Model save
Browse files- README.md +79 -0
- all_results.json +9 -0
- generation_config.json +6 -0
- train_results.json +9 -0
- trainer_state.json +679 -0
README.md
ADDED
|
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
base_model: alignment-handbook/zephyr-7b-sft-full
|
| 4 |
+
tags:
|
| 5 |
+
- trl
|
| 6 |
+
- dpo
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
model-index:
|
| 9 |
+
- name: zephyr-7b-dpo-full-magpi-reward-scale-1
|
| 10 |
+
results: []
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 14 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 15 |
+
|
| 16 |
+
# zephyr-7b-dpo-full-magpi-reward-scale-1
|
| 17 |
+
|
| 18 |
+
This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on an unknown dataset.
|
| 19 |
+
It achieves the following results on the evaluation set:
|
| 20 |
+
- Loss: 0.0001
|
| 21 |
+
- Rewards/chosen: -2.4218
|
| 22 |
+
- Rewards/rejected: -70.3140
|
| 23 |
+
- Rewards/accuracies: 1.0
|
| 24 |
+
- Rewards/margins: 67.8922
|
| 25 |
+
- Logps/rejected: -7672.1890
|
| 26 |
+
- Logps/chosen: -609.1660
|
| 27 |
+
- Logits/rejected: 2.5753
|
| 28 |
+
- Logits/chosen: -0.2188
|
| 29 |
+
|
| 30 |
+
## Model description
|
| 31 |
+
|
| 32 |
+
More information needed
|
| 33 |
+
|
| 34 |
+
## Intended uses & limitations
|
| 35 |
+
|
| 36 |
+
More information needed
|
| 37 |
+
|
| 38 |
+
## Training and evaluation data
|
| 39 |
+
|
| 40 |
+
More information needed
|
| 41 |
+
|
| 42 |
+
## Training procedure
|
| 43 |
+
|
| 44 |
+
### Training hyperparameters
|
| 45 |
+
|
| 46 |
+
The following hyperparameters were used during training:
|
| 47 |
+
- learning_rate: 5e-07
|
| 48 |
+
- train_batch_size: 8
|
| 49 |
+
- eval_batch_size: 8
|
| 50 |
+
- seed: 55
|
| 51 |
+
- distributed_type: multi-GPU
|
| 52 |
+
- num_devices: 8
|
| 53 |
+
- gradient_accumulation_steps: 2
|
| 54 |
+
- total_train_batch_size: 128
|
| 55 |
+
- total_eval_batch_size: 64
|
| 56 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 57 |
+
- lr_scheduler_type: cosine
|
| 58 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 59 |
+
- num_epochs: 1
|
| 60 |
+
|
| 61 |
+
### Training results
|
| 62 |
+
|
| 63 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 64 |
+
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 65 |
+
| 0.0055 | 0.1420 | 50 | 0.0031 | -0.8850 | -42.4340 | 0.9960 | 41.5490 | -4884.1895 | -455.4866 | -2.5280 | -2.9176 |
|
| 66 |
+
| 0.0014 | 0.2841 | 100 | 0.0003 | -1.7467 | -57.2050 | 1.0 | 55.4583 | -6361.2881 | -541.6496 | -0.1436 | -2.5818 |
|
| 67 |
+
| 0.001 | 0.4261 | 150 | 0.0002 | -2.4059 | -66.6788 | 1.0 | 64.2728 | -7308.6660 | -607.5745 | 1.9422 | -1.9097 |
|
| 68 |
+
| 0.0018 | 0.5682 | 200 | 0.0002 | -2.1797 | -67.6111 | 1.0 | 65.4314 | -7401.8965 | -584.9550 | 2.3189 | -1.0395 |
|
| 69 |
+
| 0.0009 | 0.7102 | 250 | 0.0001 | -2.4169 | -67.5787 | 1.0 | 65.1618 | -7398.6553 | -608.6732 | 2.5585 | -0.3354 |
|
| 70 |
+
| 0.0009 | 0.8523 | 300 | 0.0001 | -2.4125 | -70.2443 | 1.0 | 67.8319 | -7665.2217 | -608.2272 | 2.5751 | -0.2549 |
|
| 71 |
+
| 0.0024 | 0.9943 | 350 | 0.0001 | -2.4218 | -70.3140 | 1.0 | 67.8922 | -7672.1890 | -609.1660 | 2.5753 | -0.2188 |
|
| 72 |
+
|
| 73 |
+
|
| 74 |
+
### Framework versions
|
| 75 |
+
|
| 76 |
+
- Transformers 4.44.0.dev0
|
| 77 |
+
- Pytorch 2.1.2
|
| 78 |
+
- Datasets 2.20.0
|
| 79 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.042729352997742935,
|
| 5 |
+
"train_runtime": 10052.8601,
|
| 6 |
+
"train_samples": 45000,
|
| 7 |
+
"train_samples_per_second": 4.476,
|
| 8 |
+
"train_steps_per_second": 0.035
|
| 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": "4.44.0.dev0"
|
| 6 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.042729352997742935,
|
| 5 |
+
"train_runtime": 10052.8601,
|
| 6 |
+
"train_samples": 45000,
|
| 7 |
+
"train_samples_per_second": 4.476,
|
| 8 |
+
"train_steps_per_second": 0.035
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,679 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"eval_steps": 50,
|
| 6 |
+
"global_step": 352,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.028409090909090908,
|
| 13 |
+
"grad_norm": 45.06143403082133,
|
| 14 |
+
"learning_rate": 1.3888888888888888e-07,
|
| 15 |
+
"logits/chosen": -2.859044075012207,
|
| 16 |
+
"logits/rejected": -2.64284086227417,
|
| 17 |
+
"logps/chosen": -390.5040588378906,
|
| 18 |
+
"logps/rejected": -607.8244018554688,
|
| 19 |
+
"loss": 0.6882,
|
| 20 |
+
"rewards/accuracies": 0.637499988079071,
|
| 21 |
+
"rewards/chosen": 0.002557548927143216,
|
| 22 |
+
"rewards/margins": 0.013725275173783302,
|
| 23 |
+
"rewards/rejected": -0.01116772647947073,
|
| 24 |
+
"step": 10
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.056818181818181816,
|
| 28 |
+
"grad_norm": 19.067293358665005,
|
| 29 |
+
"learning_rate": 2.7777777777777776e-07,
|
| 30 |
+
"logits/chosen": -2.8423688411712646,
|
| 31 |
+
"logits/rejected": -2.6946520805358887,
|
| 32 |
+
"logps/chosen": -328.6228332519531,
|
| 33 |
+
"logps/rejected": -774.904541015625,
|
| 34 |
+
"loss": 0.5132,
|
| 35 |
+
"rewards/accuracies": 0.981249988079071,
|
| 36 |
+
"rewards/chosen": 0.08522307872772217,
|
| 37 |
+
"rewards/margins": 0.6598724722862244,
|
| 38 |
+
"rewards/rejected": -0.574649453163147,
|
| 39 |
+
"step": 20
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.08522727272727272,
|
| 43 |
+
"grad_norm": 4.736974625657813,
|
| 44 |
+
"learning_rate": 4.1666666666666667e-07,
|
| 45 |
+
"logits/chosen": -2.8571083545684814,
|
| 46 |
+
"logits/rejected": -2.6747167110443115,
|
| 47 |
+
"logps/chosen": -295.3691711425781,
|
| 48 |
+
"logps/rejected": -1112.830078125,
|
| 49 |
+
"loss": 0.1792,
|
| 50 |
+
"rewards/accuracies": 1.0,
|
| 51 |
+
"rewards/chosen": 0.36912840604782104,
|
| 52 |
+
"rewards/margins": 4.717885494232178,
|
| 53 |
+
"rewards/rejected": -4.348757266998291,
|
| 54 |
+
"step": 30
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.11363636363636363,
|
| 58 |
+
"grad_norm": 1.4117830566891867,
|
| 59 |
+
"learning_rate": 4.998023493068254e-07,
|
| 60 |
+
"logits/chosen": -2.834780216217041,
|
| 61 |
+
"logits/rejected": -2.644721508026123,
|
| 62 |
+
"logps/chosen": -314.90582275390625,
|
| 63 |
+
"logps/rejected": -2130.47900390625,
|
| 64 |
+
"loss": 0.0415,
|
| 65 |
+
"rewards/accuracies": 1.0,
|
| 66 |
+
"rewards/chosen": 0.31849661469459534,
|
| 67 |
+
"rewards/margins": 14.544331550598145,
|
| 68 |
+
"rewards/rejected": -14.225835800170898,
|
| 69 |
+
"step": 40
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.14204545454545456,
|
| 73 |
+
"grad_norm": 0.09320830768011595,
|
| 74 |
+
"learning_rate": 4.975823666181255e-07,
|
| 75 |
+
"logits/chosen": -2.814044713973999,
|
| 76 |
+
"logits/rejected": -2.548596143722534,
|
| 77 |
+
"logps/chosen": -437.7023010253906,
|
| 78 |
+
"logps/rejected": -4283.9775390625,
|
| 79 |
+
"loss": 0.0055,
|
| 80 |
+
"rewards/accuracies": 1.0,
|
| 81 |
+
"rewards/chosen": -0.7506707906723022,
|
| 82 |
+
"rewards/margins": 35.5496940612793,
|
| 83 |
+
"rewards/rejected": -36.30036926269531,
|
| 84 |
+
"step": 50
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.14204545454545456,
|
| 88 |
+
"eval_logits/chosen": -2.9176247119903564,
|
| 89 |
+
"eval_logits/rejected": -2.528038263320923,
|
| 90 |
+
"eval_logps/chosen": -455.486572265625,
|
| 91 |
+
"eval_logps/rejected": -4884.189453125,
|
| 92 |
+
"eval_loss": 0.0030528793577104807,
|
| 93 |
+
"eval_rewards/accuracies": 0.9959677457809448,
|
| 94 |
+
"eval_rewards/chosen": -0.8850486874580383,
|
| 95 |
+
"eval_rewards/margins": 41.54896545410156,
|
| 96 |
+
"eval_rewards/rejected": -42.43401336669922,
|
| 97 |
+
"eval_runtime": 193.4773,
|
| 98 |
+
"eval_samples_per_second": 20.183,
|
| 99 |
+
"eval_steps_per_second": 0.32,
|
| 100 |
+
"step": 50
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.17045454545454544,
|
| 104 |
+
"grad_norm": 1.0625936242535339,
|
| 105 |
+
"learning_rate": 4.929173350101024e-07,
|
| 106 |
+
"logits/chosen": -2.990762710571289,
|
| 107 |
+
"logits/rejected": -2.484398603439331,
|
| 108 |
+
"logps/chosen": -463.26251220703125,
|
| 109 |
+
"logps/rejected": -5167.1591796875,
|
| 110 |
+
"loss": 0.0033,
|
| 111 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 112 |
+
"rewards/chosen": -1.033675193786621,
|
| 113 |
+
"rewards/margins": 44.20937728881836,
|
| 114 |
+
"rewards/rejected": -45.2430534362793,
|
| 115 |
+
"step": 60
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.19886363636363635,
|
| 119 |
+
"grad_norm": 0.16389340246703632,
|
| 120 |
+
"learning_rate": 4.858533249305336e-07,
|
| 121 |
+
"logits/chosen": -2.814481735229492,
|
| 122 |
+
"logits/rejected": -1.7779146432876587,
|
| 123 |
+
"logps/chosen": -507.0504455566406,
|
| 124 |
+
"logps/rejected": -5547.2275390625,
|
| 125 |
+
"loss": 0.0031,
|
| 126 |
+
"rewards/accuracies": 1.0,
|
| 127 |
+
"rewards/chosen": -1.2788976430892944,
|
| 128 |
+
"rewards/margins": 48.066017150878906,
|
| 129 |
+
"rewards/rejected": -49.34491729736328,
|
| 130 |
+
"step": 70
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.22727272727272727,
|
| 134 |
+
"grad_norm": 15.588141585226081,
|
| 135 |
+
"learning_rate": 4.764600984163808e-07,
|
| 136 |
+
"logits/chosen": -3.000598669052124,
|
| 137 |
+
"logits/rejected": -1.848249077796936,
|
| 138 |
+
"logps/chosen": -532.5420532226562,
|
| 139 |
+
"logps/rejected": -6873.24853515625,
|
| 140 |
+
"loss": 0.0039,
|
| 141 |
+
"rewards/accuracies": 1.0,
|
| 142 |
+
"rewards/chosen": -1.2622119188308716,
|
| 143 |
+
"rewards/margins": 60.732948303222656,
|
| 144 |
+
"rewards/rejected": -61.99515914916992,
|
| 145 |
+
"step": 80
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.2556818181818182,
|
| 149 |
+
"grad_norm": 0.02323120317433541,
|
| 150 |
+
"learning_rate": 4.6483042014491527e-07,
|
| 151 |
+
"logits/chosen": -3.130722761154175,
|
| 152 |
+
"logits/rejected": -2.385204792022705,
|
| 153 |
+
"logps/chosen": -520.1354370117188,
|
| 154 |
+
"logps/rejected": -6028.5537109375,
|
| 155 |
+
"loss": 0.0138,
|
| 156 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 157 |
+
"rewards/chosen": -1.5802205801010132,
|
| 158 |
+
"rewards/margins": 52.33146286010742,
|
| 159 |
+
"rewards/rejected": -53.91168212890625,
|
| 160 |
+
"step": 90
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 0.2840909090909091,
|
| 164 |
+
"grad_norm": 0.16442080948160046,
|
| 165 |
+
"learning_rate": 4.510791413176912e-07,
|
| 166 |
+
"logits/chosen": -2.8645615577697754,
|
| 167 |
+
"logits/rejected": -1.0432738065719604,
|
| 168 |
+
"logps/chosen": -532.5404052734375,
|
| 169 |
+
"logps/rejected": -6479.7802734375,
|
| 170 |
+
"loss": 0.0014,
|
| 171 |
+
"rewards/accuracies": 1.0,
|
| 172 |
+
"rewards/chosen": -1.6307029724121094,
|
| 173 |
+
"rewards/margins": 56.10929489135742,
|
| 174 |
+
"rewards/rejected": -57.73999786376953,
|
| 175 |
+
"step": 100
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.2840909090909091,
|
| 179 |
+
"eval_logits/chosen": -2.5817885398864746,
|
| 180 |
+
"eval_logits/rejected": -0.1435869336128235,
|
| 181 |
+
"eval_logps/chosen": -541.6495971679688,
|
| 182 |
+
"eval_logps/rejected": -6361.2880859375,
|
| 183 |
+
"eval_loss": 0.0003222826635465026,
|
| 184 |
+
"eval_rewards/accuracies": 1.0,
|
| 185 |
+
"eval_rewards/chosen": -1.7466782331466675,
|
| 186 |
+
"eval_rewards/margins": 55.458316802978516,
|
| 187 |
+
"eval_rewards/rejected": -57.20499801635742,
|
| 188 |
+
"eval_runtime": 190.7671,
|
| 189 |
+
"eval_samples_per_second": 20.47,
|
| 190 |
+
"eval_steps_per_second": 0.325,
|
| 191 |
+
"step": 100
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.3125,
|
| 195 |
+
"grad_norm": 0.03669115199922144,
|
| 196 |
+
"learning_rate": 4.353420654246546e-07,
|
| 197 |
+
"logits/chosen": -2.299072742462158,
|
| 198 |
+
"logits/rejected": 0.060857243835926056,
|
| 199 |
+
"logps/chosen": -552.1079711914062,
|
| 200 |
+
"logps/rejected": -6287.04736328125,
|
| 201 |
+
"loss": 0.0024,
|
| 202 |
+
"rewards/accuracies": 1.0,
|
| 203 |
+
"rewards/chosen": -1.766867995262146,
|
| 204 |
+
"rewards/margins": 55.110923767089844,
|
| 205 |
+
"rewards/rejected": -56.87778854370117,
|
| 206 |
+
"step": 110
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.3409090909090909,
|
| 210 |
+
"grad_norm": 0.03951441158135731,
|
| 211 |
+
"learning_rate": 4.177746070897592e-07,
|
| 212 |
+
"logits/chosen": -2.019587278366089,
|
| 213 |
+
"logits/rejected": 0.7790449261665344,
|
| 214 |
+
"logps/chosen": -561.4166870117188,
|
| 215 |
+
"logps/rejected": -7041.1259765625,
|
| 216 |
+
"loss": 0.0033,
|
| 217 |
+
"rewards/accuracies": 1.0,
|
| 218 |
+
"rewards/chosen": -1.798998236656189,
|
| 219 |
+
"rewards/margins": 62.23997116088867,
|
| 220 |
+
"rewards/rejected": -64.03898620605469,
|
| 221 |
+
"step": 120
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.3693181818181818,
|
| 225 |
+
"grad_norm": 0.3141553784150294,
|
| 226 |
+
"learning_rate": 3.9855025724292763e-07,
|
| 227 |
+
"logits/chosen": -2.1000161170959473,
|
| 228 |
+
"logits/rejected": 1.278324842453003,
|
| 229 |
+
"logps/chosen": -563.4060668945312,
|
| 230 |
+
"logps/rejected": -7105.3544921875,
|
| 231 |
+
"loss": 0.0027,
|
| 232 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 233 |
+
"rewards/chosen": -1.929544448852539,
|
| 234 |
+
"rewards/margins": 62.7186164855957,
|
| 235 |
+
"rewards/rejected": -64.64815521240234,
|
| 236 |
+
"step": 130
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.3977272727272727,
|
| 240 |
+
"grad_norm": 0.5821229491227392,
|
| 241 |
+
"learning_rate": 3.7785886977585555e-07,
|
| 242 |
+
"logits/chosen": -2.022021532058716,
|
| 243 |
+
"logits/rejected": 1.353015422821045,
|
| 244 |
+
"logps/chosen": -552.7689208984375,
|
| 245 |
+
"logps/rejected": -7177.63427734375,
|
| 246 |
+
"loss": 0.0017,
|
| 247 |
+
"rewards/accuracies": 1.0,
|
| 248 |
+
"rewards/chosen": -1.867326021194458,
|
| 249 |
+
"rewards/margins": 63.489112854003906,
|
| 250 |
+
"rewards/rejected": -65.35643768310547,
|
| 251 |
+
"step": 140
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.42613636363636365,
|
| 255 |
+
"grad_norm": 0.7534850925649158,
|
| 256 |
+
"learning_rate": 3.5590478660213206e-07,
|
| 257 |
+
"logits/chosen": -1.8572208881378174,
|
| 258 |
+
"logits/rejected": 1.6534769535064697,
|
| 259 |
+
"logps/chosen": -621.11181640625,
|
| 260 |
+
"logps/rejected": -7095.47802734375,
|
| 261 |
+
"loss": 0.001,
|
| 262 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 263 |
+
"rewards/chosen": -2.4304866790771484,
|
| 264 |
+
"rewards/margins": 61.404510498046875,
|
| 265 |
+
"rewards/rejected": -63.834999084472656,
|
| 266 |
+
"step": 150
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.42613636363636365,
|
| 270 |
+
"eval_logits/chosen": -1.9097061157226562,
|
| 271 |
+
"eval_logits/rejected": 1.942201852798462,
|
| 272 |
+
"eval_logps/chosen": -607.574462890625,
|
| 273 |
+
"eval_logps/rejected": -7308.666015625,
|
| 274 |
+
"eval_loss": 0.0001972884201677516,
|
| 275 |
+
"eval_rewards/accuracies": 1.0,
|
| 276 |
+
"eval_rewards/chosen": -2.4059276580810547,
|
| 277 |
+
"eval_rewards/margins": 64.27284240722656,
|
| 278 |
+
"eval_rewards/rejected": -66.67876434326172,
|
| 279 |
+
"eval_runtime": 193.1671,
|
| 280 |
+
"eval_samples_per_second": 20.216,
|
| 281 |
+
"eval_steps_per_second": 0.321,
|
| 282 |
+
"step": 150
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.45454545454545453,
|
| 286 |
+
"grad_norm": 0.5761003352865638,
|
| 287 |
+
"learning_rate": 3.3290481963801696e-07,
|
| 288 |
+
"logits/chosen": -2.1488444805145264,
|
| 289 |
+
"logits/rejected": 1.8684966564178467,
|
| 290 |
+
"logps/chosen": -605.7991943359375,
|
| 291 |
+
"logps/rejected": -7330.78759765625,
|
| 292 |
+
"loss": 0.0002,
|
| 293 |
+
"rewards/accuracies": 1.0,
|
| 294 |
+
"rewards/chosen": -2.356612205505371,
|
| 295 |
+
"rewards/margins": 64.78636169433594,
|
| 296 |
+
"rewards/rejected": -67.14297485351562,
|
| 297 |
+
"step": 160
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.48295454545454547,
|
| 301 |
+
"grad_norm": 0.0136866819292242,
|
| 302 |
+
"learning_rate": 3.0908610963322626e-07,
|
| 303 |
+
"logits/chosen": -2.361186981201172,
|
| 304 |
+
"logits/rejected": 1.371411919593811,
|
| 305 |
+
"logps/chosen": -656.9031982421875,
|
| 306 |
+
"logps/rejected": -7962.04443359375,
|
| 307 |
+
"loss": 0.0059,
|
| 308 |
+
"rewards/accuracies": 1.0,
|
| 309 |
+
"rewards/chosen": -2.6928484439849854,
|
| 310 |
+
"rewards/margins": 69.64448547363281,
|
| 311 |
+
"rewards/rejected": -72.33734130859375,
|
| 312 |
+
"step": 170
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.5113636363636364,
|
| 316 |
+
"grad_norm": 0.00510201676789354,
|
| 317 |
+
"learning_rate": 2.846838829972671e-07,
|
| 318 |
+
"logits/chosen": -2.2552402019500732,
|
| 319 |
+
"logits/rejected": 1.9691696166992188,
|
| 320 |
+
"logps/chosen": -620.7752685546875,
|
| 321 |
+
"logps/rejected": -7178.3388671875,
|
| 322 |
+
"loss": 0.0021,
|
| 323 |
+
"rewards/accuracies": 1.0,
|
| 324 |
+
"rewards/chosen": -2.5987088680267334,
|
| 325 |
+
"rewards/margins": 63.054847717285156,
|
| 326 |
+
"rewards/rejected": -65.65354919433594,
|
| 327 |
+
"step": 180
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.5397727272727273,
|
| 331 |
+
"grad_norm": 0.2102394829040594,
|
| 332 |
+
"learning_rate": 2.5993912877423147e-07,
|
| 333 |
+
"logits/chosen": -1.9847707748413086,
|
| 334 |
+
"logits/rejected": 1.6557881832122803,
|
| 335 |
+
"logps/chosen": -555.5148315429688,
|
| 336 |
+
"logps/rejected": -7729.9189453125,
|
| 337 |
+
"loss": 0.0007,
|
| 338 |
+
"rewards/accuracies": 1.0,
|
| 339 |
+
"rewards/chosen": -2.0989906787872314,
|
| 340 |
+
"rewards/margins": 68.28409576416016,
|
| 341 |
+
"rewards/rejected": -70.38308715820312,
|
| 342 |
+
"step": 190
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 0.5681818181818182,
|
| 346 |
+
"grad_norm": 0.62932736058334,
|
| 347 |
+
"learning_rate": 2.3509621870754504e-07,
|
| 348 |
+
"logits/chosen": -1.2217730283737183,
|
| 349 |
+
"logits/rejected": 2.607821226119995,
|
| 350 |
+
"logps/chosen": -571.2921142578125,
|
| 351 |
+
"logps/rejected": -6882.54833984375,
|
| 352 |
+
"loss": 0.0018,
|
| 353 |
+
"rewards/accuracies": 1.0,
|
| 354 |
+
"rewards/chosen": -2.0616230964660645,
|
| 355 |
+
"rewards/margins": 60.45543670654297,
|
| 356 |
+
"rewards/rejected": -62.517051696777344,
|
| 357 |
+
"step": 200
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.5681818181818182,
|
| 361 |
+
"eval_logits/chosen": -1.0394700765609741,
|
| 362 |
+
"eval_logits/rejected": 2.318869113922119,
|
| 363 |
+
"eval_logps/chosen": -584.9549560546875,
|
| 364 |
+
"eval_logps/rejected": -7401.896484375,
|
| 365 |
+
"eval_loss": 0.0001572135806782171,
|
| 366 |
+
"eval_rewards/accuracies": 1.0,
|
| 367 |
+
"eval_rewards/chosen": -2.179731845855713,
|
| 368 |
+
"eval_rewards/margins": 65.43135070800781,
|
| 369 |
+
"eval_rewards/rejected": -67.61107635498047,
|
| 370 |
+
"eval_runtime": 192.1225,
|
| 371 |
+
"eval_samples_per_second": 20.326,
|
| 372 |
+
"eval_steps_per_second": 0.323,
|
| 373 |
+
"step": 200
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.5965909090909091,
|
| 377 |
+
"grad_norm": 0.5427913238664086,
|
| 378 |
+
"learning_rate": 2.1040049389819624e-07,
|
| 379 |
+
"logits/chosen": -1.0268203020095825,
|
| 380 |
+
"logits/rejected": 2.479326009750366,
|
| 381 |
+
"logps/chosen": -601.0950927734375,
|
| 382 |
+
"logps/rejected": -7280.90771484375,
|
| 383 |
+
"loss": 0.0028,
|
| 384 |
+
"rewards/accuracies": 1.0,
|
| 385 |
+
"rewards/chosen": -2.104518413543701,
|
| 386 |
+
"rewards/margins": 64.38551330566406,
|
| 387 |
+
"rewards/rejected": -66.49003601074219,
|
| 388 |
+
"step": 210
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.625,
|
| 392 |
+
"grad_norm": 2.0347197022348307,
|
| 393 |
+
"learning_rate": 1.8609584188988133e-07,
|
| 394 |
+
"logits/chosen": -0.5109957456588745,
|
| 395 |
+
"logits/rejected": 2.219853401184082,
|
| 396 |
+
"logps/chosen": -642.4463500976562,
|
| 397 |
+
"logps/rejected": -6641.6826171875,
|
| 398 |
+
"loss": 0.003,
|
| 399 |
+
"rewards/accuracies": 1.0,
|
| 400 |
+
"rewards/chosen": -2.7784571647644043,
|
| 401 |
+
"rewards/margins": 57.14200973510742,
|
| 402 |
+
"rewards/rejected": -59.92046356201172,
|
| 403 |
+
"step": 220
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.6534090909090909,
|
| 407 |
+
"grad_norm": 34.246565820645614,
|
| 408 |
+
"learning_rate": 1.624222881090439e-07,
|
| 409 |
+
"logits/chosen": -0.5791199207305908,
|
| 410 |
+
"logits/rejected": 2.0261969566345215,
|
| 411 |
+
"logps/chosen": -674.46044921875,
|
| 412 |
+
"logps/rejected": -7846.1572265625,
|
| 413 |
+
"loss": 0.0099,
|
| 414 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 415 |
+
"rewards/chosen": -2.7979581356048584,
|
| 416 |
+
"rewards/margins": 68.80779266357422,
|
| 417 |
+
"rewards/rejected": -71.60575103759766,
|
| 418 |
+
"step": 230
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 0.6818181818181818,
|
| 422 |
+
"grad_norm": 0.31954869640074296,
|
| 423 |
+
"learning_rate": 1.3961362544602212e-07,
|
| 424 |
+
"logits/chosen": -0.41727203130722046,
|
| 425 |
+
"logits/rejected": 2.4299378395080566,
|
| 426 |
+
"logps/chosen": -630.17919921875,
|
| 427 |
+
"logps/rejected": -6758.9619140625,
|
| 428 |
+
"loss": 0.001,
|
| 429 |
+
"rewards/accuracies": 1.0,
|
| 430 |
+
"rewards/chosen": -2.541667938232422,
|
| 431 |
+
"rewards/margins": 59.10721969604492,
|
| 432 |
+
"rewards/rejected": -61.648887634277344,
|
| 433 |
+
"step": 240
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 0.7102272727272727,
|
| 437 |
+
"grad_norm": 0.2930942520900322,
|
| 438 |
+
"learning_rate": 1.1789510538684522e-07,
|
| 439 |
+
"logits/chosen": -0.4175181984901428,
|
| 440 |
+
"logits/rejected": 2.409147024154663,
|
| 441 |
+
"logps/chosen": -591.2493896484375,
|
| 442 |
+
"logps/rejected": -8521.580078125,
|
| 443 |
+
"loss": 0.0009,
|
| 444 |
+
"rewards/accuracies": 1.0,
|
| 445 |
+
"rewards/chosen": -2.392158031463623,
|
| 446 |
+
"rewards/margins": 75.52781677246094,
|
| 447 |
+
"rewards/rejected": -77.91997528076172,
|
| 448 |
+
"step": 250
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 0.7102272727272727,
|
| 452 |
+
"eval_logits/chosen": -0.3353612720966339,
|
| 453 |
+
"eval_logits/rejected": 2.558471918106079,
|
| 454 |
+
"eval_logps/chosen": -608.6731567382812,
|
| 455 |
+
"eval_logps/rejected": -7398.6552734375,
|
| 456 |
+
"eval_loss": 0.0001336697314400226,
|
| 457 |
+
"eval_rewards/accuracies": 1.0,
|
| 458 |
+
"eval_rewards/chosen": -2.4169137477874756,
|
| 459 |
+
"eval_rewards/margins": 65.16175079345703,
|
| 460 |
+
"eval_rewards/rejected": -67.57865905761719,
|
| 461 |
+
"eval_runtime": 191.4599,
|
| 462 |
+
"eval_samples_per_second": 20.396,
|
| 463 |
+
"eval_steps_per_second": 0.324,
|
| 464 |
+
"step": 250
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.7386363636363636,
|
| 468 |
+
"grad_norm": 0.18266288675107914,
|
| 469 |
+
"learning_rate": 9.748121349736891e-08,
|
| 470 |
+
"logits/chosen": -0.2991776764392853,
|
| 471 |
+
"logits/rejected": 2.4152047634124756,
|
| 472 |
+
"logps/chosen": -644.2991333007812,
|
| 473 |
+
"logps/rejected": -7754.68359375,
|
| 474 |
+
"loss": 0.0006,
|
| 475 |
+
"rewards/accuracies": 1.0,
|
| 476 |
+
"rewards/chosen": -2.604710578918457,
|
| 477 |
+
"rewards/margins": 68.16084289550781,
|
| 478 |
+
"rewards/rejected": -70.76555633544922,
|
| 479 |
+
"step": 260
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.7670454545454546,
|
| 483 |
+
"grad_norm": 0.02843345452840696,
|
| 484 |
+
"learning_rate": 7.857355122839673e-08,
|
| 485 |
+
"logits/chosen": -0.5736527442932129,
|
| 486 |
+
"logits/rejected": 2.2506680488586426,
|
| 487 |
+
"logps/chosen": -621.7249145507812,
|
| 488 |
+
"logps/rejected": -7290.41015625,
|
| 489 |
+
"loss": 0.0005,
|
| 490 |
+
"rewards/accuracies": 1.0,
|
| 491 |
+
"rewards/chosen": -2.408754348754883,
|
| 492 |
+
"rewards/margins": 63.805747985839844,
|
| 493 |
+
"rewards/rejected": -66.21451568603516,
|
| 494 |
+
"step": 270
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.7954545454545454,
|
| 498 |
+
"grad_norm": 0.06722410834807425,
|
| 499 |
+
"learning_rate": 6.135884496044244e-08,
|
| 500 |
+
"logits/chosen": -0.5607927441596985,
|
| 501 |
+
"logits/rejected": 2.5057239532470703,
|
| 502 |
+
"logps/chosen": -604.9693603515625,
|
| 503 |
+
"logps/rejected": -7248.1767578125,
|
| 504 |
+
"loss": 0.001,
|
| 505 |
+
"rewards/accuracies": 1.0,
|
| 506 |
+
"rewards/chosen": -2.257964611053467,
|
| 507 |
+
"rewards/margins": 63.7749137878418,
|
| 508 |
+
"rewards/rejected": -66.03287506103516,
|
| 509 |
+
"step": 280
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 0.8238636363636364,
|
| 513 |
+
"grad_norm": 0.007877917419381247,
|
| 514 |
+
"learning_rate": 4.600710195020982e-08,
|
| 515 |
+
"logits/chosen": -0.44481128454208374,
|
| 516 |
+
"logits/rejected": 2.559147357940674,
|
| 517 |
+
"logps/chosen": -628.8250732421875,
|
| 518 |
+
"logps/rejected": -7324.6533203125,
|
| 519 |
+
"loss": 0.0027,
|
| 520 |
+
"rewards/accuracies": 1.0,
|
| 521 |
+
"rewards/chosen": -2.3968186378479004,
|
| 522 |
+
"rewards/margins": 64.43711853027344,
|
| 523 |
+
"rewards/rejected": -66.83393859863281,
|
| 524 |
+
"step": 290
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 0.8522727272727273,
|
| 528 |
+
"grad_norm": 0.5905211140332576,
|
| 529 |
+
"learning_rate": 3.2669931390104374e-08,
|
| 530 |
+
"logits/chosen": -0.4180712103843689,
|
| 531 |
+
"logits/rejected": 2.4567205905914307,
|
| 532 |
+
"logps/chosen": -577.669921875,
|
| 533 |
+
"logps/rejected": -7993.05859375,
|
| 534 |
+
"loss": 0.0009,
|
| 535 |
+
"rewards/accuracies": 1.0,
|
| 536 |
+
"rewards/chosen": -2.214153289794922,
|
| 537 |
+
"rewards/margins": 71.5081787109375,
|
| 538 |
+
"rewards/rejected": -73.72233581542969,
|
| 539 |
+
"step": 300
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 0.8522727272727273,
|
| 543 |
+
"eval_logits/chosen": -0.2549493908882141,
|
| 544 |
+
"eval_logits/rejected": 2.575054407119751,
|
| 545 |
+
"eval_logps/chosen": -608.2271728515625,
|
| 546 |
+
"eval_logps/rejected": -7665.2216796875,
|
| 547 |
+
"eval_loss": 7.371166429948062e-05,
|
| 548 |
+
"eval_rewards/accuracies": 1.0,
|
| 549 |
+
"eval_rewards/chosen": -2.412454843521118,
|
| 550 |
+
"eval_rewards/margins": 67.83187866210938,
|
| 551 |
+
"eval_rewards/rejected": -70.24433898925781,
|
| 552 |
+
"eval_runtime": 192.5021,
|
| 553 |
+
"eval_samples_per_second": 20.285,
|
| 554 |
+
"eval_steps_per_second": 0.322,
|
| 555 |
+
"step": 300
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.8806818181818182,
|
| 559 |
+
"grad_norm": 0.04237285499239183,
|
| 560 |
+
"learning_rate": 2.147904716149135e-08,
|
| 561 |
+
"logits/chosen": -0.1846945583820343,
|
| 562 |
+
"logits/rejected": 2.5058579444885254,
|
| 563 |
+
"logps/chosen": -589.5569458007812,
|
| 564 |
+
"logps/rejected": -7777.6474609375,
|
| 565 |
+
"loss": 0.0018,
|
| 566 |
+
"rewards/accuracies": 1.0,
|
| 567 |
+
"rewards/chosen": -2.2749037742614746,
|
| 568 |
+
"rewards/margins": 68.49655151367188,
|
| 569 |
+
"rewards/rejected": -70.77145385742188,
|
| 570 |
+
"step": 310
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 0.9090909090909091,
|
| 574 |
+
"grad_norm": 0.016783305893164634,
|
| 575 |
+
"learning_rate": 1.254496706805433e-08,
|
| 576 |
+
"logits/chosen": -0.39586615562438965,
|
| 577 |
+
"logits/rejected": 2.4627411365509033,
|
| 578 |
+
"logps/chosen": -622.9832763671875,
|
| 579 |
+
"logps/rejected": -7786.82421875,
|
| 580 |
+
"loss": 0.0003,
|
| 581 |
+
"rewards/accuracies": 1.0,
|
| 582 |
+
"rewards/chosen": -2.403102159500122,
|
| 583 |
+
"rewards/margins": 68.9453125,
|
| 584 |
+
"rewards/rejected": -71.3484115600586,
|
| 585 |
+
"step": 320
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.9375,
|
| 589 |
+
"grad_norm": 0.149850718402827,
|
| 590 |
+
"learning_rate": 5.955921395237318e-09,
|
| 591 |
+
"logits/chosen": -0.37074390053749084,
|
| 592 |
+
"logits/rejected": 2.418095588684082,
|
| 593 |
+
"logps/chosen": -577.0670776367188,
|
| 594 |
+
"logps/rejected": -7757.859375,
|
| 595 |
+
"loss": 0.0004,
|
| 596 |
+
"rewards/accuracies": 1.0,
|
| 597 |
+
"rewards/chosen": -2.2179901599884033,
|
| 598 |
+
"rewards/margins": 68.88307189941406,
|
| 599 |
+
"rewards/rejected": -71.10105895996094,
|
| 600 |
+
"step": 330
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 0.9659090909090909,
|
| 604 |
+
"grad_norm": 0.24861872079475503,
|
| 605 |
+
"learning_rate": 1.7769815745066474e-09,
|
| 606 |
+
"logits/chosen": -0.6141588687896729,
|
| 607 |
+
"logits/rejected": 2.2593562602996826,
|
| 608 |
+
"logps/chosen": -593.7769775390625,
|
| 609 |
+
"logps/rejected": -7168.8701171875,
|
| 610 |
+
"loss": 0.0006,
|
| 611 |
+
"rewards/accuracies": 1.0,
|
| 612 |
+
"rewards/chosen": -2.2479617595672607,
|
| 613 |
+
"rewards/margins": 63.35308074951172,
|
| 614 |
+
"rewards/rejected": -65.60104370117188,
|
| 615 |
+
"step": 340
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 0.9943181818181818,
|
| 619 |
+
"grad_norm": 0.890402381245667,
|
| 620 |
+
"learning_rate": 4.9417557483610875e-11,
|
| 621 |
+
"logits/chosen": -0.3333211839199066,
|
| 622 |
+
"logits/rejected": 2.404022455215454,
|
| 623 |
+
"logps/chosen": -621.5592041015625,
|
| 624 |
+
"logps/rejected": -7657.5458984375,
|
| 625 |
+
"loss": 0.0024,
|
| 626 |
+
"rewards/accuracies": 0.9937499761581421,
|
| 627 |
+
"rewards/chosen": -2.472914457321167,
|
| 628 |
+
"rewards/margins": 67.62696075439453,
|
| 629 |
+
"rewards/rejected": -70.0998764038086,
|
| 630 |
+
"step": 350
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 0.9943181818181818,
|
| 634 |
+
"eval_logits/chosen": -0.21884487569332123,
|
| 635 |
+
"eval_logits/rejected": 2.5753109455108643,
|
| 636 |
+
"eval_logps/chosen": -609.166015625,
|
| 637 |
+
"eval_logps/rejected": -7672.18896484375,
|
| 638 |
+
"eval_loss": 7.062770600896329e-05,
|
| 639 |
+
"eval_rewards/accuracies": 1.0,
|
| 640 |
+
"eval_rewards/chosen": -2.4218428134918213,
|
| 641 |
+
"eval_rewards/margins": 67.89216613769531,
|
| 642 |
+
"eval_rewards/rejected": -70.31401062011719,
|
| 643 |
+
"eval_runtime": 191.9742,
|
| 644 |
+
"eval_samples_per_second": 20.341,
|
| 645 |
+
"eval_steps_per_second": 0.323,
|
| 646 |
+
"step": 350
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 1.0,
|
| 650 |
+
"step": 352,
|
| 651 |
+
"total_flos": 0.0,
|
| 652 |
+
"train_loss": 0.042729352997742935,
|
| 653 |
+
"train_runtime": 10052.8601,
|
| 654 |
+
"train_samples_per_second": 4.476,
|
| 655 |
+
"train_steps_per_second": 0.035
|
| 656 |
+
}
|
| 657 |
+
],
|
| 658 |
+
"logging_steps": 10,
|
| 659 |
+
"max_steps": 352,
|
| 660 |
+
"num_input_tokens_seen": 0,
|
| 661 |
+
"num_train_epochs": 1,
|
| 662 |
+
"save_steps": 100,
|
| 663 |
+
"stateful_callbacks": {
|
| 664 |
+
"TrainerControl": {
|
| 665 |
+
"args": {
|
| 666 |
+
"should_epoch_stop": false,
|
| 667 |
+
"should_evaluate": false,
|
| 668 |
+
"should_log": false,
|
| 669 |
+
"should_save": true,
|
| 670 |
+
"should_training_stop": true
|
| 671 |
+
},
|
| 672 |
+
"attributes": {}
|
| 673 |
+
}
|
| 674 |
+
},
|
| 675 |
+
"total_flos": 0.0,
|
| 676 |
+
"train_batch_size": 8,
|
| 677 |
+
"trial_name": null,
|
| 678 |
+
"trial_params": null
|
| 679 |
+
}
|