.710810661315918,
|
| 362 |
+
"logps/chosen": -0.8436716198921204,
|
| 363 |
+
"logps/rejected": -1.0012091398239136,
|
| 364 |
+
"loss": 4.113,
|
| 365 |
+
"rewards/accuracies": 0.675000011920929,
|
| 366 |
+
"rewards/chosen": -8.436716079711914,
|
| 367 |
+
"rewards/margins": 1.5753746032714844,
|
| 368 |
+
"rewards/rejected": -10.012091636657715,
|
| 369 |
+
"step": 120
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.27262813522355506,
|
| 373 |
+
"grad_norm": 113.62807962706827,
|
| 374 |
+
"learning_rate": 7.29592878468518e-07,
|
| 375 |
+
"logits/chosen": -13.345464706420898,
|
| 376 |
+
"logits/rejected": -12.9557523727417,
|
| 377 |
+
"logps/chosen": -0.8897609710693359,
|
| 378 |
+
"logps/rejected": -1.0490643978118896,
|
| 379 |
+
"loss": 3.9421,
|
| 380 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 381 |
+
"rewards/chosen": -8.897610664367676,
|
| 382 |
+
"rewards/margins": 1.5930334329605103,
|
| 383 |
+
"rewards/rejected": -10.490643501281738,
|
| 384 |
+
"step": 125
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.2835332606324973,
|
| 388 |
+
"grad_norm": 118.6845458542213,
|
| 389 |
+
"learning_rate": 7.207143083413642e-07,
|
| 390 |
+
"logits/chosen": -12.795938491821289,
|
| 391 |
+
"logits/rejected": -12.771161079406738,
|
| 392 |
+
"logps/chosen": -0.9606706500053406,
|
| 393 |
+
"logps/rejected": -1.1983877420425415,
|
| 394 |
+
"loss": 3.7793,
|
| 395 |
+
"rewards/accuracies": 0.706250011920929,
|
| 396 |
+
"rewards/chosen": -9.606706619262695,
|
| 397 |
+
"rewards/margins": 2.3771698474884033,
|
| 398 |
+
"rewards/rejected": -11.983877182006836,
|
| 399 |
+
"step": 130
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.29443838604143946,
|
| 403 |
+
"grad_norm": 112.00779108354124,
|
| 404 |
+
"learning_rate": 7.11369603926526e-07,
|
| 405 |
+
"logits/chosen": -13.919995307922363,
|
| 406 |
+
"logits/rejected": -14.025039672851562,
|
| 407 |
+
"logps/chosen": -1.1236939430236816,
|
| 408 |
+
"logps/rejected": -1.3567845821380615,
|
| 409 |
+
"loss": 3.8057,
|
| 410 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 411 |
+
"rewards/chosen": -11.236939430236816,
|
| 412 |
+
"rewards/margins": 2.3309078216552734,
|
| 413 |
+
"rewards/rejected": -13.567845344543457,
|
| 414 |
+
"step": 135
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.3053435114503817,
|
| 418 |
+
"grad_norm": 121.41466704799392,
|
| 419 |
+
"learning_rate": 7.015723470537589e-07,
|
| 420 |
+
"logits/chosen": -14.653108596801758,
|
| 421 |
+
"logits/rejected": -14.910409927368164,
|
| 422 |
+
"logps/chosen": -1.2697908878326416,
|
| 423 |
+
"logps/rejected": -1.5090426206588745,
|
| 424 |
+
"loss": 3.7783,
|
| 425 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 426 |
+
"rewards/chosen": -12.69791030883789,
|
| 427 |
+
"rewards/margins": 2.392517566680908,
|
| 428 |
+
"rewards/rejected": -15.090426445007324,
|
| 429 |
+
"step": 140
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.31624863685932386,
|
| 433 |
+
"grad_norm": 118.0007773777601,
|
| 434 |
+
"learning_rate": 6.913367773040073e-07,
|
| 435 |
+
"logits/chosen": -15.646504402160645,
|
| 436 |
+
"logits/rejected": -15.359659194946289,
|
| 437 |
+
"logps/chosen": -1.2195041179656982,
|
| 438 |
+
"logps/rejected": -1.5088005065917969,
|
| 439 |
+
"loss": 3.626,
|
| 440 |
+
"rewards/accuracies": 0.71875,
|
| 441 |
+
"rewards/chosen": -12.195042610168457,
|
| 442 |
+
"rewards/margins": 2.892961263656616,
|
| 443 |
+
"rewards/rejected": -15.088003158569336,
|
| 444 |
+
"step": 145
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.3271537622682661,
|
| 448 |
+
"grad_norm": 123.35286255999112,
|
| 449 |
+
"learning_rate": 6.806777713132373e-07,
|
| 450 |
+
"logits/chosen": -15.674619674682617,
|
| 451 |
+
"logits/rejected": -15.5663423538208,
|
| 452 |
+
"logps/chosen": -1.3831182718276978,
|
| 453 |
+
"logps/rejected": -1.60293710231781,
|
| 454 |
+
"loss": 3.76,
|
| 455 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 456 |
+
"rewards/chosen": -13.831182479858398,
|
| 457 |
+
"rewards/margins": 2.1981873512268066,
|
| 458 |
+
"rewards/rejected": -16.029369354248047,
|
| 459 |
+
"step": 150
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.33805888767720826,
|
| 463 |
+
"grad_norm": 131.11270501363964,
|
| 464 |
+
"learning_rate": 6.69610821150358e-07,
|
| 465 |
+
"logits/chosen": -15.65868854522705,
|
| 466 |
+
"logits/rejected": -15.700967788696289,
|
| 467 |
+
"logps/chosen": -1.3463469743728638,
|
| 468 |
+
"logps/rejected": -1.6862995624542236,
|
| 469 |
+
"loss": 3.4489,
|
| 470 |
+
"rewards/accuracies": 0.699999988079071,
|
| 471 |
+
"rewards/chosen": -13.463468551635742,
|
| 472 |
+
"rewards/margins": 3.3995232582092285,
|
| 473 |
+
"rewards/rejected": -16.862991333007812,
|
| 474 |
+
"step": 155
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.3489640130861505,
|
| 478 |
+
"grad_norm": 135.1222468933701,
|
| 479 |
+
"learning_rate": 6.581520118006564e-07,
|
| 480 |
+
"logits/chosen": -15.956563949584961,
|
| 481 |
+
"logits/rejected": -16.274681091308594,
|
| 482 |
+
"logps/chosen": -1.4811228513717651,
|
| 483 |
+
"logps/rejected": -1.8506397008895874,
|
| 484 |
+
"loss": 3.4319,
|
| 485 |
+
"rewards/accuracies": 0.737500011920929,
|
| 486 |
+
"rewards/chosen": -14.81122875213623,
|
| 487 |
+
"rewards/margins": 3.695168972015381,
|
| 488 |
+
"rewards/rejected": -18.506397247314453,
|
| 489 |
+
"step": 160
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.35986913849509267,
|
| 493 |
+
"grad_norm": 117.31734546028652,
|
| 494 |
+
"learning_rate": 6.46317997787473e-07,
|
| 495 |
+
"logits/chosen": -16.915264129638672,
|
| 496 |
+
"logits/rejected": -17.207149505615234,
|
| 497 |
+
"logps/chosen": -1.4308239221572876,
|
| 498 |
+
"logps/rejected": -1.7504551410675049,
|
| 499 |
+
"loss": 3.7315,
|
| 500 |
+
"rewards/accuracies": 0.699999988079071,
|
| 501 |
+
"rewards/chosen": -14.30823802947998,
|
| 502 |
+
"rewards/margins": 3.1963133811950684,
|
| 503 |
+
"rewards/rejected": -17.504552841186523,
|
| 504 |
+
"step": 165
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 0.3707742639040349,
|
| 508 |
+
"grad_norm": 135.9602274797816,
|
| 509 |
+
"learning_rate": 6.341259789660969e-07,
|
| 510 |
+
"logits/chosen": -16.72784996032715,
|
| 511 |
+
"logits/rejected": -16.691822052001953,
|
| 512 |
+
"logps/chosen": -1.5182801485061646,
|
| 513 |
+
"logps/rejected": -1.922546148300171,
|
| 514 |
+
"loss": 3.3263,
|
| 515 |
+
"rewards/accuracies": 0.737500011920929,
|
| 516 |
+
"rewards/chosen": -15.182802200317383,
|
| 517 |
+
"rewards/margins": 4.042659759521484,
|
| 518 |
+
"rewards/rejected": -19.225460052490234,
|
| 519 |
+
"step": 170
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.3816793893129771,
|
| 523 |
+
"grad_norm": 149.4264363196396,
|
| 524 |
+
"learning_rate": 6.215936755250596e-07,
|
| 525 |
+
"logits/chosen": -17.34011459350586,
|
| 526 |
+
"logits/rejected": -17.421663284301758,
|
| 527 |
+
"logps/chosen": -1.496767282485962,
|
| 528 |
+
"logps/rejected": -1.85422682762146,
|
| 529 |
+
"loss": 3.5029,
|
| 530 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 531 |
+
"rewards/chosen": -14.967672348022461,
|
| 532 |
+
"rewards/margins": 3.574597120285034,
|
| 533 |
+
"rewards/rejected": -18.54227066040039,
|
| 534 |
+
"step": 175
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.3925845147219193,
|
| 538 |
+
"grad_norm": 116.14395125879456,
|
| 539 |
+
"learning_rate": 6.087393022311666e-07,
|
| 540 |
+
"logits/chosen": -16.09074592590332,
|
| 541 |
+
"logits/rejected": -15.661008834838867,
|
| 542 |
+
"logps/chosen": -1.6729342937469482,
|
| 543 |
+
"logps/rejected": -2.0850119590759277,
|
| 544 |
+
"loss": 3.3119,
|
| 545 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 546 |
+
"rewards/chosen": -16.729345321655273,
|
| 547 |
+
"rewards/margins": 4.1207756996154785,
|
| 548 |
+
"rewards/rejected": -20.85011863708496,
|
| 549 |
+
"step": 180
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.4034896401308615,
|
| 553 |
+
"grad_norm": 302.66398320568743,
|
| 554 |
+
"learning_rate": 5.955815419556934e-07,
|
| 555 |
+
"logits/chosen": -17.55354118347168,
|
| 556 |
+
"logits/rejected": -17.935550689697266,
|
| 557 |
+
"logps/chosen": -1.8943233489990234,
|
| 558 |
+
"logps/rejected": -2.240093231201172,
|
| 559 |
+
"loss": 3.3613,
|
| 560 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 561 |
+
"rewards/chosen": -18.943233489990234,
|
| 562 |
+
"rewards/margins": 3.4576992988586426,
|
| 563 |
+
"rewards/rejected": -22.40093231201172,
|
| 564 |
+
"step": 185
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.4143947655398037,
|
| 568 |
+
"grad_norm": 125.3361496668939,
|
| 569 |
+
"learning_rate": 5.821395185202285e-07,
|
| 570 |
+
"logits/chosen": -15.341293334960938,
|
| 571 |
+
"logits/rejected": -15.44407844543457,
|
| 572 |
+
"logps/chosen": -1.8814849853515625,
|
| 573 |
+
"logps/rejected": -2.2990918159484863,
|
| 574 |
+
"loss": 3.1174,
|
| 575 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 576 |
+
"rewards/chosen": -18.814849853515625,
|
| 577 |
+
"rewards/margins": 4.17606782913208,
|
| 578 |
+
"rewards/rejected": -22.990917205810547,
|
| 579 |
+
"step": 190
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.42529989094874593,
|
| 583 |
+
"grad_norm": 123.17711415812771,
|
| 584 |
+
"learning_rate": 5.684327689016264e-07,
|
| 585 |
+
"logits/chosen": -15.80346393585205,
|
| 586 |
+
"logits/rejected": -15.893457412719727,
|
| 587 |
+
"logps/chosen": -1.6922333240509033,
|
| 588 |
+
"logps/rejected": -2.1358566284179688,
|
| 589 |
+
"loss": 3.3943,
|
| 590 |
+
"rewards/accuracies": 0.762499988079071,
|
| 591 |
+
"rewards/chosen": -16.922334671020508,
|
| 592 |
+
"rewards/margins": 4.436235427856445,
|
| 593 |
+
"rewards/rejected": -21.358566284179688,
|
| 594 |
+
"step": 195
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 0.4362050163576881,
|
| 598 |
+
"grad_norm": 121.59808879397914,
|
| 599 |
+
"learning_rate": 5.544812148364731e-07,
|
| 600 |
+
"logits/chosen": -16.505403518676758,
|
| 601 |
+
"logits/rejected": -16.2810001373291,
|
| 602 |
+
"logps/chosen": -1.7912696599960327,
|
| 603 |
+
"logps/rejected": -2.283963680267334,
|
| 604 |
+
"loss": 3.3937,
|
| 605 |
+
"rewards/accuracies": 0.768750011920929,
|
| 606 |
+
"rewards/chosen": -17.912700653076172,
|
| 607 |
+
"rewards/margins": 4.926938533782959,
|
| 608 |
+
"rewards/rejected": -22.839635848999023,
|
| 609 |
+
"step": 200
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.44711014176663033,
|
| 613 |
+
"grad_norm": 130.03760301081726,
|
| 614 |
+
"learning_rate": 5.40305133866328e-07,
|
| 615 |
+
"logits/chosen": -16.51849365234375,
|
| 616 |
+
"logits/rejected": -16.17244529724121,
|
| 617 |
+
"logps/chosen": -1.8261501789093018,
|
| 618 |
+
"logps/rejected": -2.216474771499634,
|
| 619 |
+
"loss": 3.3439,
|
| 620 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 621 |
+
"rewards/chosen": -18.26150131225586,
|
| 622 |
+
"rewards/margins": 3.903247117996216,
|
| 623 |
+
"rewards/rejected": -22.164745330810547,
|
| 624 |
+
"step": 205
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 0.4580152671755725,
|
| 628 |
+
"grad_norm": 107.25457872042979,
|
| 629 |
+
"learning_rate": 5.259251298658339e-07,
|
| 630 |
+
"logits/chosen": -14.851339340209961,
|
| 631 |
+
"logits/rejected": -14.814165115356445,
|
| 632 |
+
"logps/chosen": -1.7905569076538086,
|
| 633 |
+
"logps/rejected": -2.320906162261963,
|
| 634 |
+
"loss": 3.1591,
|
| 635 |
+
"rewards/accuracies": 0.762499988079071,
|
| 636 |
+
"rewards/chosen": -17.905569076538086,
|
| 637 |
+
"rewards/margins": 5.303492546081543,
|
| 638 |
+
"rewards/rejected": -23.209064483642578,
|
| 639 |
+
"step": 210
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.46892039258451473,
|
| 643 |
+
"grad_norm": 129.6812695651334,
|
| 644 |
+
"learning_rate": 5.113621030965238e-07,
|
| 645 |
+
"logits/chosen": -16.215179443359375,
|
| 646 |
+
"logits/rejected": -16.335643768310547,
|
| 647 |
+
"logps/chosen": -1.8237041234970093,
|
| 648 |
+
"logps/rejected": -2.207712411880493,
|
| 649 |
+
"loss": 3.2475,
|
| 650 |
+
"rewards/accuracies": 0.737500011920929,
|
| 651 |
+
"rewards/chosen": -18.23703956604004,
|
| 652 |
+
"rewards/margins": 3.840085506439209,
|
| 653 |
+
"rewards/rejected": -22.07712745666504,
|
| 654 |
+
"step": 215
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.4798255179934569,
|
| 658 |
+
"grad_norm": 103.25783765301819,
|
| 659 |
+
"learning_rate": 4.96637219829852e-07,
|
| 660 |
+
"logits/chosen": -14.638501167297363,
|
| 661 |
+
"logits/rejected": -14.695428848266602,
|
| 662 |
+
"logps/chosen": -1.8283414840698242,
|
| 663 |
+
"logps/rejected": -2.376028537750244,
|
| 664 |
+
"loss": 3.3144,
|
| 665 |
+
"rewards/accuracies": 0.737500011920929,
|
| 666 |
+
"rewards/chosen": -18.283416748046875,
|
| 667 |
+
"rewards/margins": 5.47686767578125,
|
| 668 |
+
"rewards/rejected": -23.760284423828125,
|
| 669 |
+
"step": 220
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.49073064340239914,
|
| 673 |
+
"grad_norm": 132.036942227645,
|
| 674 |
+
"learning_rate": 4.817718815835998e-07,
|
| 675 |
+
"logits/chosen": -15.56849479675293,
|
| 676 |
+
"logits/rejected": -15.743542671203613,
|
| 677 |
+
"logps/chosen": -1.9234107732772827,
|
| 678 |
+
"logps/rejected": -2.3087854385375977,
|
| 679 |
+
"loss": 3.2906,
|
| 680 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 681 |
+
"rewards/chosen": -19.234111785888672,
|
| 682 |
+
"rewards/margins": 3.8537425994873047,
|
| 683 |
+
"rewards/rejected": -23.087854385375977,
|
| 684 |
+
"step": 225
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 0.5016357688113413,
|
| 688 |
+
"grad_norm": 132.28323726857968,
|
| 689 |
+
"learning_rate": 4.6678769401636887e-07,
|
| 690 |
+
"logits/chosen": -15.513163566589355,
|
| 691 |
+
"logits/rejected": -15.330879211425781,
|
| 692 |
+
"logps/chosen": -1.868847131729126,
|
| 693 |
+
"logps/rejected": -2.3147101402282715,
|
| 694 |
+
"loss": 3.2416,
|
| 695 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 696 |
+
"rewards/chosen": -18.688472747802734,
|
| 697 |
+
"rewards/margins": 4.4586310386657715,
|
| 698 |
+
"rewards/rejected": -23.14710235595703,
|
| 699 |
+
"step": 230
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.5125408942202835,
|
| 703 |
+
"grad_norm": 116.14658388209503,
|
| 704 |
+
"learning_rate": 4.517064355253696e-07,
|
| 705 |
+
"logits/chosen": -14.959882736206055,
|
| 706 |
+
"logits/rejected": -14.972663879394531,
|
| 707 |
+
"logps/chosen": -1.9306243658065796,
|
| 708 |
+
"logps/rejected": -2.457444667816162,
|
| 709 |
+
"loss": 2.9094,
|
| 710 |
+
"rewards/accuracies": 0.75,
|
| 711 |
+
"rewards/chosen": -19.306243896484375,
|
| 712 |
+
"rewards/margins": 5.268204689025879,
|
| 713 |
+
"rewards/rejected": -24.574447631835938,
|
| 714 |
+
"step": 235
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 0.5234460196292258,
|
| 718 |
+
"grad_norm": 125.188125846376,
|
| 719 |
+
"learning_rate": 4.365500255931484e-07,
|
| 720 |
+
"logits/chosen": -15.104351997375488,
|
| 721 |
+
"logits/rejected": -15.062860488891602,
|
| 722 |
+
"logps/chosen": -1.9797313213348389,
|
| 723 |
+
"logps/rejected": -2.4472575187683105,
|
| 724 |
+
"loss": 3.0593,
|
| 725 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 726 |
+
"rewards/chosen": -19.797313690185547,
|
| 727 |
+
"rewards/margins": 4.675261497497559,
|
| 728 |
+
"rewards/rejected": -24.472576141357422,
|
| 729 |
+
"step": 240
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.5343511450381679,
|
| 733 |
+
"grad_norm": 135.15384009467002,
|
| 734 |
+
"learning_rate": 4.213404929292575e-07,
|
| 735 |
+
"logits/chosen": -15.561657905578613,
|
| 736 |
+
"logits/rejected": -15.077102661132812,
|
| 737 |
+
"logps/chosen": -2.040621280670166,
|
| 738 |
+
"logps/rejected": -2.5414981842041016,
|
| 739 |
+
"loss": 2.9158,
|
| 740 |
+
"rewards/accuracies": 0.762499988079071,
|
| 741 |
+
"rewards/chosen": -20.40620994567871,
|
| 742 |
+
"rewards/margins": 5.008772373199463,
|
| 743 |
+
"rewards/rejected": -25.41498374938965,
|
| 744 |
+
"step": 245
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.5452562704471101,
|
| 748 |
+
"grad_norm": 124.60869295390538,
|
| 749 |
+
"learning_rate": 4.060999434531704e-07,
|
| 750 |
+
"logits/chosen": -15.397348403930664,
|
| 751 |
+
"logits/rejected": -15.1492280960083,
|
| 752 |
+
"logps/chosen": -1.8068050146102905,
|
| 753 |
+
"logps/rejected": -2.3928465843200684,
|
| 754 |
+
"loss": 2.8634,
|
| 755 |
+
"rewards/accuracies": 0.793749988079071,
|
| 756 |
+
"rewards/chosen": -18.068050384521484,
|
| 757 |
+
"rewards/margins": 5.860415935516357,
|
| 758 |
+
"rewards/rejected": -23.928464889526367,
|
| 759 |
+
"step": 250
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.5561613958560524,
|
| 763 |
+
"grad_norm": 129.14805119503163,
|
| 764 |
+
"learning_rate": 3.908505281649805e-07,
|
| 765 |
+
"logits/chosen": -14.593152046203613,
|
| 766 |
+
"logits/rejected": -14.538106918334961,
|
| 767 |
+
"logps/chosen": -1.871093988418579,
|
| 768 |
+
"logps/rejected": -2.36173152923584,
|
| 769 |
+
"loss": 2.9231,
|
| 770 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 771 |
+
"rewards/chosen": -18.710941314697266,
|
| 772 |
+
"rewards/margins": 4.906374454498291,
|
| 773 |
+
"rewards/rejected": -23.617313385009766,
|
| 774 |
+
"step": 255
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 0.5670665212649946,
|
| 778 |
+
"grad_norm": 119.36679390601662,
|
| 779 |
+
"learning_rate": 3.756144109505764e-07,
|
| 780 |
+
"logits/chosen": -15.056838989257812,
|
| 781 |
+
"logits/rejected": -15.148704528808594,
|
| 782 |
+
"logps/chosen": -1.8173519372940063,
|
| 783 |
+
"logps/rejected": -2.4098479747772217,
|
| 784 |
+
"loss": 2.8089,
|
| 785 |
+
"rewards/accuracies": 0.800000011920929,
|
| 786 |
+
"rewards/chosen": -18.173521041870117,
|
| 787 |
+
"rewards/margins": 5.924956321716309,
|
| 788 |
+
"rewards/rejected": -24.098478317260742,
|
| 789 |
+
"step": 260
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.5779716466739367,
|
| 793 |
+
"grad_norm": 119.03491048259538,
|
| 794 |
+
"learning_rate": 3.6041373636809017e-07,
|
| 795 |
+
"logits/chosen": -15.791028022766113,
|
| 796 |
+
"logits/rejected": -15.623140335083008,
|
| 797 |
+
"logps/chosen": -2.1847808361053467,
|
| 798 |
+
"logps/rejected": -2.698356866836548,
|
| 799 |
+
"loss": 3.1097,
|
| 800 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 801 |
+
"rewards/chosen": -21.847808837890625,
|
| 802 |
+
"rewards/margins": 5.135758399963379,
|
| 803 |
+
"rewards/rejected": -26.983570098876953,
|
| 804 |
+
"step": 265
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 0.5888767720828789,
|
| 808 |
+
"grad_norm": 119.59861773726368,
|
| 809 |
+
"learning_rate": 3.4527059746243596e-07,
|
| 810 |
+
"logits/chosen": -15.602317810058594,
|
| 811 |
+
"logits/rejected": -15.679903030395508,
|
| 812 |
+
"logps/chosen": -2.001981735229492,
|
| 813 |
+
"logps/rejected": -2.4404501914978027,
|
| 814 |
+
"loss": 3.1726,
|
| 815 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 816 |
+
"rewards/chosen": -20.019817352294922,
|
| 817 |
+
"rewards/margins": 4.384683132171631,
|
| 818 |
+
"rewards/rejected": -24.404499053955078,
|
| 819 |
+
"step": 270
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 0.5997818974918212,
|
| 823 |
+
"grad_norm": 143.27392642046235,
|
| 824 |
+
"learning_rate": 3.302070036547201e-07,
|
| 825 |
+
"logits/chosen": -15.435728073120117,
|
| 826 |
+
"logits/rejected": -15.314961433410645,
|
| 827 |
+
"logps/chosen": -2.168989896774292,
|
| 828 |
+
"logps/rejected": -2.737366199493408,
|
| 829 |
+
"loss": 3.0792,
|
| 830 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 831 |
+
"rewards/chosen": -21.689899444580078,
|
| 832 |
+
"rewards/margins": 5.683760643005371,
|
| 833 |
+
"rewards/rejected": -27.3736572265625,
|
| 834 |
+
"step": 275
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 0.6106870229007634,
|
| 838 |
+
"grad_norm": 114.06559627918872,
|
| 839 |
+
"learning_rate": 3.152448487531915e-07,
|
| 840 |
+
"logits/chosen": -15.740002632141113,
|
| 841 |
+
"logits/rejected": -15.62346076965332,
|
| 842 |
+
"logps/chosen": -1.9302055835723877,
|
| 843 |
+
"logps/rejected": -2.405974864959717,
|
| 844 |
+
"loss": 2.9702,
|
| 845 |
+
"rewards/accuracies": 0.762499988079071,
|
| 846 |
+
"rewards/chosen": -19.302053451538086,
|
| 847 |
+
"rewards/margins": 4.757693767547607,
|
| 848 |
+
"rewards/rejected": -24.059749603271484,
|
| 849 |
+
"step": 280
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.6215921483097055,
|
| 853 |
+
"grad_norm": 107.47439430691443,
|
| 854 |
+
"learning_rate": 3.0040587913222725e-07,
|
| 855 |
+
"logits/chosen": -14.441022872924805,
|
| 856 |
+
"logits/rejected": -14.633172988891602,
|
| 857 |
+
"logps/chosen": -1.8620984554290771,
|
| 858 |
+
"logps/rejected": -2.345489263534546,
|
| 859 |
+
"loss": 3.0044,
|
| 860 |
+
"rewards/accuracies": 0.824999988079071,
|
| 861 |
+
"rewards/chosen": -18.620983123779297,
|
| 862 |
+
"rewards/margins": 4.833909034729004,
|
| 863 |
+
"rewards/rejected": -23.45489501953125,
|
| 864 |
+
"step": 285
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.6324972737186477,
|
| 868 |
+
"grad_norm": 154.10880275569383,
|
| 869 |
+
"learning_rate": 2.857116621256018e-07,
|
| 870 |
+
"logits/chosen": -15.512109756469727,
|
| 871 |
+
"logits/rejected": -15.243085861206055,
|
| 872 |
+
"logps/chosen": -1.9362213611602783,
|
| 873 |
+
"logps/rejected": -2.402963399887085,
|
| 874 |
+
"loss": 3.2181,
|
| 875 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 876 |
+
"rewards/chosen": -19.36220932006836,
|
| 877 |
+
"rewards/margins": 4.667425155639648,
|
| 878 |
+
"rewards/rejected": -24.02963638305664,
|
| 879 |
+
"step": 290
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 0.64340239912759,
|
| 883 |
+
"grad_norm": 112.54594135095479,
|
| 884 |
+
"learning_rate": 2.7118355467997835e-07,
|
| 885 |
+
"logits/chosen": -15.574625968933105,
|
| 886 |
+
"logits/rejected": -15.6203031539917,
|
| 887 |
+
"logps/chosen": -1.9695098400115967,
|
| 888 |
+
"logps/rejected": -2.388803482055664,
|
| 889 |
+
"loss": 2.9531,
|
| 890 |
+
"rewards/accuracies": 0.6875,
|
| 891 |
+
"rewards/chosen": -19.695100784301758,
|
| 892 |
+
"rewards/margins": 4.192935943603516,
|
| 893 |
+
"rewards/rejected": -23.888032913208008,
|
| 894 |
+
"step": 295
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 0.6543075245365322,
|
| 898 |
+
"grad_norm": 131.96864010640354,
|
| 899 |
+
"learning_rate": 2.5684267231418333e-07,
|
| 900 |
+
"logits/chosen": -16.529542922973633,
|
| 901 |
+
"logits/rejected": -16.21666145324707,
|
| 902 |
+
"logps/chosen": -2.111536741256714,
|
| 903 |
+
"logps/rejected": -2.6558353900909424,
|
| 904 |
+
"loss": 2.8683,
|
| 905 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 906 |
+
"rewards/chosen": -21.11536979675293,
|
| 907 |
+
"rewards/margins": 5.442984580993652,
|
| 908 |
+
"rewards/rejected": -26.5583553314209,
|
| 909 |
+
"step": 300
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 0.6652126499454744,
|
| 913 |
+
"grad_norm": 131.182887672942,
|
| 914 |
+
"learning_rate": 2.427098584293759e-07,
|
| 915 |
+
"logits/chosen": -15.954015731811523,
|
| 916 |
+
"logits/rejected": -15.863182067871094,
|
| 917 |
+
"logps/chosen": -2.109534978866577,
|
| 918 |
+
"logps/rejected": -2.553351640701294,
|
| 919 |
+
"loss": 2.9835,
|
| 920 |
+
"rewards/accuracies": 0.737500011920929,
|
| 921 |
+
"rewards/chosen": -21.09535026550293,
|
| 922 |
+
"rewards/margins": 4.438165187835693,
|
| 923 |
+
"rewards/rejected": -25.533517837524414,
|
| 924 |
+
"step": 305
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 0.6761177753544165,
|
| 928 |
+
"grad_norm": 117.78663856490914,
|
| 929 |
+
"learning_rate": 2.2880565401472287e-07,
|
| 930 |
+
"logits/chosen": -15.363965034484863,
|
| 931 |
+
"logits/rejected": -15.453814506530762,
|
| 932 |
+
"logps/chosen": -2.0754928588867188,
|
| 933 |
+
"logps/rejected": -2.7302043437957764,
|
| 934 |
+
"loss": 2.9351,
|
| 935 |
+
"rewards/accuracies": 0.793749988079071,
|
| 936 |
+
"rewards/chosen": -20.754932403564453,
|
| 937 |
+
"rewards/margins": 6.547112464904785,
|
| 938 |
+
"rewards/rejected": -27.302043914794922,
|
| 939 |
+
"step": 310
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 0.6870229007633588,
|
| 943 |
+
"grad_norm": 102.64755949094341,
|
| 944 |
+
"learning_rate": 2.151502677926042e-07,
|
| 945 |
+
"logits/chosen": -15.314285278320312,
|
| 946 |
+
"logits/rejected": -15.252153396606445,
|
| 947 |
+
"logps/chosen": -2.0483040809631348,
|
| 948 |
+
"logps/rejected": -2.753251552581787,
|
| 949 |
+
"loss": 2.7283,
|
| 950 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 951 |
+
"rewards/chosen": -20.483041763305664,
|
| 952 |
+
"rewards/margins": 7.049475193023682,
|
| 953 |
+
"rewards/rejected": -27.532512664794922,
|
| 954 |
+
"step": 315
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.697928026172301,
|
| 958 |
+
"grad_norm": 101.22013067633767,
|
| 959 |
+
"learning_rate": 2.0176354684674575e-07,
|
| 960 |
+
"logits/chosen": -16.12703514099121,
|
| 961 |
+
"logits/rejected": -15.803072929382324,
|
| 962 |
+
"logps/chosen": -2.151562213897705,
|
| 963 |
+
"logps/rejected": -2.721626043319702,
|
| 964 |
+
"loss": 2.9477,
|
| 965 |
+
"rewards/accuracies": 0.793749988079071,
|
| 966 |
+
"rewards/chosen": -21.515621185302734,
|
| 967 |
+
"rewards/margins": 5.700636863708496,
|
| 968 |
+
"rewards/rejected": -27.216259002685547,
|
| 969 |
+
"step": 320
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.7088331515812432,
|
| 973 |
+
"grad_norm": 125.44812959020139,
|
| 974 |
+
"learning_rate": 1.8866494777596534e-07,
|
| 975 |
+
"logits/chosen": -15.385473251342773,
|
| 976 |
+
"logits/rejected": -15.099942207336426,
|
| 977 |
+
"logps/chosen": -2.194483518600464,
|
| 978 |
+
"logps/rejected": -2.6979141235351562,
|
| 979 |
+
"loss": 3.0708,
|
| 980 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 981 |
+
"rewards/chosen": -21.944835662841797,
|
| 982 |
+
"rewards/margins": 5.034304141998291,
|
| 983 |
+
"rewards/rejected": -26.979141235351562,
|
| 984 |
+
"step": 325
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 0.7197382769901853,
|
| 988 |
+
"grad_norm": 95.2552912244021,
|
| 989 |
+
"learning_rate": 1.7587350841546007e-07,
|
| 990 |
+
"logits/chosen": -15.96777629852295,
|
| 991 |
+
"logits/rejected": -16.115123748779297,
|
| 992 |
+
"logps/chosen": -1.9400516748428345,
|
| 993 |
+
"logps/rejected": -2.51167368888855,
|
| 994 |
+
"loss": 2.8539,
|
| 995 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 996 |
+
"rewards/chosen": -19.400516510009766,
|
| 997 |
+
"rewards/margins": 5.716220855712891,
|
| 998 |
+
"rewards/rejected": -25.11673927307129,
|
| 999 |
+
"step": 330
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 0.7306434023991276,
|
| 1003 |
+
"grad_norm": 128.737726612481,
|
| 1004 |
+
"learning_rate": 1.634078201667347e-07,
|
| 1005 |
+
"logits/chosen": -15.98449420928955,
|
| 1006 |
+
"logits/rejected": -16.243240356445312,
|
| 1007 |
+
"logps/chosen": -2.1536316871643066,
|
| 1008 |
+
"logps/rejected": -2.7533793449401855,
|
| 1009 |
+
"loss": 2.9433,
|
| 1010 |
+
"rewards/accuracies": 0.8125,
|
| 1011 |
+
"rewards/chosen": -21.53631591796875,
|
| 1012 |
+
"rewards/margins": 5.9974775314331055,
|
| 1013 |
+
"rewards/rejected": -27.53379249572754,
|
| 1014 |
+
"step": 335
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 0.7415485278080698,
|
| 1018 |
+
"grad_norm": 150.4063701643456,
|
| 1019 |
+
"learning_rate": 1.512860009763891e-07,
|
| 1020 |
+
"logits/chosen": -15.605964660644531,
|
| 1021 |
+
"logits/rejected": -16.06555938720703,
|
| 1022 |
+
"logps/chosen": -1.8955068588256836,
|
| 1023 |
+
"logps/rejected": -2.390796422958374,
|
| 1024 |
+
"loss": 2.9364,
|
| 1025 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1026 |
+
"rewards/chosen": -18.955068588256836,
|
| 1027 |
+
"rewards/margins": 4.952897071838379,
|
| 1028 |
+
"rewards/rejected": -23.9079647064209,
|
| 1029 |
+
"step": 340
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 0.752453653217012,
|
| 1033 |
+
"grad_norm": 135.6732671225431,
|
| 1034 |
+
"learning_rate": 1.3952566900303602e-07,
|
| 1035 |
+
"logits/chosen": -16.359243392944336,
|
| 1036 |
+
"logits/rejected": -16.223447799682617,
|
| 1037 |
+
"logps/chosen": -2.149503469467163,
|
| 1038 |
+
"logps/rejected": -2.7064743041992188,
|
| 1039 |
+
"loss": 2.929,
|
| 1040 |
+
"rewards/accuracies": 0.75,
|
| 1041 |
+
"rewards/chosen": -21.49503517150879,
|
| 1042 |
+
"rewards/margins": 5.569708347320557,
|
| 1043 |
+
"rewards/rejected": -27.064743041992188,
|
| 1044 |
+
"step": 345
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 0.7633587786259542,
|
| 1048 |
+
"grad_norm": 125.758255019621,
|
| 1049 |
+
"learning_rate": 1.2814391701062391e-07,
|
| 1050 |
+
"logits/chosen": -14.762056350708008,
|
| 1051 |
+
"logits/rejected": -14.951733589172363,
|
| 1052 |
+
"logps/chosen": -2.00016450881958,
|
| 1053 |
+
"logps/rejected": -2.528573989868164,
|
| 1054 |
+
"loss": 3.0167,
|
| 1055 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1056 |
+
"rewards/chosen": -20.00164794921875,
|
| 1057 |
+
"rewards/margins": 5.284093379974365,
|
| 1058 |
+
"rewards/rejected": -25.285737991333008,
|
| 1059 |
+
"step": 350
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.7742639040348964,
|
| 1063 |
+
"grad_norm": 123.77860608266465,
|
| 1064 |
+
"learning_rate": 1.17157287525381e-07,
|
| 1065 |
+
"logits/chosen": -14.972272872924805,
|
| 1066 |
+
"logits/rejected": -15.062446594238281,
|
| 1067 |
+
"logps/chosen": -2.0419890880584717,
|
| 1068 |
+
"logps/rejected": -2.7083749771118164,
|
| 1069 |
+
"loss": 3.1472,
|
| 1070 |
+
"rewards/accuracies": 0.793749988079071,
|
| 1071 |
+
"rewards/chosen": -20.419891357421875,
|
| 1072 |
+
"rewards/margins": 6.663858890533447,
|
| 1073 |
+
"rewards/rejected": -27.083751678466797,
|
| 1074 |
+
"step": 355
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.7851690294438386,
|
| 1078 |
+
"grad_norm": 128.49716646084636,
|
| 1079 |
+
"learning_rate": 1.0658174879249e-07,
|
| 1080 |
+
"logits/chosen": -15.742474555969238,
|
| 1081 |
+
"logits/rejected": -15.778717041015625,
|
| 1082 |
+
"logps/chosen": -2.0881805419921875,
|
| 1083 |
+
"logps/rejected": -2.6481823921203613,
|
| 1084 |
+
"loss": 2.9629,
|
| 1085 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1086 |
+
"rewards/chosen": -20.881805419921875,
|
| 1087 |
+
"rewards/margins": 5.6000165939331055,
|
| 1088 |
+
"rewards/rejected": -26.481822967529297,
|
| 1089 |
+
"step": 360
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 0.7960741548527808,
|
| 1093 |
+
"grad_norm": 105.81239577262055,
|
| 1094 |
+
"learning_rate": 9.643267156743626e-08,
|
| 1095 |
+
"logits/chosen": -15.465214729309082,
|
| 1096 |
+
"logits/rejected": -15.414011001586914,
|
| 1097 |
+
"logps/chosen": -2.0494282245635986,
|
| 1098 |
+
"logps/rejected": -2.6396915912628174,
|
| 1099 |
+
"loss": 2.9365,
|
| 1100 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1101 |
+
"rewards/chosen": -20.49428367614746,
|
| 1102 |
+
"rewards/margins": 5.9026336669921875,
|
| 1103 |
+
"rewards/rejected": -26.39691734313965,
|
| 1104 |
+
"step": 365
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 0.806979280261723,
|
| 1108 |
+
"grad_norm": 106.05628455304455,
|
| 1109 |
+
"learning_rate": 8.672480677576266e-08,
|
| 1110 |
+
"logits/chosen": -15.015769958496094,
|
| 1111 |
+
"logits/rejected": -14.94062614440918,
|
| 1112 |
+
"logps/chosen": -2.0296902656555176,
|
| 1113 |
+
"logps/rejected": -2.648134231567383,
|
| 1114 |
+
"loss": 2.8337,
|
| 1115 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1116 |
+
"rewards/chosen": -20.29690170288086,
|
| 1117 |
+
"rewards/margins": 6.184438705444336,
|
| 1118 |
+
"rewards/rejected": -26.481342315673828,
|
| 1119 |
+
"step": 370
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 0.8178844056706652,
|
| 1123 |
+
"grad_norm": 139.7590362680712,
|
| 1124 |
+
"learning_rate": 7.7472264073702e-08,
|
| 1125 |
+
"logits/chosen": -15.844355583190918,
|
| 1126 |
+
"logits/rejected": -15.914880752563477,
|
| 1127 |
+
"logps/chosen": -2.140533208847046,
|
| 1128 |
+
"logps/rejected": -2.731921434402466,
|
| 1129 |
+
"loss": 2.7732,
|
| 1130 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1131 |
+
"rewards/chosen": -21.405330657958984,
|
| 1132 |
+
"rewards/margins": 5.913882255554199,
|
| 1133 |
+
"rewards/rejected": -27.3192138671875,
|
| 1134 |
+
"step": 375
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 0.8287895310796074,
|
| 1138 |
+
"grad_norm": 123.27756798269391,
|
| 1139 |
+
"learning_rate": 6.868849134084534e-08,
|
| 1140 |
+
"logits/chosen": -14.827451705932617,
|
| 1141 |
+
"logits/rejected": -14.896784782409668,
|
| 1142 |
+
"logps/chosen": -2.1212401390075684,
|
| 1143 |
+
"logps/rejected": -2.6644959449768066,
|
| 1144 |
+
"loss": 2.7211,
|
| 1145 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1146 |
+
"rewards/chosen": -21.21240234375,
|
| 1147 |
+
"rewards/margins": 5.432553291320801,
|
| 1148 |
+
"rewards/rejected": -26.644954681396484,
|
| 1149 |
+
"step": 380
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 0.8396946564885496,
|
| 1153 |
+
"grad_norm": 102.3035020753843,
|
| 1154 |
+
"learning_rate": 6.038625513465372e-08,
|
| 1155 |
+
"logits/chosen": -15.892682075500488,
|
| 1156 |
+
"logits/rejected": -15.998703002929688,
|
| 1157 |
+
"logps/chosen": -2.0303637981414795,
|
| 1158 |
+
"logps/rejected": -2.4150891304016113,
|
| 1159 |
+
"loss": 2.9474,
|
| 1160 |
+
"rewards/accuracies": 0.6875,
|
| 1161 |
+
"rewards/chosen": -20.303638458251953,
|
| 1162 |
+
"rewards/margins": 3.8472514152526855,
|
| 1163 |
+
"rewards/rejected": -24.15088653564453,
|
| 1164 |
+
"step": 385
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.8505997818974919,
|
| 1168 |
+
"grad_norm": 110.56211402431573,
|
| 1169 |
+
"learning_rate": 5.2577622135220546e-08,
|
| 1170 |
+
"logits/chosen": -15.678448677062988,
|
| 1171 |
+
"logits/rejected": -15.579752922058105,
|
| 1172 |
+
"logps/chosen": -2.146989345550537,
|
| 1173 |
+
"logps/rejected": -2.7087996006011963,
|
| 1174 |
+
"loss": 2.8812,
|
| 1175 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 1176 |
+
"rewards/chosen": -21.469894409179688,
|
| 1177 |
+
"rewards/margins": 5.618099689483643,
|
| 1178 |
+
"rewards/rejected": -27.087993621826172,
|
| 1179 |
+
"step": 390
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.861504907306434,
|
| 1183 |
+
"grad_norm": 136.5376747108113,
|
| 1184 |
+
"learning_rate": 4.527394160725375e-08,
|
| 1185 |
+
"logits/chosen": -15.390423774719238,
|
| 1186 |
+
"logits/rejected": -15.318705558776855,
|
| 1187 |
+
"logps/chosen": -2.0882272720336914,
|
| 1188 |
+
"logps/rejected": -2.5445287227630615,
|
| 1189 |
+
"loss": 2.9208,
|
| 1190 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 1191 |
+
"rewards/chosen": -20.882274627685547,
|
| 1192 |
+
"rewards/margins": 4.563012599945068,
|
| 1193 |
+
"rewards/rejected": -25.44528579711914,
|
| 1194 |
+
"step": 395
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 0.8724100327153762,
|
| 1198 |
+
"grad_norm": 110.11074526975227,
|
| 1199 |
+
"learning_rate": 3.848582890476728e-08,
|
| 1200 |
+
"logits/chosen": -13.988882064819336,
|
| 1201 |
+
"logits/rejected": -14.337841987609863,
|
| 1202 |
+
"logps/chosen": -1.9440772533416748,
|
| 1203 |
+
"logps/rejected": -2.5067763328552246,
|
| 1204 |
+
"loss": 2.7931,
|
| 1205 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1206 |
+
"rewards/chosen": -19.440773010253906,
|
| 1207 |
+
"rewards/margins": 5.626988410949707,
|
| 1208 |
+
"rewards/rejected": -25.067760467529297,
|
| 1209 |
+
"step": 400
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 0.8724100327153762,
|
| 1213 |
+
"eval_logits/chosen": -13.542856216430664,
|
| 1214 |
+
"eval_logits/rejected": -13.604004859924316,
|
| 1215 |
+
"eval_logps/chosen": -1.9483120441436768,
|
| 1216 |
+
"eval_logps/rejected": -2.4730513095855713,
|
| 1217 |
+
"eval_loss": 2.762831449508667,
|
| 1218 |
+
"eval_rewards/accuracies": 0.7807376980781555,
|
| 1219 |
+
"eval_rewards/chosen": -19.48311996459961,
|
| 1220 |
+
"eval_rewards/margins": 5.247389316558838,
|
| 1221 |
+
"eval_rewards/rejected": -24.73050880432129,
|
| 1222 |
+
"eval_runtime": 66.0141,
|
| 1223 |
+
"eval_samples_per_second": 29.403,
|
| 1224 |
+
"eval_steps_per_second": 1.848,
|
| 1225 |
+
"step": 400
|
| 1226 |
+
},
|
| 1227 |
+
{
|
| 1228 |
+
"epoch": 0.8833151581243184,
|
| 1229 |
+
"grad_norm": 106.81510695706733,
|
| 1230 |
+
"learning_rate": 3.222315004245715e-08,
|
| 1231 |
+
"logits/chosen": -15.698140144348145,
|
| 1232 |
+
"logits/rejected": -15.643528938293457,
|
| 1233 |
+
"logps/chosen": -2.1120400428771973,
|
| 1234 |
+
"logps/rejected": -2.7991552352905273,
|
| 1235 |
+
"loss": 2.7377,
|
| 1236 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1237 |
+
"rewards/chosen": -21.120403289794922,
|
| 1238 |
+
"rewards/margins": 6.871151924133301,
|
| 1239 |
+
"rewards/rejected": -27.991552352905273,
|
| 1240 |
+
"step": 405
|
| 1241 |
+
},
|
| 1242 |
+
{
|
| 1243 |
+
"epoch": 0.8942202835332607,
|
| 1244 |
+
"grad_norm": 127.02527681895255,
|
| 1245 |
+
"learning_rate": 2.6495007356185817e-08,
|
| 1246 |
+
"logits/chosen": -14.854022026062012,
|
| 1247 |
+
"logits/rejected": -15.155965805053711,
|
| 1248 |
+
"logps/chosen": -2.028907299041748,
|
| 1249 |
+
"logps/rejected": -2.589961528778076,
|
| 1250 |
+
"loss": 2.7722,
|
| 1251 |
+
"rewards/accuracies": 0.793749988079071,
|
| 1252 |
+
"rewards/chosen": -20.289072036743164,
|
| 1253 |
+
"rewards/margins": 5.610544204711914,
|
| 1254 |
+
"rewards/rejected": -25.899616241455078,
|
| 1255 |
+
"step": 410
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 0.9051254089422028,
|
| 1259 |
+
"grad_norm": 118.59066087169252,
|
| 1260 |
+
"learning_rate": 2.1309726273417605e-08,
|
| 1261 |
+
"logits/chosen": -15.23388385772705,
|
| 1262 |
+
"logits/rejected": -15.077451705932617,
|
| 1263 |
+
"logps/chosen": -2.019240379333496,
|
| 1264 |
+
"logps/rejected": -2.622464895248413,
|
| 1265 |
+
"loss": 2.7385,
|
| 1266 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 1267 |
+
"rewards/chosen": -20.192401885986328,
|
| 1268 |
+
"rewards/margins": 6.0322442054748535,
|
| 1269 |
+
"rewards/rejected": -26.224645614624023,
|
| 1270 |
+
"step": 415
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 0.916030534351145,
|
| 1274 |
+
"grad_norm": 125.5340583771102,
|
| 1275 |
+
"learning_rate": 1.667484321283146e-08,
|
| 1276 |
+
"logits/chosen": -16.14217758178711,
|
| 1277 |
+
"logits/rejected": -15.975908279418945,
|
| 1278 |
+
"logps/chosen": -2.1546969413757324,
|
| 1279 |
+
"logps/rejected": -2.7095894813537598,
|
| 1280 |
+
"loss": 2.8351,
|
| 1281 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1282 |
+
"rewards/chosen": -21.54697036743164,
|
| 1283 |
+
"rewards/margins": 5.548920631408691,
|
| 1284 |
+
"rewards/rejected": -27.09589195251465,
|
| 1285 |
+
"step": 420
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 0.9269356597600873,
|
| 1289 |
+
"grad_norm": 126.63955014497344,
|
| 1290 |
+
"learning_rate": 1.2597094630699156e-08,
|
| 1291 |
+
"logits/chosen": -16.528745651245117,
|
| 1292 |
+
"logits/rejected": -16.543407440185547,
|
| 1293 |
+
"logps/chosen": -1.9707733392715454,
|
| 1294 |
+
"logps/rejected": -2.48225474357605,
|
| 1295 |
+
"loss": 2.7718,
|
| 1296 |
+
"rewards/accuracies": 0.8125,
|
| 1297 |
+
"rewards/chosen": -19.707733154296875,
|
| 1298 |
+
"rewards/margins": 5.1148176193237305,
|
| 1299 |
+
"rewards/rejected": -24.822551727294922,
|
| 1300 |
+
"step": 425
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"epoch": 0.9378407851690295,
|
| 1304 |
+
"grad_norm": 141.60001641865918,
|
| 1305 |
+
"learning_rate": 9.082407229950017e-09,
|
| 1306 |
+
"logits/chosen": -16.1947021484375,
|
| 1307 |
+
"logits/rejected": -16.0598087310791,
|
| 1308 |
+
"logps/chosen": -2.160557985305786,
|
| 1309 |
+
"logps/rejected": -2.6326889991760254,
|
| 1310 |
+
"loss": 3.0444,
|
| 1311 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1312 |
+
"rewards/chosen": -21.605579376220703,
|
| 1313 |
+
"rewards/margins": 4.721311092376709,
|
| 1314 |
+
"rewards/rejected": -26.326889038085938,
|
| 1315 |
+
"step": 430
|
| 1316 |
+
},
|
| 1317 |
+
{
|
| 1318 |
+
"epoch": 0.9487459105779716,
|
| 1319 |
+
"grad_norm": 114.72393277887394,
|
| 1320 |
+
"learning_rate": 6.1358893461505916e-09,
|
| 1321 |
+
"logits/chosen": -15.125299453735352,
|
| 1322 |
+
"logits/rejected": -15.178013801574707,
|
| 1323 |
+
"logps/chosen": -2.1593708992004395,
|
| 1324 |
+
"logps/rejected": -2.785780429840088,
|
| 1325 |
+
"loss": 2.7684,
|
| 1326 |
+
"rewards/accuracies": 0.78125,
|
| 1327 |
+
"rewards/chosen": -21.593708038330078,
|
| 1328 |
+
"rewards/margins": 6.264092445373535,
|
| 1329 |
+
"rewards/rejected": -27.857803344726562,
|
| 1330 |
+
"step": 435
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"epoch": 0.9596510359869138,
|
| 1334 |
+
"grad_norm": 120.19496730357073,
|
| 1335 |
+
"learning_rate": 3.761823522920426e-09,
|
| 1336 |
+
"logits/chosen": -15.59558391571045,
|
| 1337 |
+
"logits/rejected": -15.761117935180664,
|
| 1338 |
+
"logps/chosen": -2.194023847579956,
|
| 1339 |
+
"logps/rejected": -2.795337677001953,
|
| 1340 |
+
"loss": 2.8552,
|
| 1341 |
+
"rewards/accuracies": 0.75,
|
| 1342 |
+
"rewards/chosen": -21.940235137939453,
|
| 1343 |
+
"rewards/margins": 6.013139247894287,
|
| 1344 |
+
"rewards/rejected": -27.953380584716797,
|
| 1345 |
+
"step": 440
|
| 1346 |
+
},
|
| 1347 |
+
{
|
| 1348 |
+
"epoch": 0.9705561613958561,
|
| 1349 |
+
"grad_norm": 139.41969950436,
|
| 1350 |
+
"learning_rate": 1.963660287575086e-09,
|
| 1351 |
+
"logits/chosen": -14.188136100769043,
|
| 1352 |
+
"logits/rejected": -14.617622375488281,
|
| 1353 |
+
"logps/chosen": -2.016371011734009,
|
| 1354 |
+
"logps/rejected": -2.6547458171844482,
|
| 1355 |
+
"loss": 2.746,
|
| 1356 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 1357 |
+
"rewards/chosen": -20.16370964050293,
|
| 1358 |
+
"rewards/margins": 6.383749961853027,
|
| 1359 |
+
"rewards/rejected": -26.547454833984375,
|
| 1360 |
+
"step": 445
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 0.9814612868047983,
|
| 1364 |
+
"grad_norm": 121.93850210912117,
|
| 1365 |
+
"learning_rate": 7.44013136042465e-10,
|
| 1366 |
+
"logits/chosen": -15.886639595031738,
|
| 1367 |
+
"logits/rejected": -16.0024356842041,
|
| 1368 |
+
"logps/chosen": -2.115778684616089,
|
| 1369 |
+
"logps/rejected": -2.6602773666381836,
|
| 1370 |
+
"loss": 2.9859,
|
| 1371 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1372 |
+
"rewards/chosen": -21.157787322998047,
|
| 1373 |
+
"rewards/margins": 5.444985389709473,
|
| 1374 |
+
"rewards/rejected": -26.602771759033203,
|
| 1375 |
+
"step": 450
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 0.9923664122137404,
|
| 1379 |
+
"grad_norm": 122.9917872545958,
|
| 1380 |
+
"learning_rate": 1.0465473434155824e-10,
|
| 1381 |
+
"logits/chosen": -13.859865188598633,
|
| 1382 |
+
"logits/rejected": -13.9098539352417,
|
| 1383 |
+
"logps/chosen": -2.006869077682495,
|
| 1384 |
+
"logps/rejected": -2.4212822914123535,
|
| 1385 |
+
"loss": 3.0127,
|
| 1386 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1387 |
+
"rewards/chosen": -20.06869125366211,
|
| 1388 |
+
"rewards/margins": 4.144131660461426,
|
| 1389 |
+
"rewards/rejected": -24.21282386779785,
|
| 1390 |
+
"step": 455
|
| 1391 |
+
},
|
| 1392 |
+
{
|
| 1393 |
+
"epoch": 0.9989094874591058,
|
| 1394 |
+
"step": 458,
|
| 1395 |
+
"total_flos": 0.0,
|
| 1396 |
+
"train_loss": 3.523275049492782,
|
| 1397 |
+
"train_runtime": 6692.3835,
|
| 1398 |
+
"train_samples_per_second": 8.769,
|
| 1399 |
+
"train_steps_per_second": 0.068
|
| 1400 |
+
}
|
| 1401 |
+
],
|
| 1402 |
+
"logging_steps": 5,
|
| 1403 |
+
"max_steps": 458,
|
| 1404 |
+
"num_input_tokens_seen": 0,
|
| 1405 |
+
"num_train_epochs": 1,
|
| 1406 |
+
"save_steps": 1000000,
|
| 1407 |
+
"stateful_callbacks": {
|
| 1408 |
+
"TrainerControl": {
|
| 1409 |
+
"args": {
|
| 1410 |
+
"should_epoch_stop": false,
|
| 1411 |
+
"should_evaluate": false,
|
| 1412 |
+
"should_log": false,
|
| 1413 |
+
"should_save": true,
|
| 1414 |
+
"should_training_stop": true
|
| 1415 |
+
},
|
| 1416 |
+
"attributes": {}
|
| 1417 |
+
}
|
| 1418 |
+
},
|
| 1419 |
+
"total_flos": 0.0,
|
| 1420 |
+
"train_batch_size": 2,
|
| 1421 |
+
"trial_name": null,
|
| 1422 |
+
"trial_params": null
|
| 1423 |
+
}
|