Model save
Browse files- README.md +61 -0
- all_results.json +9 -0
- generation_config.json +6 -0
- train_results.json +9 -0
- trainer_state.json +732 -0
README.md
ADDED
|
@@ -0,0 +1,61 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
license: apache-2.0
|
| 4 |
+
base_model: mistralai/Mistral-7B-Instruct-v0.2
|
| 5 |
+
tags:
|
| 6 |
+
- trl
|
| 7 |
+
- simpo
|
| 8 |
+
- generated_from_trainer
|
| 9 |
+
model-index:
|
| 10 |
+
- name: mistral-7b-instruct-v0.2-simpo-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 |
+
# mistral-7b-instruct-v0.2-simpo-full
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [mistralai/Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) on an unknown dataset.
|
| 20 |
+
|
| 21 |
+
## Model description
|
| 22 |
+
|
| 23 |
+
More information needed
|
| 24 |
+
|
| 25 |
+
## Intended uses & limitations
|
| 26 |
+
|
| 27 |
+
More information needed
|
| 28 |
+
|
| 29 |
+
## Training and evaluation data
|
| 30 |
+
|
| 31 |
+
More information needed
|
| 32 |
+
|
| 33 |
+
## Training procedure
|
| 34 |
+
|
| 35 |
+
### Training hyperparameters
|
| 36 |
+
|
| 37 |
+
The following hyperparameters were used during training:
|
| 38 |
+
- learning_rate: 1e-06
|
| 39 |
+
- train_batch_size: 2
|
| 40 |
+
- eval_batch_size: 4
|
| 41 |
+
- seed: 42
|
| 42 |
+
- distributed_type: multi-GPU
|
| 43 |
+
- num_devices: 8
|
| 44 |
+
- gradient_accumulation_steps: 16
|
| 45 |
+
- total_train_batch_size: 256
|
| 46 |
+
- total_eval_batch_size: 32
|
| 47 |
+
- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
|
| 48 |
+
- lr_scheduler_type: cosine
|
| 49 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 50 |
+
- num_epochs: 1
|
| 51 |
+
|
| 52 |
+
### Training results
|
| 53 |
+
|
| 54 |
+
|
| 55 |
+
|
| 56 |
+
### Framework versions
|
| 57 |
+
|
| 58 |
+
- Transformers 4.46.3
|
| 59 |
+
- Pytorch 2.7.1+cu126
|
| 60 |
+
- Datasets 4.0.0
|
| 61 |
+
- Tokenizers 0.20.3
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.998660594695955,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.7728924853607309,
|
| 5 |
+
"train_runtime": 5369.4124,
|
| 6 |
+
"train_samples": 59720,
|
| 7 |
+
"train_samples_per_second": 11.122,
|
| 8 |
+
"train_steps_per_second": 0.043
|
| 9 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"eos_token_id": 2,
|
| 5 |
+
"transformers_version": "4.46.3"
|
| 6 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.998660594695955,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.7728924853607309,
|
| 5 |
+
"train_runtime": 5369.4124,
|
| 6 |
+
"train_samples": 59720,
|
| 7 |
+
"train_samples_per_second": 11.122,
|
| 8 |
+
"train_steps_per_second": 0.043
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,732 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.998660594695955,
|
| 5 |
+
"eval_steps": 400,
|
| 6 |
+
"global_step": 233,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.021430484864720063,
|
| 13 |
+
"grad_norm": 20.883335931995553,
|
| 14 |
+
"learning_rate": 2.0833333333333333e-07,
|
| 15 |
+
"logits/chosen": -2.265625,
|
| 16 |
+
"logits/rejected": -2.28125,
|
| 17 |
+
"logps/chosen": -0.59765625,
|
| 18 |
+
"logps/rejected": -0.58984375,
|
| 19 |
+
"loss": 0.8615,
|
| 20 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 21 |
+
"rewards/chosen": -1.4921875,
|
| 22 |
+
"rewards/margins": -0.0164794921875,
|
| 23 |
+
"rewards/rejected": -1.4765625,
|
| 24 |
+
"step": 5
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.042860969729440125,
|
| 28 |
+
"grad_norm": 19.90222092204048,
|
| 29 |
+
"learning_rate": 4.1666666666666667e-07,
|
| 30 |
+
"logits/chosen": -2.3125,
|
| 31 |
+
"logits/rejected": -2.296875,
|
| 32 |
+
"logps/chosen": -0.59765625,
|
| 33 |
+
"logps/rejected": -0.609375,
|
| 34 |
+
"loss": 0.8473,
|
| 35 |
+
"rewards/accuracies": 0.518750011920929,
|
| 36 |
+
"rewards/chosen": -1.4921875,
|
| 37 |
+
"rewards/margins": 0.034423828125,
|
| 38 |
+
"rewards/rejected": -1.5234375,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.06429145459416019,
|
| 43 |
+
"grad_norm": 14.62873307673371,
|
| 44 |
+
"learning_rate": 6.249999999999999e-07,
|
| 45 |
+
"logits/chosen": -2.359375,
|
| 46 |
+
"logits/rejected": -2.34375,
|
| 47 |
+
"logps/chosen": -0.5078125,
|
| 48 |
+
"logps/rejected": -0.5390625,
|
| 49 |
+
"loss": 0.8339,
|
| 50 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 51 |
+
"rewards/chosen": -1.265625,
|
| 52 |
+
"rewards/margins": 0.07568359375,
|
| 53 |
+
"rewards/rejected": -1.34375,
|
| 54 |
+
"step": 15
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.08572193945888025,
|
| 58 |
+
"grad_norm": 19.932354741180678,
|
| 59 |
+
"learning_rate": 8.333333333333333e-07,
|
| 60 |
+
"logits/chosen": -2.3125,
|
| 61 |
+
"logits/rejected": -2.328125,
|
| 62 |
+
"logps/chosen": -0.50390625,
|
| 63 |
+
"logps/rejected": -0.52734375,
|
| 64 |
+
"loss": 0.839,
|
| 65 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 66 |
+
"rewards/chosen": -1.2578125,
|
| 67 |
+
"rewards/margins": 0.053955078125,
|
| 68 |
+
"rewards/rejected": -1.3125,
|
| 69 |
+
"step": 20
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.10715242432360032,
|
| 73 |
+
"grad_norm": 14.148190772420111,
|
| 74 |
+
"learning_rate": 9.999435142363483e-07,
|
| 75 |
+
"logits/chosen": -2.28125,
|
| 76 |
+
"logits/rejected": -2.28125,
|
| 77 |
+
"logps/chosen": -0.494140625,
|
| 78 |
+
"logps/rejected": -0.498046875,
|
| 79 |
+
"loss": 0.8281,
|
| 80 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 81 |
+
"rewards/chosen": -1.234375,
|
| 82 |
+
"rewards/margins": 0.01104736328125,
|
| 83 |
+
"rewards/rejected": -1.2421875,
|
| 84 |
+
"step": 25
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.12858290918832038,
|
| 88 |
+
"grad_norm": 13.826503591038982,
|
| 89 |
+
"learning_rate": 9.97967852255038e-07,
|
| 90 |
+
"logits/chosen": -2.296875,
|
| 91 |
+
"logits/rejected": -2.3125,
|
| 92 |
+
"logps/chosen": -0.47265625,
|
| 93 |
+
"logps/rejected": -0.48046875,
|
| 94 |
+
"loss": 0.8253,
|
| 95 |
+
"rewards/accuracies": 0.4749999940395355,
|
| 96 |
+
"rewards/chosen": -1.1796875,
|
| 97 |
+
"rewards/margins": 0.018798828125,
|
| 98 |
+
"rewards/rejected": -1.203125,
|
| 99 |
+
"step": 30
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.15001339405304046,
|
| 103 |
+
"grad_norm": 14.367978791172906,
|
| 104 |
+
"learning_rate": 9.931806517013612e-07,
|
| 105 |
+
"logits/chosen": -2.359375,
|
| 106 |
+
"logits/rejected": -2.390625,
|
| 107 |
+
"logps/chosen": -0.462890625,
|
| 108 |
+
"logps/rejected": -0.462890625,
|
| 109 |
+
"loss": 0.8095,
|
| 110 |
+
"rewards/accuracies": 0.4749999940395355,
|
| 111 |
+
"rewards/chosen": -1.15625,
|
| 112 |
+
"rewards/margins": 0.002044677734375,
|
| 113 |
+
"rewards/rejected": -1.15625,
|
| 114 |
+
"step": 35
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.1714438789177605,
|
| 118 |
+
"grad_norm": 23.918114882300664,
|
| 119 |
+
"learning_rate": 9.856089412257604e-07,
|
| 120 |
+
"logits/chosen": -2.421875,
|
| 121 |
+
"logits/rejected": -2.40625,
|
| 122 |
+
"logps/chosen": -0.486328125,
|
| 123 |
+
"logps/rejected": -0.5234375,
|
| 124 |
+
"loss": 0.8151,
|
| 125 |
+
"rewards/accuracies": 0.550000011920929,
|
| 126 |
+
"rewards/chosen": -1.21875,
|
| 127 |
+
"rewards/margins": 0.095703125,
|
| 128 |
+
"rewards/rejected": -1.3125,
|
| 129 |
+
"step": 40
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.19287436378248057,
|
| 133 |
+
"grad_norm": 21.40358486661956,
|
| 134 |
+
"learning_rate": 9.752954708892377e-07,
|
| 135 |
+
"logits/chosen": -2.34375,
|
| 136 |
+
"logits/rejected": -2.359375,
|
| 137 |
+
"logps/chosen": -0.5,
|
| 138 |
+
"logps/rejected": -0.5234375,
|
| 139 |
+
"loss": 0.8191,
|
| 140 |
+
"rewards/accuracies": 0.5625,
|
| 141 |
+
"rewards/chosen": -1.25,
|
| 142 |
+
"rewards/margins": 0.054443359375,
|
| 143 |
+
"rewards/rejected": -1.3046875,
|
| 144 |
+
"step": 45
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.21430484864720065,
|
| 148 |
+
"grad_norm": 13.919989253092268,
|
| 149 |
+
"learning_rate": 9.62298470795473e-07,
|
| 150 |
+
"logits/chosen": -2.34375,
|
| 151 |
+
"logits/rejected": -2.328125,
|
| 152 |
+
"logps/chosen": -0.490234375,
|
| 153 |
+
"logps/rejected": -0.53515625,
|
| 154 |
+
"loss": 0.801,
|
| 155 |
+
"rewards/accuracies": 0.59375,
|
| 156 |
+
"rewards/chosen": -1.2265625,
|
| 157 |
+
"rewards/margins": 0.11083984375,
|
| 158 |
+
"rewards/rejected": -1.3359375,
|
| 159 |
+
"step": 50
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.23573533351192072,
|
| 163 |
+
"grad_norm": 15.390687236299986,
|
| 164 |
+
"learning_rate": 9.466913223222465e-07,
|
| 165 |
+
"logits/chosen": -2.28125,
|
| 166 |
+
"logits/rejected": -2.25,
|
| 167 |
+
"logps/chosen": -0.51171875,
|
| 168 |
+
"logps/rejected": -0.51953125,
|
| 169 |
+
"loss": 0.8121,
|
| 170 |
+
"rewards/accuracies": 0.4625000059604645,
|
| 171 |
+
"rewards/chosen": -1.28125,
|
| 172 |
+
"rewards/margins": 0.0235595703125,
|
| 173 |
+
"rewards/rejected": -1.3046875,
|
| 174 |
+
"step": 55
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.25716581837664076,
|
| 178 |
+
"grad_norm": 15.533470825408608,
|
| 179 |
+
"learning_rate": 9.285621438083997e-07,
|
| 180 |
+
"logits/chosen": -2.34375,
|
| 181 |
+
"logits/rejected": -2.34375,
|
| 182 |
+
"logps/chosen": -0.5234375,
|
| 183 |
+
"logps/rejected": -0.5703125,
|
| 184 |
+
"loss": 0.8084,
|
| 185 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 186 |
+
"rewards/chosen": -1.3125,
|
| 187 |
+
"rewards/margins": 0.12060546875,
|
| 188 |
+
"rewards/rejected": -1.4296875,
|
| 189 |
+
"step": 60
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.2785963032413608,
|
| 193 |
+
"grad_norm": 20.15952117472032,
|
| 194 |
+
"learning_rate": 9.080132930355566e-07,
|
| 195 |
+
"logits/chosen": -2.328125,
|
| 196 |
+
"logits/rejected": -2.34375,
|
| 197 |
+
"logps/chosen": -0.53515625,
|
| 198 |
+
"logps/rejected": -0.6015625,
|
| 199 |
+
"loss": 0.7938,
|
| 200 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 201 |
+
"rewards/chosen": -1.34375,
|
| 202 |
+
"rewards/margins": 0.1640625,
|
| 203 |
+
"rewards/rejected": -1.5078125,
|
| 204 |
+
"step": 65
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.3000267881060809,
|
| 208 |
+
"grad_norm": 20.257235368327233,
|
| 209 |
+
"learning_rate": 8.851607893136064e-07,
|
| 210 |
+
"logits/chosen": -2.390625,
|
| 211 |
+
"logits/rejected": -2.40625,
|
| 212 |
+
"logps/chosen": -0.55078125,
|
| 213 |
+
"logps/rejected": -0.625,
|
| 214 |
+
"loss": 0.7934,
|
| 215 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 216 |
+
"rewards/chosen": -1.3828125,
|
| 217 |
+
"rewards/margins": 0.18359375,
|
| 218 |
+
"rewards/rejected": -1.5625,
|
| 219 |
+
"step": 70
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.32145727297080096,
|
| 223 |
+
"grad_norm": 20.838334942567865,
|
| 224 |
+
"learning_rate": 8.601336584328658e-07,
|
| 225 |
+
"logits/chosen": -2.5625,
|
| 226 |
+
"logits/rejected": -2.53125,
|
| 227 |
+
"logps/chosen": -0.578125,
|
| 228 |
+
"logps/rejected": -0.6328125,
|
| 229 |
+
"loss": 0.7834,
|
| 230 |
+
"rewards/accuracies": 0.59375,
|
| 231 |
+
"rewards/chosen": -1.453125,
|
| 232 |
+
"rewards/margins": 0.1337890625,
|
| 233 |
+
"rewards/rejected": -1.5859375,
|
| 234 |
+
"step": 75
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.342887757835521,
|
| 238 |
+
"grad_norm": 20.73361765612086,
|
| 239 |
+
"learning_rate": 8.330732041813366e-07,
|
| 240 |
+
"logits/chosen": -2.515625,
|
| 241 |
+
"logits/rejected": -2.515625,
|
| 242 |
+
"logps/chosen": -0.5859375,
|
| 243 |
+
"logps/rejected": -0.63671875,
|
| 244 |
+
"loss": 0.7917,
|
| 245 |
+
"rewards/accuracies": 0.550000011920929,
|
| 246 |
+
"rewards/chosen": -1.4609375,
|
| 247 |
+
"rewards/margins": 0.130859375,
|
| 248 |
+
"rewards/rejected": -1.59375,
|
| 249 |
+
"step": 80
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.3643182427002411,
|
| 253 |
+
"grad_norm": 19.794702155542176,
|
| 254 |
+
"learning_rate": 8.041322105400921e-07,
|
| 255 |
+
"logits/chosen": -2.53125,
|
| 256 |
+
"logits/rejected": -2.546875,
|
| 257 |
+
"logps/chosen": -0.6171875,
|
| 258 |
+
"logps/rejected": -0.625,
|
| 259 |
+
"loss": 0.793,
|
| 260 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 261 |
+
"rewards/chosen": -1.546875,
|
| 262 |
+
"rewards/margins": 0.01544189453125,
|
| 263 |
+
"rewards/rejected": -1.5625,
|
| 264 |
+
"step": 85
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.38574872756496115,
|
| 268 |
+
"grad_norm": 19.137565727745777,
|
| 269 |
+
"learning_rate": 7.734740790612136e-07,
|
| 270 |
+
"logits/chosen": -2.5,
|
| 271 |
+
"logits/rejected": -2.5,
|
| 272 |
+
"logps/chosen": -0.58203125,
|
| 273 |
+
"logps/rejected": -0.6328125,
|
| 274 |
+
"loss": 0.7773,
|
| 275 |
+
"rewards/accuracies": 0.625,
|
| 276 |
+
"rewards/chosen": -1.453125,
|
| 277 |
+
"rewards/margins": 0.1240234375,
|
| 278 |
+
"rewards/rejected": -1.578125,
|
| 279 |
+
"step": 90
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.40717921242968125,
|
| 283 |
+
"grad_norm": 18.51359957650475,
|
| 284 |
+
"learning_rate": 7.412719062986631e-07,
|
| 285 |
+
"logits/chosen": -2.53125,
|
| 286 |
+
"logits/rejected": -2.5,
|
| 287 |
+
"logps/chosen": -0.6171875,
|
| 288 |
+
"logps/rejected": -0.6640625,
|
| 289 |
+
"loss": 0.7898,
|
| 290 |
+
"rewards/accuracies": 0.550000011920929,
|
| 291 |
+
"rewards/chosen": -1.5390625,
|
| 292 |
+
"rewards/margins": 0.12109375,
|
| 293 |
+
"rewards/rejected": -1.6640625,
|
| 294 |
+
"step": 95
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.4286096972944013,
|
| 298 |
+
"grad_norm": 20.516762220606736,
|
| 299 |
+
"learning_rate": 7.077075065009433e-07,
|
| 300 |
+
"logits/chosen": -2.640625,
|
| 301 |
+
"logits/rejected": -2.625,
|
| 302 |
+
"logps/chosen": -0.65625,
|
| 303 |
+
"logps/rejected": -0.7265625,
|
| 304 |
+
"loss": 0.7906,
|
| 305 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 306 |
+
"rewards/chosen": -1.6484375,
|
| 307 |
+
"rewards/margins": 0.1708984375,
|
| 308 |
+
"rewards/rejected": -1.8125,
|
| 309 |
+
"step": 100
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.45004018215912134,
|
| 313 |
+
"grad_norm": 23.39610952973476,
|
| 314 |
+
"learning_rate": 6.72970385083438e-07,
|
| 315 |
+
"logits/chosen": -2.65625,
|
| 316 |
+
"logits/rejected": -2.625,
|
| 317 |
+
"logps/chosen": -0.671875,
|
| 318 |
+
"logps/rejected": -0.734375,
|
| 319 |
+
"loss": 0.7652,
|
| 320 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 321 |
+
"rewards/chosen": -1.6796875,
|
| 322 |
+
"rewards/margins": 0.1611328125,
|
| 323 |
+
"rewards/rejected": -1.8359375,
|
| 324 |
+
"step": 105
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.47147066702384144,
|
| 328 |
+
"grad_norm": 21.60030365380359,
|
| 329 |
+
"learning_rate": 6.372566686762426e-07,
|
| 330 |
+
"logits/chosen": -2.671875,
|
| 331 |
+
"logits/rejected": -2.703125,
|
| 332 |
+
"logps/chosen": -0.68359375,
|
| 333 |
+
"logps/rejected": -0.75,
|
| 334 |
+
"loss": 0.7744,
|
| 335 |
+
"rewards/accuracies": 0.643750011920929,
|
| 336 |
+
"rewards/chosen": -1.7109375,
|
| 337 |
+
"rewards/margins": 0.1708984375,
|
| 338 |
+
"rewards/rejected": -1.875,
|
| 339 |
+
"step": 110
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.4929011518885615,
|
| 343 |
+
"grad_norm": 27.735737738499445,
|
| 344 |
+
"learning_rate": 6.00767997788451e-07,
|
| 345 |
+
"logits/chosen": -2.609375,
|
| 346 |
+
"logits/rejected": -2.625,
|
| 347 |
+
"logps/chosen": -0.76171875,
|
| 348 |
+
"logps/rejected": -0.80859375,
|
| 349 |
+
"loss": 0.7672,
|
| 350 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 351 |
+
"rewards/chosen": -1.90625,
|
| 352 |
+
"rewards/margins": 0.11083984375,
|
| 353 |
+
"rewards/rejected": -2.015625,
|
| 354 |
+
"step": 115
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.5143316367532815,
|
| 358 |
+
"grad_norm": 24.358513667557904,
|
| 359 |
+
"learning_rate": 5.637103883409525e-07,
|
| 360 |
+
"logits/chosen": -2.59375,
|
| 361 |
+
"logits/rejected": -2.59375,
|
| 362 |
+
"logps/chosen": -0.81640625,
|
| 363 |
+
"logps/rejected": -0.87890625,
|
| 364 |
+
"loss": 0.7625,
|
| 365 |
+
"rewards/accuracies": 0.606249988079071,
|
| 366 |
+
"rewards/chosen": -2.03125,
|
| 367 |
+
"rewards/margins": 0.1640625,
|
| 368 |
+
"rewards/rejected": -2.203125,
|
| 369 |
+
"step": 120
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.5357621216180016,
|
| 373 |
+
"grad_norm": 25.06386819321774,
|
| 374 |
+
"learning_rate": 5.262930684955438e-07,
|
| 375 |
+
"logits/chosen": -2.46875,
|
| 376 |
+
"logits/rejected": -2.453125,
|
| 377 |
+
"logps/chosen": -0.90234375,
|
| 378 |
+
"logps/rejected": -1.0390625,
|
| 379 |
+
"loss": 0.7596,
|
| 380 |
+
"rewards/accuracies": 0.643750011920929,
|
| 381 |
+
"rewards/chosen": -2.25,
|
| 382 |
+
"rewards/margins": 0.333984375,
|
| 383 |
+
"rewards/rejected": -2.59375,
|
| 384 |
+
"step": 125
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.5571926064827216,
|
| 388 |
+
"grad_norm": 30.74239972927362,
|
| 389 |
+
"learning_rate": 4.88727297347654e-07,
|
| 390 |
+
"logits/chosen": -2.5625,
|
| 391 |
+
"logits/rejected": -2.546875,
|
| 392 |
+
"logps/chosen": -0.859375,
|
| 393 |
+
"logps/rejected": -0.9375,
|
| 394 |
+
"loss": 0.7626,
|
| 395 |
+
"rewards/accuracies": 0.637499988079071,
|
| 396 |
+
"rewards/chosen": -2.140625,
|
| 397 |
+
"rewards/margins": 0.1953125,
|
| 398 |
+
"rewards/rejected": -2.34375,
|
| 399 |
+
"step": 130
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.5786230913474417,
|
| 403 |
+
"grad_norm": 28.54486073469317,
|
| 404 |
+
"learning_rate": 4.512251721523659e-07,
|
| 405 |
+
"logits/chosen": -2.53125,
|
| 406 |
+
"logits/rejected": -2.546875,
|
| 407 |
+
"logps/chosen": -0.984375,
|
| 408 |
+
"logps/rejected": -1.078125,
|
| 409 |
+
"loss": 0.7557,
|
| 410 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 411 |
+
"rewards/chosen": -2.46875,
|
| 412 |
+
"rewards/margins": 0.2333984375,
|
| 413 |
+
"rewards/rejected": -2.6875,
|
| 414 |
+
"step": 135
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.6000535762121618,
|
| 418 |
+
"grad_norm": 34.81427967339666,
|
| 419 |
+
"learning_rate": 4.139984308181708e-07,
|
| 420 |
+
"logits/chosen": -2.5,
|
| 421 |
+
"logits/rejected": -2.5,
|
| 422 |
+
"logps/chosen": -1.0078125,
|
| 423 |
+
"logps/rejected": -1.1640625,
|
| 424 |
+
"loss": 0.7408,
|
| 425 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 426 |
+
"rewards/chosen": -2.515625,
|
| 427 |
+
"rewards/margins": 0.3984375,
|
| 428 |
+
"rewards/rejected": -2.90625,
|
| 429 |
+
"step": 140
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.6214840610768818,
|
| 433 |
+
"grad_norm": 36.00644023249742,
|
| 434 |
+
"learning_rate": 3.772572564296004e-07,
|
| 435 |
+
"logits/chosen": -2.5,
|
| 436 |
+
"logits/rejected": -2.484375,
|
| 437 |
+
"logps/chosen": -1.0859375,
|
| 438 |
+
"logps/rejected": -1.2109375,
|
| 439 |
+
"loss": 0.7467,
|
| 440 |
+
"rewards/accuracies": 0.643750011920929,
|
| 441 |
+
"rewards/chosen": -2.703125,
|
| 442 |
+
"rewards/margins": 0.31640625,
|
| 443 |
+
"rewards/rejected": -3.015625,
|
| 444 |
+
"step": 145
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.6429145459416019,
|
| 448 |
+
"grad_norm": 32.41309299717272,
|
| 449 |
+
"learning_rate": 3.412090905484337e-07,
|
| 450 |
+
"logits/chosen": -2.453125,
|
| 451 |
+
"logits/rejected": -2.453125,
|
| 452 |
+
"logps/chosen": -1.109375,
|
| 453 |
+
"logps/rejected": -1.3046875,
|
| 454 |
+
"loss": 0.7513,
|
| 455 |
+
"rewards/accuracies": 0.731249988079071,
|
| 456 |
+
"rewards/chosen": -2.78125,
|
| 457 |
+
"rewards/margins": 0.4765625,
|
| 458 |
+
"rewards/rejected": -3.25,
|
| 459 |
+
"step": 150
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.664345030806322,
|
| 463 |
+
"grad_norm": 36.106876811155786,
|
| 464 |
+
"learning_rate": 3.060574619936075e-07,
|
| 465 |
+
"logits/chosen": -2.5,
|
| 466 |
+
"logits/rejected": -2.484375,
|
| 467 |
+
"logps/chosen": -1.1171875,
|
| 468 |
+
"logps/rejected": -1.2421875,
|
| 469 |
+
"loss": 0.7541,
|
| 470 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 471 |
+
"rewards/chosen": -2.796875,
|
| 472 |
+
"rewards/margins": 0.31640625,
|
| 473 |
+
"rewards/rejected": -3.109375,
|
| 474 |
+
"step": 155
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.685775515671042,
|
| 478 |
+
"grad_norm": 29.82009232975657,
|
| 479 |
+
"learning_rate": 2.720008377125682e-07,
|
| 480 |
+
"logits/chosen": -2.515625,
|
| 481 |
+
"logits/rejected": -2.53125,
|
| 482 |
+
"logps/chosen": -1.0859375,
|
| 483 |
+
"logps/rejected": -1.1796875,
|
| 484 |
+
"loss": 0.7298,
|
| 485 |
+
"rewards/accuracies": 0.699999988079071,
|
| 486 |
+
"rewards/chosen": -2.71875,
|
| 487 |
+
"rewards/margins": 0.2412109375,
|
| 488 |
+
"rewards/rejected": -2.953125,
|
| 489 |
+
"step": 160
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.7072060005357621,
|
| 493 |
+
"grad_norm": 33.878683796382994,
|
| 494 |
+
"learning_rate": 2.3923150223207173e-07,
|
| 495 |
+
"logits/chosen": -2.5625,
|
| 496 |
+
"logits/rejected": -2.578125,
|
| 497 |
+
"logps/chosen": -1.0625,
|
| 498 |
+
"logps/rejected": -1.1875,
|
| 499 |
+
"loss": 0.7428,
|
| 500 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 501 |
+
"rewards/chosen": -2.640625,
|
| 502 |
+
"rewards/margins": 0.318359375,
|
| 503 |
+
"rewards/rejected": -2.96875,
|
| 504 |
+
"step": 165
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 0.7286364854004822,
|
| 508 |
+
"grad_norm": 28.349316658677726,
|
| 509 |
+
"learning_rate": 2.0793447201508286e-07,
|
| 510 |
+
"logits/chosen": -2.53125,
|
| 511 |
+
"logits/rejected": -2.546875,
|
| 512 |
+
"logps/chosen": -1.140625,
|
| 513 |
+
"logps/rejected": -1.2734375,
|
| 514 |
+
"loss": 0.7299,
|
| 515 |
+
"rewards/accuracies": 0.71875,
|
| 516 |
+
"rewards/chosen": -2.859375,
|
| 517 |
+
"rewards/margins": 0.333984375,
|
| 518 |
+
"rewards/rejected": -3.1875,
|
| 519 |
+
"step": 170
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.7500669702652023,
|
| 523 |
+
"grad_norm": 29.876505430358502,
|
| 524 |
+
"learning_rate": 1.7828645085333644e-07,
|
| 525 |
+
"logits/chosen": -2.609375,
|
| 526 |
+
"logits/rejected": -2.59375,
|
| 527 |
+
"logps/chosen": -1.1015625,
|
| 528 |
+
"logps/rejected": -1.25,
|
| 529 |
+
"loss": 0.7457,
|
| 530 |
+
"rewards/accuracies": 0.65625,
|
| 531 |
+
"rewards/chosen": -2.75,
|
| 532 |
+
"rewards/margins": 0.3671875,
|
| 533 |
+
"rewards/rejected": -3.125,
|
| 534 |
+
"step": 175
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.7714974551299223,
|
| 538 |
+
"grad_norm": 27.71938507760261,
|
| 539 |
+
"learning_rate": 1.5045483219344385e-07,
|
| 540 |
+
"logits/chosen": -2.546875,
|
| 541 |
+
"logits/rejected": -2.546875,
|
| 542 |
+
"logps/chosen": -1.125,
|
| 543 |
+
"logps/rejected": -1.28125,
|
| 544 |
+
"loss": 0.7361,
|
| 545 |
+
"rewards/accuracies": 0.675000011920929,
|
| 546 |
+
"rewards/chosen": -2.8125,
|
| 547 |
+
"rewards/margins": 0.388671875,
|
| 548 |
+
"rewards/rejected": -3.203125,
|
| 549 |
+
"step": 180
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.7929279399946424,
|
| 553 |
+
"grad_norm": 27.85983890657696,
|
| 554 |
+
"learning_rate": 1.2459675402943288e-07,
|
| 555 |
+
"logits/chosen": -2.578125,
|
| 556 |
+
"logits/rejected": -2.5625,
|
| 557 |
+
"logps/chosen": -1.1328125,
|
| 558 |
+
"logps/rejected": -1.21875,
|
| 559 |
+
"loss": 0.7261,
|
| 560 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 561 |
+
"rewards/chosen": -2.828125,
|
| 562 |
+
"rewards/margins": 0.21484375,
|
| 563 |
+
"rewards/rejected": -3.046875,
|
| 564 |
+
"step": 185
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.8143584248593625,
|
| 568 |
+
"grad_norm": 37.85316797275618,
|
| 569 |
+
"learning_rate": 1.0085821169782199e-07,
|
| 570 |
+
"logits/chosen": -2.578125,
|
| 571 |
+
"logits/rejected": -2.59375,
|
| 572 |
+
"logps/chosen": -1.1796875,
|
| 573 |
+
"logps/rejected": -1.265625,
|
| 574 |
+
"loss": 0.7512,
|
| 575 |
+
"rewards/accuracies": 0.612500011920929,
|
| 576 |
+
"rewards/chosen": -2.953125,
|
| 577 |
+
"rewards/margins": 0.2255859375,
|
| 578 |
+
"rewards/rejected": -3.171875,
|
| 579 |
+
"step": 190
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.8357889097240825,
|
| 583 |
+
"grad_norm": 36.34792913013898,
|
| 584 |
+
"learning_rate": 7.937323358440934e-08,
|
| 585 |
+
"logits/chosen": -2.5625,
|
| 586 |
+
"logits/rejected": -2.5625,
|
| 587 |
+
"logps/chosen": -1.15625,
|
| 588 |
+
"logps/rejected": -1.265625,
|
| 589 |
+
"loss": 0.7437,
|
| 590 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 591 |
+
"rewards/chosen": -2.890625,
|
| 592 |
+
"rewards/margins": 0.27734375,
|
| 593 |
+
"rewards/rejected": -3.171875,
|
| 594 |
+
"step": 195
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 0.8572193945888026,
|
| 598 |
+
"grad_norm": 38.47006137486786,
|
| 599 |
+
"learning_rate": 6.026312439675551e-08,
|
| 600 |
+
"logits/chosen": -2.609375,
|
| 601 |
+
"logits/rejected": -2.609375,
|
| 602 |
+
"logps/chosen": -1.1328125,
|
| 603 |
+
"logps/rejected": -1.3359375,
|
| 604 |
+
"loss": 0.7203,
|
| 605 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 606 |
+
"rewards/chosen": -2.828125,
|
| 607 |
+
"rewards/margins": 0.5078125,
|
| 608 |
+
"rewards/rejected": -3.34375,
|
| 609 |
+
"step": 200
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.8786498794535227,
|
| 613 |
+
"grad_norm": 32.23204631755238,
|
| 614 |
+
"learning_rate": 4.3635780274861864e-08,
|
| 615 |
+
"logits/chosen": -2.59375,
|
| 616 |
+
"logits/rejected": -2.59375,
|
| 617 |
+
"logps/chosen": -1.1328125,
|
| 618 |
+
"logps/rejected": -1.2890625,
|
| 619 |
+
"loss": 0.7247,
|
| 620 |
+
"rewards/accuracies": 0.731249988079071,
|
| 621 |
+
"rewards/chosen": -2.828125,
|
| 622 |
+
"rewards/margins": 0.40625,
|
| 623 |
+
"rewards/rejected": -3.234375,
|
| 624 |
+
"step": 205
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 0.9000803643182427,
|
| 628 |
+
"grad_norm": 34.213864254754874,
|
| 629 |
+
"learning_rate": 2.958507960694784e-08,
|
| 630 |
+
"logits/chosen": -2.546875,
|
| 631 |
+
"logits/rejected": -2.5625,
|
| 632 |
+
"logps/chosen": -1.2265625,
|
| 633 |
+
"logps/rejected": -1.453125,
|
| 634 |
+
"loss": 0.7235,
|
| 635 |
+
"rewards/accuracies": 0.706250011920929,
|
| 636 |
+
"rewards/chosen": -3.078125,
|
| 637 |
+
"rewards/margins": 0.55078125,
|
| 638 |
+
"rewards/rejected": -3.625,
|
| 639 |
+
"step": 210
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.9215108491829628,
|
| 643 |
+
"grad_norm": 31.906095133000242,
|
| 644 |
+
"learning_rate": 1.8190352989793322e-08,
|
| 645 |
+
"logits/chosen": -2.578125,
|
| 646 |
+
"logits/rejected": -2.578125,
|
| 647 |
+
"logps/chosen": -1.1640625,
|
| 648 |
+
"logps/rejected": -1.296875,
|
| 649 |
+
"loss": 0.7316,
|
| 650 |
+
"rewards/accuracies": 0.637499988079071,
|
| 651 |
+
"rewards/chosen": -2.90625,
|
| 652 |
+
"rewards/margins": 0.333984375,
|
| 653 |
+
"rewards/rejected": -3.25,
|
| 654 |
+
"step": 215
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.9429413340476829,
|
| 658 |
+
"grad_norm": 35.354867146517954,
|
| 659 |
+
"learning_rate": 9.515935326265378e-09,
|
| 660 |
+
"logits/chosen": -2.5,
|
| 661 |
+
"logits/rejected": -2.5,
|
| 662 |
+
"logps/chosen": -1.25,
|
| 663 |
+
"logps/rejected": -1.421875,
|
| 664 |
+
"loss": 0.729,
|
| 665 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 666 |
+
"rewards/chosen": -3.125,
|
| 667 |
+
"rewards/margins": 0.42578125,
|
| 668 |
+
"rewards/rejected": -3.546875,
|
| 669 |
+
"step": 220
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.9643718189124029,
|
| 673 |
+
"grad_norm": 32.177844585654924,
|
| 674 |
+
"learning_rate": 3.6108025888958447e-09,
|
| 675 |
+
"logits/chosen": -2.578125,
|
| 676 |
+
"logits/rejected": -2.59375,
|
| 677 |
+
"logps/chosen": -1.140625,
|
| 678 |
+
"logps/rejected": -1.3203125,
|
| 679 |
+
"loss": 0.7181,
|
| 680 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 681 |
+
"rewards/chosen": -2.859375,
|
| 682 |
+
"rewards/margins": 0.443359375,
|
| 683 |
+
"rewards/rejected": -3.3125,
|
| 684 |
+
"step": 225
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 0.985802303777123,
|
| 688 |
+
"grad_norm": 36.438384324409185,
|
| 689 |
+
"learning_rate": 5.082953003528456e-10,
|
| 690 |
+
"logits/chosen": -2.546875,
|
| 691 |
+
"logits/rejected": -2.546875,
|
| 692 |
+
"logps/chosen": -1.2109375,
|
| 693 |
+
"logps/rejected": -1.3828125,
|
| 694 |
+
"loss": 0.7353,
|
| 695 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 696 |
+
"rewards/chosen": -3.015625,
|
| 697 |
+
"rewards/margins": 0.435546875,
|
| 698 |
+
"rewards/rejected": -3.453125,
|
| 699 |
+
"step": 230
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.998660594695955,
|
| 703 |
+
"step": 233,
|
| 704 |
+
"total_flos": 0.0,
|
| 705 |
+
"train_loss": 0.7728924853607309,
|
| 706 |
+
"train_runtime": 5369.4124,
|
| 707 |
+
"train_samples_per_second": 11.122,
|
| 708 |
+
"train_steps_per_second": 0.043
|
| 709 |
+
}
|
| 710 |
+
],
|
| 711 |
+
"logging_steps": 5,
|
| 712 |
+
"max_steps": 233,
|
| 713 |
+
"num_input_tokens_seen": 0,
|
| 714 |
+
"num_train_epochs": 1,
|
| 715 |
+
"save_steps": 1000000,
|
| 716 |
+
"stateful_callbacks": {
|
| 717 |
+
"TrainerControl": {
|
| 718 |
+
"args": {
|
| 719 |
+
"should_epoch_stop": false,
|
| 720 |
+
"should_evaluate": false,
|
| 721 |
+
"should_log": false,
|
| 722 |
+
"should_save": true,
|
| 723 |
+
"should_training_stop": true
|
| 724 |
+
},
|
| 725 |
+
"attributes": {}
|
| 726 |
+
}
|
| 727 |
+
},
|
| 728 |
+
"total_flos": 0.0,
|
| 729 |
+
"train_batch_size": 2,
|
| 730 |
+
"trial_name": null,
|
| 731 |
+
"trial_params": null
|
| 732 |
+
}
|