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 +1717 -0
README.md
ADDED
|
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: mit
|
| 3 |
+
base_model: EleutherAI/gpt-neo-125M
|
| 4 |
+
tags:
|
| 5 |
+
- trl
|
| 6 |
+
- dpo
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
model-index:
|
| 9 |
+
- name: model
|
| 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 |
+
# model
|
| 17 |
+
|
| 18 |
+
This model is a fine-tuned version of [EleutherAI/gpt-neo-125M](https://huggingface.co/EleutherAI/gpt-neo-125M) on the None dataset.
|
| 19 |
+
It achieves the following results on the evaluation set:
|
| 20 |
+
- Loss: 0.6955
|
| 21 |
+
- Rewards/chosen: -0.0079
|
| 22 |
+
- Rewards/rejected: -0.0080
|
| 23 |
+
- Rewards/accuracies: 0.4813
|
| 24 |
+
- Rewards/margins: 0.0001
|
| 25 |
+
- Logps/rejected: -478.8612
|
| 26 |
+
- Logps/chosen: -494.2958
|
| 27 |
+
- Logits/rejected: -18.3633
|
| 28 |
+
- Logits/chosen: -18.4819
|
| 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: 42
|
| 51 |
+
- gradient_accumulation_steps: 8
|
| 52 |
+
- total_train_batch_size: 64
|
| 53 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 54 |
+
- lr_scheduler_type: cosine
|
| 55 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 56 |
+
- num_epochs: 3
|
| 57 |
+
|
| 58 |
+
### Training results
|
| 59 |
+
|
| 60 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 61 |
+
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 62 |
+
| 0.6955 | 0.2992 | 100 | 0.6958 | -0.0017 | -0.0008 | 0.4701 | -0.0008 | -478.7900 | -494.2336 | -18.3637 | -18.4824 |
|
| 63 |
+
| 0.6906 | 0.5984 | 200 | 0.6962 | -0.0028 | -0.0016 | 0.4744 | -0.0013 | -478.7974 | -494.2453 | -18.3625 | -18.4806 |
|
| 64 |
+
| 0.6985 | 0.8975 | 300 | 0.6959 | -0.0222 | -0.0214 | 0.4738 | -0.0008 | -478.9952 | -494.4388 | -18.3624 | -18.4809 |
|
| 65 |
+
| 0.6946 | 1.1967 | 400 | 0.6955 | 0.0015 | 0.0015 | 0.4753 | 0.0000 | -478.7664 | -494.2018 | -18.3628 | -18.4811 |
|
| 66 |
+
| 0.6946 | 1.4959 | 500 | 0.6960 | -0.0046 | -0.0040 | 0.4791 | -0.0006 | -478.8223 | -494.2634 | -18.3631 | -18.4816 |
|
| 67 |
+
| 0.6952 | 1.7951 | 600 | 0.6951 | -0.0047 | -0.0057 | 0.4882 | 0.0011 | -478.8391 | -494.2639 | -18.3636 | -18.4821 |
|
| 68 |
+
| 0.6947 | 2.0942 | 700 | 0.6955 | -0.0053 | -0.0056 | 0.4822 | 0.0003 | -478.8379 | -494.2701 | -18.3634 | -18.4820 |
|
| 69 |
+
| 0.6995 | 2.3934 | 800 | 0.6948 | -0.0060 | -0.0076 | 0.4918 | 0.0015 | -478.8574 | -494.2774 | -18.3632 | -18.4818 |
|
| 70 |
+
| 0.6932 | 2.6926 | 900 | 0.6952 | -0.0080 | -0.0087 | 0.4837 | 0.0008 | -478.8692 | -494.2970 | -18.3633 | -18.4817 |
|
| 71 |
+
| 0.6964 | 2.9918 | 1000 | 0.6955 | -0.0079 | -0.0080 | 0.4813 | 0.0001 | -478.8612 | -494.2958 | -18.3633 | -18.4819 |
|
| 72 |
+
|
| 73 |
+
|
| 74 |
+
### Framework versions
|
| 75 |
+
|
| 76 |
+
- Transformers 4.41.2
|
| 77 |
+
- Pytorch 2.3.0+cu121
|
| 78 |
+
- Datasets 2.19.1
|
| 79 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.99775617053104,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.6953648589328377,
|
| 5 |
+
"train_runtime": 32693.7742,
|
| 6 |
+
"train_samples": 21390,
|
| 7 |
+
"train_samples_per_second": 1.963,
|
| 8 |
+
"train_steps_per_second": 0.031
|
| 9 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 50256,
|
| 4 |
+
"eos_token_id": 50256,
|
| 5 |
+
"transformers_version": "4.41.2"
|
| 6 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 2.99775617053104,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.6953648589328377,
|
| 5 |
+
"train_runtime": 32693.7742,
|
| 6 |
+
"train_samples": 21390,
|
| 7 |
+
"train_samples_per_second": 1.963,
|
| 8 |
+
"train_steps_per_second": 0.031
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1717 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 2.99775617053104,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 1002,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0029917726252804786,
|
| 13 |
+
"grad_norm": 12.0,
|
| 14 |
+
"learning_rate": 4.95049504950495e-09,
|
| 15 |
+
"logits/chosen": -17.95128631591797,
|
| 16 |
+
"logits/rejected": -17.9676513671875,
|
| 17 |
+
"logps/chosen": -470.5423889160156,
|
| 18 |
+
"logps/rejected": -476.4247131347656,
|
| 19 |
+
"loss": 0.6931,
|
| 20 |
+
"rewards/accuracies": 0.0,
|
| 21 |
+
"rewards/chosen": 0.0,
|
| 22 |
+
"rewards/margins": 0.0,
|
| 23 |
+
"rewards/rejected": 0.0,
|
| 24 |
+
"step": 1
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.029917726252804786,
|
| 28 |
+
"grad_norm": 12.75,
|
| 29 |
+
"learning_rate": 4.950495049504951e-08,
|
| 30 |
+
"logits/chosen": -18.26405906677246,
|
| 31 |
+
"logits/rejected": -18.187000274658203,
|
| 32 |
+
"logps/chosen": -499.1011962890625,
|
| 33 |
+
"logps/rejected": -476.3055419921875,
|
| 34 |
+
"loss": 0.6925,
|
| 35 |
+
"rewards/accuracies": 0.1753472238779068,
|
| 36 |
+
"rewards/chosen": -0.0009479601285420358,
|
| 37 |
+
"rewards/margins": 0.0018414496444165707,
|
| 38 |
+
"rewards/rejected": -0.002789410063996911,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.05983545250560957,
|
| 43 |
+
"grad_norm": 14.5,
|
| 44 |
+
"learning_rate": 9.900990099009901e-08,
|
| 45 |
+
"logits/chosen": -18.466373443603516,
|
| 46 |
+
"logits/rejected": -18.416664123535156,
|
| 47 |
+
"logps/chosen": -518.89794921875,
|
| 48 |
+
"logps/rejected": -501.88494873046875,
|
| 49 |
+
"loss": 0.6937,
|
| 50 |
+
"rewards/accuracies": 0.48124998807907104,
|
| 51 |
+
"rewards/chosen": 0.004138890188187361,
|
| 52 |
+
"rewards/margins": 0.0013234459329396486,
|
| 53 |
+
"rewards/rejected": 0.0028154447209089994,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.08975317875841436,
|
| 58 |
+
"grad_norm": 12.125,
|
| 59 |
+
"learning_rate": 1.485148514851485e-07,
|
| 60 |
+
"logits/chosen": -18.632816314697266,
|
| 61 |
+
"logits/rejected": -18.364212036132812,
|
| 62 |
+
"logps/chosen": -518.1107177734375,
|
| 63 |
+
"logps/rejected": -474.103271484375,
|
| 64 |
+
"loss": 0.6978,
|
| 65 |
+
"rewards/accuracies": 0.44218748807907104,
|
| 66 |
+
"rewards/chosen": -0.0032092046458274126,
|
| 67 |
+
"rewards/margins": -0.005843925289809704,
|
| 68 |
+
"rewards/rejected": 0.0026347211096435785,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.11967090501121914,
|
| 73 |
+
"grad_norm": 12.1875,
|
| 74 |
+
"learning_rate": 1.9801980198019803e-07,
|
| 75 |
+
"logits/chosen": -18.572345733642578,
|
| 76 |
+
"logits/rejected": -18.411455154418945,
|
| 77 |
+
"logps/chosen": -499.1954650878906,
|
| 78 |
+
"logps/rejected": -475.9703674316406,
|
| 79 |
+
"loss": 0.6935,
|
| 80 |
+
"rewards/accuracies": 0.484375,
|
| 81 |
+
"rewards/chosen": -0.007920954376459122,
|
| 82 |
+
"rewards/margins": 0.003416565014049411,
|
| 83 |
+
"rewards/rejected": -0.011337519623339176,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.14958863126402394,
|
| 88 |
+
"grad_norm": 14.0625,
|
| 89 |
+
"learning_rate": 2.475247524752475e-07,
|
| 90 |
+
"logits/chosen": -18.346946716308594,
|
| 91 |
+
"logits/rejected": -18.163400650024414,
|
| 92 |
+
"logps/chosen": -488.15997314453125,
|
| 93 |
+
"logps/rejected": -466.6705017089844,
|
| 94 |
+
"loss": 0.6953,
|
| 95 |
+
"rewards/accuracies": 0.4781250059604645,
|
| 96 |
+
"rewards/chosen": -0.0012789719039574265,
|
| 97 |
+
"rewards/margins": -0.00010849386308109388,
|
| 98 |
+
"rewards/rejected": -0.0011704775970429182,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.17950635751682872,
|
| 103 |
+
"grad_norm": 14.375,
|
| 104 |
+
"learning_rate": 2.97029702970297e-07,
|
| 105 |
+
"logits/chosen": -18.437816619873047,
|
| 106 |
+
"logits/rejected": -18.432828903198242,
|
| 107 |
+
"logps/chosen": -494.74176025390625,
|
| 108 |
+
"logps/rejected": -490.072021484375,
|
| 109 |
+
"loss": 0.6937,
|
| 110 |
+
"rewards/accuracies": 0.48124998807907104,
|
| 111 |
+
"rewards/chosen": 0.005937687586992979,
|
| 112 |
+
"rewards/margins": 0.003491427283734083,
|
| 113 |
+
"rewards/rejected": 0.0024462605360895395,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.2094240837696335,
|
| 118 |
+
"grad_norm": 12.9375,
|
| 119 |
+
"learning_rate": 3.465346534653465e-07,
|
| 120 |
+
"logits/chosen": -18.613513946533203,
|
| 121 |
+
"logits/rejected": -18.44839859008789,
|
| 122 |
+
"logps/chosen": -496.1435546875,
|
| 123 |
+
"logps/rejected": -476.2933044433594,
|
| 124 |
+
"loss": 0.6983,
|
| 125 |
+
"rewards/accuracies": 0.45781248807907104,
|
| 126 |
+
"rewards/chosen": -0.006354107521474361,
|
| 127 |
+
"rewards/margins": -0.00639874953776598,
|
| 128 |
+
"rewards/rejected": 4.464127050596289e-05,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.2393418100224383,
|
| 133 |
+
"grad_norm": 13.0,
|
| 134 |
+
"learning_rate": 3.9603960396039606e-07,
|
| 135 |
+
"logits/chosen": -18.565101623535156,
|
| 136 |
+
"logits/rejected": -18.38371467590332,
|
| 137 |
+
"logps/chosen": -515.5604858398438,
|
| 138 |
+
"logps/rejected": -464.8475646972656,
|
| 139 |
+
"loss": 0.6953,
|
| 140 |
+
"rewards/accuracies": 0.46875,
|
| 141 |
+
"rewards/chosen": -0.010281904600560665,
|
| 142 |
+
"rewards/margins": 0.0007311844383366406,
|
| 143 |
+
"rewards/rejected": -0.01101309061050415,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.26925953627524307,
|
| 148 |
+
"grad_norm": 12.75,
|
| 149 |
+
"learning_rate": 4.4554455445544555e-07,
|
| 150 |
+
"logits/chosen": -18.506877899169922,
|
| 151 |
+
"logits/rejected": -18.355941772460938,
|
| 152 |
+
"logps/chosen": -496.4579162597656,
|
| 153 |
+
"logps/rejected": -481.5419921875,
|
| 154 |
+
"loss": 0.696,
|
| 155 |
+
"rewards/accuracies": 0.4765625,
|
| 156 |
+
"rewards/chosen": -0.008175070397555828,
|
| 157 |
+
"rewards/margins": -0.0013799158623442054,
|
| 158 |
+
"rewards/rejected": -0.006795152090489864,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.2991772625280479,
|
| 163 |
+
"grad_norm": 13.25,
|
| 164 |
+
"learning_rate": 4.95049504950495e-07,
|
| 165 |
+
"logits/chosen": -18.576337814331055,
|
| 166 |
+
"logits/rejected": -18.462890625,
|
| 167 |
+
"logps/chosen": -507.07696533203125,
|
| 168 |
+
"logps/rejected": -471.42303466796875,
|
| 169 |
+
"loss": 0.6955,
|
| 170 |
+
"rewards/accuracies": 0.4859375059604645,
|
| 171 |
+
"rewards/chosen": 0.007285754196345806,
|
| 172 |
+
"rewards/margins": 0.0004216151428408921,
|
| 173 |
+
"rewards/rejected": 0.0068641407415270805,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.2991772625280479,
|
| 178 |
+
"eval_logits/chosen": -18.482393264770508,
|
| 179 |
+
"eval_logits/rejected": -18.36368751525879,
|
| 180 |
+
"eval_logps/chosen": -494.2335510253906,
|
| 181 |
+
"eval_logps/rejected": -478.78997802734375,
|
| 182 |
+
"eval_loss": 0.6958388686180115,
|
| 183 |
+
"eval_rewards/accuracies": 0.4701046347618103,
|
| 184 |
+
"eval_rewards/chosen": -0.001653533661738038,
|
| 185 |
+
"eval_rewards/margins": -0.0008236159919761121,
|
| 186 |
+
"eval_rewards/rejected": -0.0008299172623082995,
|
| 187 |
+
"eval_runtime": 844.1668,
|
| 188 |
+
"eval_samples_per_second": 6.335,
|
| 189 |
+
"eval_steps_per_second": 0.792,
|
| 190 |
+
"step": 100
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.32909498878085264,
|
| 194 |
+
"grad_norm": 14.25,
|
| 195 |
+
"learning_rate": 4.998769137458909e-07,
|
| 196 |
+
"logits/chosen": -18.438594818115234,
|
| 197 |
+
"logits/rejected": -18.30778694152832,
|
| 198 |
+
"logps/chosen": -488.8753967285156,
|
| 199 |
+
"logps/rejected": -469.21240234375,
|
| 200 |
+
"loss": 0.693,
|
| 201 |
+
"rewards/accuracies": 0.49687498807907104,
|
| 202 |
+
"rewards/chosen": -0.0037526420783251524,
|
| 203 |
+
"rewards/margins": 0.005193626508116722,
|
| 204 |
+
"rewards/rejected": -0.008946266956627369,
|
| 205 |
+
"step": 110
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.35901271503365745,
|
| 209 |
+
"grad_norm": 13.375,
|
| 210 |
+
"learning_rate": 4.994515860008407e-07,
|
| 211 |
+
"logits/chosen": -18.361738204956055,
|
| 212 |
+
"logits/rejected": -18.35593032836914,
|
| 213 |
+
"logps/chosen": -502.665771484375,
|
| 214 |
+
"logps/rejected": -485.8645935058594,
|
| 215 |
+
"loss": 0.6939,
|
| 216 |
+
"rewards/accuracies": 0.4921875,
|
| 217 |
+
"rewards/chosen": -0.011272026225924492,
|
| 218 |
+
"rewards/margins": 0.0029675140976905823,
|
| 219 |
+
"rewards/rejected": -0.0142395393922925,
|
| 220 |
+
"step": 120
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.3889304412864622,
|
| 224 |
+
"grad_norm": 14.4375,
|
| 225 |
+
"learning_rate": 4.987230141051212e-07,
|
| 226 |
+
"logits/chosen": -18.523672103881836,
|
| 227 |
+
"logits/rejected": -18.332469940185547,
|
| 228 |
+
"logps/chosen": -497.9520568847656,
|
| 229 |
+
"logps/rejected": -494.5643615722656,
|
| 230 |
+
"loss": 0.6936,
|
| 231 |
+
"rewards/accuracies": 0.48124998807907104,
|
| 232 |
+
"rewards/chosen": 0.0009459417196922004,
|
| 233 |
+
"rewards/margins": 0.003786131041124463,
|
| 234 |
+
"rewards/rejected": -0.0028401894960552454,
|
| 235 |
+
"step": 130
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.418848167539267,
|
| 239 |
+
"grad_norm": 13.6875,
|
| 240 |
+
"learning_rate": 4.976920837422805e-07,
|
| 241 |
+
"logits/chosen": -18.39594078063965,
|
| 242 |
+
"logits/rejected": -18.280231475830078,
|
| 243 |
+
"logps/chosen": -499.1576232910156,
|
| 244 |
+
"logps/rejected": -480.56964111328125,
|
| 245 |
+
"loss": 0.692,
|
| 246 |
+
"rewards/accuracies": 0.504687488079071,
|
| 247 |
+
"rewards/chosen": -0.008311725221574306,
|
| 248 |
+
"rewards/margins": 0.006742014549672604,
|
| 249 |
+
"rewards/rejected": -0.01505373977124691,
|
| 250 |
+
"step": 140
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.4487658937920718,
|
| 254 |
+
"grad_norm": 13.5,
|
| 255 |
+
"learning_rate": 4.963600481559129e-07,
|
| 256 |
+
"logits/chosen": -18.300243377685547,
|
| 257 |
+
"logits/rejected": -18.194604873657227,
|
| 258 |
+
"logps/chosen": -495.46173095703125,
|
| 259 |
+
"logps/rejected": -483.1366271972656,
|
| 260 |
+
"loss": 0.6985,
|
| 261 |
+
"rewards/accuracies": 0.4671874940395355,
|
| 262 |
+
"rewards/chosen": -0.014861522242426872,
|
| 263 |
+
"rewards/margins": -0.005774814635515213,
|
| 264 |
+
"rewards/rejected": -0.009086708538234234,
|
| 265 |
+
"step": 150
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.4786836200448766,
|
| 269 |
+
"grad_norm": 14.9375,
|
| 270 |
+
"learning_rate": 4.947285266261623e-07,
|
| 271 |
+
"logits/chosen": -18.28691864013672,
|
| 272 |
+
"logits/rejected": -18.09464454650879,
|
| 273 |
+
"logps/chosen": -501.8270568847656,
|
| 274 |
+
"logps/rejected": -481.210693359375,
|
| 275 |
+
"loss": 0.6892,
|
| 276 |
+
"rewards/accuracies": 0.542187511920929,
|
| 277 |
+
"rewards/chosen": 0.004282023757696152,
|
| 278 |
+
"rewards/margins": 0.012691694311797619,
|
| 279 |
+
"rewards/rejected": -0.008409671485424042,
|
| 280 |
+
"step": 160
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.5086013462976814,
|
| 284 |
+
"grad_norm": 12.375,
|
| 285 |
+
"learning_rate": 4.927995025012548e-07,
|
| 286 |
+
"logits/chosen": -18.599102020263672,
|
| 287 |
+
"logits/rejected": -18.47110366821289,
|
| 288 |
+
"logps/chosen": -497.13409423828125,
|
| 289 |
+
"logps/rejected": -462.9967346191406,
|
| 290 |
+
"loss": 0.6976,
|
| 291 |
+
"rewards/accuracies": 0.47187501192092896,
|
| 292 |
+
"rewards/chosen": -0.015234975144267082,
|
| 293 |
+
"rewards/margins": -0.004649547394365072,
|
| 294 |
+
"rewards/rejected": -0.010585429146885872,
|
| 295 |
+
"step": 170
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 0.5385190725504861,
|
| 299 |
+
"grad_norm": 18.0,
|
| 300 |
+
"learning_rate": 4.905753207864547e-07,
|
| 301 |
+
"logits/chosen": -18.513076782226562,
|
| 302 |
+
"logits/rejected": -18.358755111694336,
|
| 303 |
+
"logps/chosen": -499.5277404785156,
|
| 304 |
+
"logps/rejected": -484.95782470703125,
|
| 305 |
+
"loss": 0.6963,
|
| 306 |
+
"rewards/accuracies": 0.47968751192092896,
|
| 307 |
+
"rewards/chosen": -0.002763888565823436,
|
| 308 |
+
"rewards/margins": -0.0016130201984196901,
|
| 309 |
+
"rewards/rejected": -0.0011508697643876076,
|
| 310 |
+
"step": 180
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.5684367988032909,
|
| 314 |
+
"grad_norm": 14.0625,
|
| 315 |
+
"learning_rate": 4.880586852933747e-07,
|
| 316 |
+
"logits/chosen": -18.47112464904785,
|
| 317 |
+
"logits/rejected": -18.298198699951172,
|
| 318 |
+
"logps/chosen": -504.328857421875,
|
| 319 |
+
"logps/rejected": -473.43359375,
|
| 320 |
+
"loss": 0.6939,
|
| 321 |
+
"rewards/accuracies": 0.512499988079071,
|
| 322 |
+
"rewards/chosen": -0.0022486697416752577,
|
| 323 |
+
"rewards/margins": 0.00311539089307189,
|
| 324 |
+
"rewards/rejected": -0.005364061798900366,
|
| 325 |
+
"step": 190
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.5983545250560958,
|
| 329 |
+
"grad_norm": 13.9375,
|
| 330 |
+
"learning_rate": 4.852526553531054e-07,
|
| 331 |
+
"logits/chosen": -18.492422103881836,
|
| 332 |
+
"logits/rejected": -18.396549224853516,
|
| 333 |
+
"logps/chosen": -498.78741455078125,
|
| 334 |
+
"logps/rejected": -481.09405517578125,
|
| 335 |
+
"loss": 0.6906,
|
| 336 |
+
"rewards/accuracies": 0.4781250059604645,
|
| 337 |
+
"rewards/chosen": -0.0009043412283062935,
|
| 338 |
+
"rewards/margins": 0.010298773646354675,
|
| 339 |
+
"rewards/rejected": -0.01120311301201582,
|
| 340 |
+
"step": 200
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.5983545250560958,
|
| 344 |
+
"eval_logits/chosen": -18.480627059936523,
|
| 345 |
+
"eval_logits/rejected": -18.362462997436523,
|
| 346 |
+
"eval_logps/chosen": -494.2452697753906,
|
| 347 |
+
"eval_logps/rejected": -478.79736328125,
|
| 348 |
+
"eval_loss": 0.6961907148361206,
|
| 349 |
+
"eval_rewards/accuracies": 0.4744020998477936,
|
| 350 |
+
"eval_rewards/chosen": -0.0028220494277775288,
|
| 351 |
+
"eval_rewards/margins": -0.001257332507520914,
|
| 352 |
+
"eval_rewards/rejected": -0.0015647169202566147,
|
| 353 |
+
"eval_runtime": 842.3412,
|
| 354 |
+
"eval_samples_per_second": 6.349,
|
| 355 |
+
"eval_steps_per_second": 0.794,
|
| 356 |
+
"step": 200
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.6282722513089005,
|
| 360 |
+
"grad_norm": 13.6875,
|
| 361 |
+
"learning_rate": 4.8216064209716e-07,
|
| 362 |
+
"logits/chosen": -18.475950241088867,
|
| 363 |
+
"logits/rejected": -18.323816299438477,
|
| 364 |
+
"logps/chosen": -510.2359313964844,
|
| 365 |
+
"logps/rejected": -491.6826171875,
|
| 366 |
+
"loss": 0.6954,
|
| 367 |
+
"rewards/accuracies": 0.4703125059604645,
|
| 368 |
+
"rewards/chosen": -0.00869015883654356,
|
| 369 |
+
"rewards/margins": 0.0011458725202828646,
|
| 370 |
+
"rewards/rejected": -0.009836031123995781,
|
| 371 |
+
"step": 210
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 0.6581899775617053,
|
| 375 |
+
"grad_norm": 13.75,
|
| 376 |
+
"learning_rate": 4.787864043107546e-07,
|
| 377 |
+
"logits/chosen": -18.483421325683594,
|
| 378 |
+
"logits/rejected": -18.315481185913086,
|
| 379 |
+
"logps/chosen": -505.24383544921875,
|
| 380 |
+
"logps/rejected": -480.8597717285156,
|
| 381 |
+
"loss": 0.6945,
|
| 382 |
+
"rewards/accuracies": 0.4984374940395355,
|
| 383 |
+
"rewards/chosen": -0.007081210613250732,
|
| 384 |
+
"rewards/margins": 0.002295692218467593,
|
| 385 |
+
"rewards/rejected": -0.009376903995871544,
|
| 386 |
+
"step": 220
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.6881077038145101,
|
| 390 |
+
"grad_norm": 12.3125,
|
| 391 |
+
"learning_rate": 4.751340438634669e-07,
|
| 392 |
+
"logits/chosen": -18.307737350463867,
|
| 393 |
+
"logits/rejected": -18.23043441772461,
|
| 394 |
+
"logps/chosen": -492.05596923828125,
|
| 395 |
+
"logps/rejected": -478.38238525390625,
|
| 396 |
+
"loss": 0.6932,
|
| 397 |
+
"rewards/accuracies": 0.4828124940395355,
|
| 398 |
+
"rewards/chosen": -0.005866049323230982,
|
| 399 |
+
"rewards/margins": 0.005141937639564276,
|
| 400 |
+
"rewards/rejected": -0.011007988825440407,
|
| 401 |
+
"step": 230
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.7180254300673149,
|
| 405 |
+
"grad_norm": 13.8125,
|
| 406 |
+
"learning_rate": 4.712080007228254e-07,
|
| 407 |
+
"logits/chosen": -18.248706817626953,
|
| 408 |
+
"logits/rejected": -18.240238189697266,
|
| 409 |
+
"logps/chosen": -498.770263671875,
|
| 410 |
+
"logps/rejected": -494.4297790527344,
|
| 411 |
+
"loss": 0.6899,
|
| 412 |
+
"rewards/accuracies": 0.5015624761581421,
|
| 413 |
+
"rewards/chosen": -0.0009179557673633099,
|
| 414 |
+
"rewards/margins": 0.01158793456852436,
|
| 415 |
+
"rewards/rejected": -0.012505888938903809,
|
| 416 |
+
"step": 240
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.7479431563201197,
|
| 420 |
+
"grad_norm": 15.5625,
|
| 421 |
+
"learning_rate": 4.670130475568927e-07,
|
| 422 |
+
"logits/chosen": -18.496627807617188,
|
| 423 |
+
"logits/rejected": -18.331436157226562,
|
| 424 |
+
"logps/chosen": -501.021728515625,
|
| 425 |
+
"logps/rejected": -485.9256896972656,
|
| 426 |
+
"loss": 0.693,
|
| 427 |
+
"rewards/accuracies": 0.4859375059604645,
|
| 428 |
+
"rewards/chosen": 0.0024902846198529005,
|
| 429 |
+
"rewards/margins": 0.005465887952595949,
|
| 430 |
+
"rewards/rejected": -0.002975603099912405,
|
| 431 |
+
"step": 250
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.7778608825729244,
|
| 435 |
+
"grad_norm": 12.625,
|
| 436 |
+
"learning_rate": 4.6255428393240354e-07,
|
| 437 |
+
"logits/chosen": -18.292850494384766,
|
| 438 |
+
"logits/rejected": -18.20071029663086,
|
| 439 |
+
"logps/chosen": -479.3639221191406,
|
| 440 |
+
"logps/rejected": -463.76776123046875,
|
| 441 |
+
"loss": 0.6947,
|
| 442 |
+
"rewards/accuracies": 0.484375,
|
| 443 |
+
"rewards/chosen": -0.019604574888944626,
|
| 444 |
+
"rewards/margins": 0.0011741012567654252,
|
| 445 |
+
"rewards/rejected": -0.020778674632310867,
|
| 446 |
+
"step": 260
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.8077786088257293,
|
| 450 |
+
"grad_norm": 13.4375,
|
| 451 |
+
"learning_rate": 4.578371301155106e-07,
|
| 452 |
+
"logits/chosen": -18.332660675048828,
|
| 453 |
+
"logits/rejected": -18.25689697265625,
|
| 454 |
+
"logps/chosen": -497.0006408691406,
|
| 455 |
+
"logps/rejected": -493.8580017089844,
|
| 456 |
+
"loss": 0.6925,
|
| 457 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 458 |
+
"rewards/chosen": -0.017484821379184723,
|
| 459 |
+
"rewards/margins": 0.006399748381227255,
|
| 460 |
+
"rewards/rejected": -0.023884572088718414,
|
| 461 |
+
"step": 270
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.837696335078534,
|
| 465 |
+
"grad_norm": 13.625,
|
| 466 |
+
"learning_rate": 4.528673204826744e-07,
|
| 467 |
+
"logits/chosen": -18.48120880126953,
|
| 468 |
+
"logits/rejected": -18.390308380126953,
|
| 469 |
+
"logps/chosen": -513.1058959960938,
|
| 470 |
+
"logps/rejected": -496.99578857421875,
|
| 471 |
+
"loss": 0.6956,
|
| 472 |
+
"rewards/accuracies": 0.4515624940395355,
|
| 473 |
+
"rewards/chosen": -0.023203711956739426,
|
| 474 |
+
"rewards/margins": 0.0008115891250781715,
|
| 475 |
+
"rewards/rejected": -0.02401530183851719,
|
| 476 |
+
"step": 280
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.8676140613313388,
|
| 480 |
+
"grad_norm": 15.3125,
|
| 481 |
+
"learning_rate": 4.476508965497067e-07,
|
| 482 |
+
"logits/chosen": -18.398761749267578,
|
| 483 |
+
"logits/rejected": -18.283004760742188,
|
| 484 |
+
"logps/chosen": -498.1253967285156,
|
| 485 |
+
"logps/rejected": -489.4436950683594,
|
| 486 |
+
"loss": 0.697,
|
| 487 |
+
"rewards/accuracies": 0.4749999940395355,
|
| 488 |
+
"rewards/chosen": -0.015816329047083855,
|
| 489 |
+
"rewards/margins": -0.002141923177987337,
|
| 490 |
+
"rewards/rejected": -0.013674402609467506,
|
| 491 |
+
"step": 290
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.8975317875841436,
|
| 495 |
+
"grad_norm": 15.0,
|
| 496 |
+
"learning_rate": 4.421941996274423e-07,
|
| 497 |
+
"logits/chosen": -18.34426498413086,
|
| 498 |
+
"logits/rejected": -18.241085052490234,
|
| 499 |
+
"logps/chosen": -527.1797485351562,
|
| 500 |
+
"logps/rejected": -507.3211975097656,
|
| 501 |
+
"loss": 0.6985,
|
| 502 |
+
"rewards/accuracies": 0.4593749940395355,
|
| 503 |
+
"rewards/chosen": -0.020052501931786537,
|
| 504 |
+
"rewards/margins": -0.005515648517757654,
|
| 505 |
+
"rewards/rejected": -0.014536852948367596,
|
| 506 |
+
"step": 300
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.8975317875841436,
|
| 510 |
+
"eval_logits/chosen": -18.48088836669922,
|
| 511 |
+
"eval_logits/rejected": -18.36236000061035,
|
| 512 |
+
"eval_logps/chosen": -494.4387512207031,
|
| 513 |
+
"eval_logps/rejected": -478.9952392578125,
|
| 514 |
+
"eval_loss": 0.6959420442581177,
|
| 515 |
+
"eval_rewards/accuracies": 0.4738415479660034,
|
| 516 |
+
"eval_rewards/chosen": -0.022165408357977867,
|
| 517 |
+
"eval_rewards/margins": -0.000810055120382458,
|
| 518 |
+
"eval_rewards/rejected": -0.02135535702109337,
|
| 519 |
+
"eval_runtime": 842.4585,
|
| 520 |
+
"eval_samples_per_second": 6.348,
|
| 521 |
+
"eval_steps_per_second": 0.794,
|
| 522 |
+
"step": 300
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.9274495138369484,
|
| 526 |
+
"grad_norm": 14.0625,
|
| 527 |
+
"learning_rate": 4.3650386311296715e-07,
|
| 528 |
+
"logits/chosen": -18.639049530029297,
|
| 529 |
+
"logits/rejected": -18.473182678222656,
|
| 530 |
+
"logps/chosen": -519.571533203125,
|
| 531 |
+
"logps/rejected": -493.08416748046875,
|
| 532 |
+
"loss": 0.7007,
|
| 533 |
+
"rewards/accuracies": 0.44218748807907104,
|
| 534 |
+
"rewards/chosen": -0.020600352436304092,
|
| 535 |
+
"rewards/margins": -0.009581638500094414,
|
| 536 |
+
"rewards/rejected": -0.011018714867532253,
|
| 537 |
+
"step": 310
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.9573672400897532,
|
| 541 |
+
"grad_norm": 14.625,
|
| 542 |
+
"learning_rate": 4.305868044257734e-07,
|
| 543 |
+
"logits/chosen": -18.35852813720703,
|
| 544 |
+
"logits/rejected": -18.237619400024414,
|
| 545 |
+
"logps/chosen": -494.0035705566406,
|
| 546 |
+
"logps/rejected": -474.61175537109375,
|
| 547 |
+
"loss": 0.6962,
|
| 548 |
+
"rewards/accuracies": 0.4921875,
|
| 549 |
+
"rewards/chosen": -0.03291434049606323,
|
| 550 |
+
"rewards/margins": -0.0014624244067817926,
|
| 551 |
+
"rewards/rejected": -0.031451914459466934,
|
| 552 |
+
"step": 320
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 0.9872849663425579,
|
| 556 |
+
"grad_norm": 15.0,
|
| 557 |
+
"learning_rate": 4.244502165986448e-07,
|
| 558 |
+
"logits/chosen": -18.50986099243164,
|
| 559 |
+
"logits/rejected": -18.264389038085938,
|
| 560 |
+
"logps/chosen": -487.98284912109375,
|
| 561 |
+
"logps/rejected": -478.84002685546875,
|
| 562 |
+
"loss": 0.6944,
|
| 563 |
+
"rewards/accuracies": 0.4859375059604645,
|
| 564 |
+
"rewards/chosen": -0.018264159560203552,
|
| 565 |
+
"rewards/margins": 0.0021764684934169054,
|
| 566 |
+
"rewards/rejected": -0.02044062688946724,
|
| 567 |
+
"step": 330
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.0172026925953628,
|
| 571 |
+
"grad_norm": 13.0625,
|
| 572 |
+
"learning_rate": 4.1810155953349445e-07,
|
| 573 |
+
"logits/chosen": -18.608068466186523,
|
| 574 |
+
"logits/rejected": -18.53306007385254,
|
| 575 |
+
"logps/chosen": -499.0755310058594,
|
| 576 |
+
"logps/rejected": -490.98486328125,
|
| 577 |
+
"loss": 0.6936,
|
| 578 |
+
"rewards/accuracies": 0.47343748807907104,
|
| 579 |
+
"rewards/chosen": -0.01962715946137905,
|
| 580 |
+
"rewards/margins": 0.00391949201002717,
|
| 581 |
+
"rewards/rejected": -0.023546651005744934,
|
| 582 |
+
"step": 340
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 1.0471204188481675,
|
| 586 |
+
"grad_norm": 12.1875,
|
| 587 |
+
"learning_rate": 4.115485509327851e-07,
|
| 588 |
+
"logits/chosen": -18.467256546020508,
|
| 589 |
+
"logits/rejected": -18.209827423095703,
|
| 590 |
+
"logps/chosen": -505.059814453125,
|
| 591 |
+
"logps/rejected": -475.1127014160156,
|
| 592 |
+
"loss": 0.6934,
|
| 593 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 594 |
+
"rewards/chosen": -0.009037208743393421,
|
| 595 |
+
"rewards/margins": 0.004168554209172726,
|
| 596 |
+
"rewards/rejected": -0.013205763883888721,
|
| 597 |
+
"step": 350
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 1.0770381451009723,
|
| 601 |
+
"grad_norm": 12.4375,
|
| 602 |
+
"learning_rate": 4.0479915691755583e-07,
|
| 603 |
+
"logits/chosen": -18.381351470947266,
|
| 604 |
+
"logits/rejected": -18.23720359802246,
|
| 605 |
+
"logps/chosen": -503.64044189453125,
|
| 606 |
+
"logps/rejected": -486.0143127441406,
|
| 607 |
+
"loss": 0.7002,
|
| 608 |
+
"rewards/accuracies": 0.4703125059604645,
|
| 609 |
+
"rewards/chosen": -0.015495906583964825,
|
| 610 |
+
"rewards/margins": -0.009308911859989166,
|
| 611 |
+
"rewards/rejected": -0.006186993792653084,
|
| 612 |
+
"step": 360
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 1.106955871353777,
|
| 616 |
+
"grad_norm": 13.8125,
|
| 617 |
+
"learning_rate": 3.9786158234346e-07,
|
| 618 |
+
"logits/chosen": -18.671598434448242,
|
| 619 |
+
"logits/rejected": -18.52102279663086,
|
| 620 |
+
"logps/chosen": -491.046142578125,
|
| 621 |
+
"logps/rejected": -473.78057861328125,
|
| 622 |
+
"loss": 0.692,
|
| 623 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 624 |
+
"rewards/chosen": -0.0006091233226470649,
|
| 625 |
+
"rewards/margins": 0.007144009228795767,
|
| 626 |
+
"rewards/rejected": -0.007753132376819849,
|
| 627 |
+
"step": 370
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 1.136873597606582,
|
| 631 |
+
"grad_norm": 14.375,
|
| 632 |
+
"learning_rate": 3.90744260826588e-07,
|
| 633 |
+
"logits/chosen": -18.341663360595703,
|
| 634 |
+
"logits/rejected": -18.282123565673828,
|
| 635 |
+
"logps/chosen": -503.53521728515625,
|
| 636 |
+
"logps/rejected": -478.75726318359375,
|
| 637 |
+
"loss": 0.6975,
|
| 638 |
+
"rewards/accuracies": 0.4828124940395355,
|
| 639 |
+
"rewards/chosen": -0.014238146133720875,
|
| 640 |
+
"rewards/margins": -0.004205497447401285,
|
| 641 |
+
"rewards/rejected": -0.010032649151980877,
|
| 642 |
+
"step": 380
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 1.1667913238593868,
|
| 646 |
+
"grad_norm": 13.75,
|
| 647 |
+
"learning_rate": 3.834558444911977e-07,
|
| 648 |
+
"logits/chosen": -18.506235122680664,
|
| 649 |
+
"logits/rejected": -18.454498291015625,
|
| 650 |
+
"logps/chosen": -501.6888122558594,
|
| 651 |
+
"logps/rejected": -492.0193786621094,
|
| 652 |
+
"loss": 0.6937,
|
| 653 |
+
"rewards/accuracies": 0.4671874940395355,
|
| 654 |
+
"rewards/chosen": -0.013215325772762299,
|
| 655 |
+
"rewards/margins": 0.004076224751770496,
|
| 656 |
+
"rewards/rejected": -0.01729154959321022,
|
| 657 |
+
"step": 390
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.1967090501121915,
|
| 661 |
+
"grad_norm": 11.4375,
|
| 662 |
+
"learning_rate": 3.760051934518178e-07,
|
| 663 |
+
"logits/chosen": -18.325563430786133,
|
| 664 |
+
"logits/rejected": -18.207347869873047,
|
| 665 |
+
"logps/chosen": -490.45599365234375,
|
| 666 |
+
"logps/rejected": -484.768798828125,
|
| 667 |
+
"loss": 0.6946,
|
| 668 |
+
"rewards/accuracies": 0.503125011920929,
|
| 669 |
+
"rewards/chosen": -0.012493303045630455,
|
| 670 |
+
"rewards/margins": 0.0020187748596072197,
|
| 671 |
+
"rewards/rejected": -0.0145120769739151,
|
| 672 |
+
"step": 400
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 1.1967090501121915,
|
| 676 |
+
"eval_logits/chosen": -18.481069564819336,
|
| 677 |
+
"eval_logits/rejected": -18.362834930419922,
|
| 678 |
+
"eval_logps/chosen": -494.2017822265625,
|
| 679 |
+
"eval_logps/rejected": -478.7664489746094,
|
| 680 |
+
"eval_loss": 0.6955097913742065,
|
| 681 |
+
"eval_rewards/accuracies": 0.47533631324768066,
|
| 682 |
+
"eval_rewards/chosen": 0.0015334173804149032,
|
| 683 |
+
"eval_rewards/margins": 5.264274932414992e-06,
|
| 684 |
+
"eval_rewards/rejected": 0.0015281536616384983,
|
| 685 |
+
"eval_runtime": 842.6072,
|
| 686 |
+
"eval_samples_per_second": 6.347,
|
| 687 |
+
"eval_steps_per_second": 0.794,
|
| 688 |
+
"step": 400
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 1.2266267763649963,
|
| 692 |
+
"grad_norm": 12.5,
|
| 693 |
+
"learning_rate": 3.6840136504250895e-07,
|
| 694 |
+
"logits/chosen": -18.419193267822266,
|
| 695 |
+
"logits/rejected": -18.217782974243164,
|
| 696 |
+
"logps/chosen": -504.786865234375,
|
| 697 |
+
"logps/rejected": -477.596923828125,
|
| 698 |
+
"loss": 0.6965,
|
| 699 |
+
"rewards/accuracies": 0.4984374940395355,
|
| 700 |
+
"rewards/chosen": -0.006161923985928297,
|
| 701 |
+
"rewards/margins": -0.001232474809512496,
|
| 702 |
+
"rewards/rejected": -0.004929449409246445,
|
| 703 |
+
"step": 410
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 1.256544502617801,
|
| 707 |
+
"grad_norm": 13.0625,
|
| 708 |
+
"learning_rate": 3.6065360280637635e-07,
|
| 709 |
+
"logits/chosen": -18.372119903564453,
|
| 710 |
+
"logits/rejected": -18.320249557495117,
|
| 711 |
+
"logps/chosen": -484.0809020996094,
|
| 712 |
+
"logps/rejected": -474.786376953125,
|
| 713 |
+
"loss": 0.6929,
|
| 714 |
+
"rewards/accuracies": 0.47343748807907104,
|
| 715 |
+
"rewards/chosen": -0.009603636339306831,
|
| 716 |
+
"rewards/margins": 0.00548522686585784,
|
| 717 |
+
"rewards/rejected": -0.015088860876858234,
|
| 718 |
+
"step": 420
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 1.2864622288706058,
|
| 722 |
+
"grad_norm": 16.375,
|
| 723 |
+
"learning_rate": 3.527713252587179e-07,
|
| 724 |
+
"logits/chosen": -18.48366928100586,
|
| 725 |
+
"logits/rejected": -18.39108657836914,
|
| 726 |
+
"logps/chosen": -501.91314697265625,
|
| 727 |
+
"logps/rejected": -480.6852111816406,
|
| 728 |
+
"loss": 0.6968,
|
| 729 |
+
"rewards/accuracies": 0.4671874940395355,
|
| 730 |
+
"rewards/chosen": -0.004988933447748423,
|
| 731 |
+
"rewards/margins": -0.0021209451369941235,
|
| 732 |
+
"rewards/rejected": -0.002867989707738161,
|
| 733 |
+
"step": 430
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 1.3163799551234106,
|
| 737 |
+
"grad_norm": 14.0625,
|
| 738 |
+
"learning_rate": 3.447641144374697e-07,
|
| 739 |
+
"logits/chosen": -18.41563606262207,
|
| 740 |
+
"logits/rejected": -18.326993942260742,
|
| 741 |
+
"logps/chosen": -495.029541015625,
|
| 742 |
+
"logps/rejected": -485.95654296875,
|
| 743 |
+
"loss": 0.6938,
|
| 744 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 745 |
+
"rewards/chosen": -0.0021522603929042816,
|
| 746 |
+
"rewards/margins": 0.003724109847098589,
|
| 747 |
+
"rewards/rejected": -0.005876368843019009,
|
| 748 |
+
"step": 440
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 1.3462976813762153,
|
| 752 |
+
"grad_norm": 12.8125,
|
| 753 |
+
"learning_rate": 3.3664170425486577e-07,
|
| 754 |
+
"logits/chosen": -18.418684005737305,
|
| 755 |
+
"logits/rejected": -18.22823143005371,
|
| 756 |
+
"logps/chosen": -508.91619873046875,
|
| 757 |
+
"logps/rejected": -493.2528381347656,
|
| 758 |
+
"loss": 0.6962,
|
| 759 |
+
"rewards/accuracies": 0.4859375059604645,
|
| 760 |
+
"rewards/chosen": -0.016366075724363327,
|
| 761 |
+
"rewards/margins": -0.0013246103189885616,
|
| 762 |
+
"rewards/rejected": -0.015041463077068329,
|
| 763 |
+
"step": 450
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 1.37621540762902,
|
| 767 |
+
"grad_norm": 12.25,
|
| 768 |
+
"learning_rate": 3.2841396866447216e-07,
|
| 769 |
+
"logits/chosen": -18.291414260864258,
|
| 770 |
+
"logits/rejected": -18.216777801513672,
|
| 771 |
+
"logps/chosen": -494.11651611328125,
|
| 772 |
+
"logps/rejected": -479.777587890625,
|
| 773 |
+
"loss": 0.6941,
|
| 774 |
+
"rewards/accuracies": 0.4749999940395355,
|
| 775 |
+
"rewards/chosen": 0.010070102289319038,
|
| 776 |
+
"rewards/margins": 0.002756762085482478,
|
| 777 |
+
"rewards/rejected": 0.007313339505344629,
|
| 778 |
+
"step": 460
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 1.406133133881825,
|
| 782 |
+
"grad_norm": 12.125,
|
| 783 |
+
"learning_rate": 3.200909096579822e-07,
|
| 784 |
+
"logits/chosen": -18.547330856323242,
|
| 785 |
+
"logits/rejected": -18.329965591430664,
|
| 786 |
+
"logps/chosen": -502.29052734375,
|
| 787 |
+
"logps/rejected": -481.80035400390625,
|
| 788 |
+
"loss": 0.6937,
|
| 789 |
+
"rewards/accuracies": 0.484375,
|
| 790 |
+
"rewards/chosen": 0.006962032523006201,
|
| 791 |
+
"rewards/margins": 0.004513027612119913,
|
| 792 |
+
"rewards/rejected": 0.0024490035139024258,
|
| 793 |
+
"step": 470
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 1.4360508601346298,
|
| 797 |
+
"grad_norm": 13.75,
|
| 798 |
+
"learning_rate": 3.116826451063618e-07,
|
| 799 |
+
"logits/chosen": -18.396873474121094,
|
| 800 |
+
"logits/rejected": -18.2117862701416,
|
| 801 |
+
"logps/chosen": -498.18048095703125,
|
| 802 |
+
"logps/rejected": -463.1722106933594,
|
| 803 |
+
"loss": 0.6994,
|
| 804 |
+
"rewards/accuracies": 0.4625000059604645,
|
| 805 |
+
"rewards/chosen": -0.007130985148251057,
|
| 806 |
+
"rewards/margins": -0.007668208330869675,
|
| 807 |
+
"rewards/rejected": 0.0005372242303565145,
|
| 808 |
+
"step": 480
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 1.4659685863874345,
|
| 812 |
+
"grad_norm": 13.4375,
|
| 813 |
+
"learning_rate": 3.031993964601274e-07,
|
| 814 |
+
"logits/chosen": -18.40250015258789,
|
| 815 |
+
"logits/rejected": -18.297494888305664,
|
| 816 |
+
"logps/chosen": -503.1748962402344,
|
| 817 |
+
"logps/rejected": -499.47900390625,
|
| 818 |
+
"loss": 0.6945,
|
| 819 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 820 |
+
"rewards/chosen": -0.006336678750813007,
|
| 821 |
+
"rewards/margins": 0.0019012052798643708,
|
| 822 |
+
"rewards/rejected": -0.008237885311245918,
|
| 823 |
+
"step": 490
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 1.4958863126402393,
|
| 827 |
+
"grad_norm": 12.6875,
|
| 828 |
+
"learning_rate": 2.9465147632370795e-07,
|
| 829 |
+
"logits/chosen": -18.3398380279541,
|
| 830 |
+
"logits/rejected": -18.301984786987305,
|
| 831 |
+
"logps/chosen": -485.8975524902344,
|
| 832 |
+
"logps/rejected": -474.5433654785156,
|
| 833 |
+
"loss": 0.6946,
|
| 834 |
+
"rewards/accuracies": 0.4765625,
|
| 835 |
+
"rewards/chosen": -0.00577941257506609,
|
| 836 |
+
"rewards/margins": 0.0024705647956579924,
|
| 837 |
+
"rewards/rejected": -0.008249977603554726,
|
| 838 |
+
"step": 500
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 1.4958863126402393,
|
| 842 |
+
"eval_logits/chosen": -18.481569290161133,
|
| 843 |
+
"eval_logits/rejected": -18.363122940063477,
|
| 844 |
+
"eval_logps/chosen": -494.26336669921875,
|
| 845 |
+
"eval_logps/rejected": -478.822265625,
|
| 846 |
+
"eval_loss": 0.6959648728370667,
|
| 847 |
+
"eval_rewards/accuracies": 0.47907325625419617,
|
| 848 |
+
"eval_rewards/chosen": -0.0046310219913721085,
|
| 849 |
+
"eval_rewards/margins": -0.0005819797515869141,
|
| 850 |
+
"eval_rewards/rejected": -0.0040490408428013325,
|
| 851 |
+
"eval_runtime": 842.3853,
|
| 852 |
+
"eval_samples_per_second": 6.349,
|
| 853 |
+
"eval_steps_per_second": 0.794,
|
| 854 |
+
"step": 500
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 1.5258040388930443,
|
| 858 |
+
"grad_norm": 15.875,
|
| 859 |
+
"learning_rate": 2.86049275918997e-07,
|
| 860 |
+
"logits/chosen": -18.267154693603516,
|
| 861 |
+
"logits/rejected": -18.09113883972168,
|
| 862 |
+
"logps/chosen": -506.2464904785156,
|
| 863 |
+
"logps/rejected": -472.58721923828125,
|
| 864 |
+
"loss": 0.6937,
|
| 865 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 866 |
+
"rewards/chosen": 0.00022939778864383698,
|
| 867 |
+
"rewards/margins": 0.003868215484544635,
|
| 868 |
+
"rewards/rejected": -0.003638815600425005,
|
| 869 |
+
"step": 510
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 1.555721765145849,
|
| 873 |
+
"grad_norm": 11.9375,
|
| 874 |
+
"learning_rate": 2.774032524533323e-07,
|
| 875 |
+
"logits/chosen": -18.302738189697266,
|
| 876 |
+
"logits/rejected": -18.214998245239258,
|
| 877 |
+
"logps/chosen": -485.1527404785156,
|
| 878 |
+
"logps/rejected": -474.42120361328125,
|
| 879 |
+
"loss": 0.6912,
|
| 880 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 881 |
+
"rewards/chosen": -0.0037696107756346464,
|
| 882 |
+
"rewards/margins": 0.008755478076636791,
|
| 883 |
+
"rewards/rejected": -0.01252509094774723,
|
| 884 |
+
"step": 520
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 1.5856394913986538,
|
| 888 |
+
"grad_norm": 13.25,
|
| 889 |
+
"learning_rate": 2.6872391640726276e-07,
|
| 890 |
+
"logits/chosen": -18.349544525146484,
|
| 891 |
+
"logits/rejected": -18.272260665893555,
|
| 892 |
+
"logps/chosen": -496.89117431640625,
|
| 893 |
+
"logps/rejected": -471.0453186035156,
|
| 894 |
+
"loss": 0.696,
|
| 895 |
+
"rewards/accuracies": 0.46406251192092896,
|
| 896 |
+
"rewards/chosen": -0.016991907730698586,
|
| 897 |
+
"rewards/margins": -0.0011090601328760386,
|
| 898 |
+
"rewards/rejected": -0.01588284783065319,
|
| 899 |
+
"step": 530
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 1.6155572176514585,
|
| 903 |
+
"grad_norm": 14.5625,
|
| 904 |
+
"learning_rate": 2.6002181875755233e-07,
|
| 905 |
+
"logits/chosen": -18.489253997802734,
|
| 906 |
+
"logits/rejected": -18.443098068237305,
|
| 907 |
+
"logps/chosen": -492.5693359375,
|
| 908 |
+
"logps/rejected": -482.33306884765625,
|
| 909 |
+
"loss": 0.6967,
|
| 910 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 911 |
+
"rewards/chosen": -0.0029235139954835176,
|
| 912 |
+
"rewards/margins": -0.0017206415068358183,
|
| 913 |
+
"rewards/rejected": -0.0012028723722323775,
|
| 914 |
+
"step": 540
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 1.6454749439042633,
|
| 918 |
+
"grad_norm": 12.6875,
|
| 919 |
+
"learning_rate": 2.5130753815095647e-07,
|
| 920 |
+
"logits/chosen": -18.483600616455078,
|
| 921 |
+
"logits/rejected": -18.2879581451416,
|
| 922 |
+
"logps/chosen": -510.51348876953125,
|
| 923 |
+
"logps/rejected": -490.70196533203125,
|
| 924 |
+
"loss": 0.6949,
|
| 925 |
+
"rewards/accuracies": 0.4921875,
|
| 926 |
+
"rewards/chosen": -0.0024643188808113337,
|
| 927 |
+
"rewards/margins": 0.001509314402937889,
|
| 928 |
+
"rewards/rejected": -0.003973634447902441,
|
| 929 |
+
"step": 550
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 1.675392670157068,
|
| 933 |
+
"grad_norm": 12.0,
|
| 934 |
+
"learning_rate": 2.4259166804436003e-07,
|
| 935 |
+
"logits/chosen": -18.499126434326172,
|
| 936 |
+
"logits/rejected": -18.366363525390625,
|
| 937 |
+
"logps/chosen": -499.53240966796875,
|
| 938 |
+
"logps/rejected": -479.3192443847656,
|
| 939 |
+
"loss": 0.6907,
|
| 940 |
+
"rewards/accuracies": 0.5171874761581421,
|
| 941 |
+
"rewards/chosen": -0.0017261719331145287,
|
| 942 |
+
"rewards/margins": 0.009601183235645294,
|
| 943 |
+
"rewards/rejected": -0.011327354237437248,
|
| 944 |
+
"step": 560
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 1.7053103964098728,
|
| 948 |
+
"grad_norm": 13.5,
|
| 949 |
+
"learning_rate": 2.338848038269135e-07,
|
| 950 |
+
"logits/chosen": -18.419811248779297,
|
| 951 |
+
"logits/rejected": -18.180389404296875,
|
| 952 |
+
"logps/chosen": -502.84686279296875,
|
| 953 |
+
"logps/rejected": -476.781494140625,
|
| 954 |
+
"loss": 0.692,
|
| 955 |
+
"rewards/accuracies": 0.49531251192092896,
|
| 956 |
+
"rewards/chosen": 0.001199430669657886,
|
| 957 |
+
"rewards/margins": 0.007095737848430872,
|
| 958 |
+
"rewards/rejected": -0.005896307062357664,
|
| 959 |
+
"step": 570
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 1.7352281226626776,
|
| 963 |
+
"grad_norm": 14.125,
|
| 964 |
+
"learning_rate": 2.2519752993981804e-07,
|
| 965 |
+
"logits/chosen": -18.500263214111328,
|
| 966 |
+
"logits/rejected": -18.394670486450195,
|
| 967 |
+
"logps/chosen": -514.1810913085938,
|
| 968 |
+
"logps/rejected": -490.6826171875,
|
| 969 |
+
"loss": 0.7018,
|
| 970 |
+
"rewards/accuracies": 0.42656248807907104,
|
| 971 |
+
"rewards/chosen": -0.007822760380804539,
|
| 972 |
+
"rewards/margins": -0.011961746029555798,
|
| 973 |
+
"rewards/rejected": 0.004138986114412546,
|
| 974 |
+
"step": 580
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 1.7651458489154823,
|
| 978 |
+
"grad_norm": 14.0,
|
| 979 |
+
"learning_rate": 2.1654040700942112e-07,
|
| 980 |
+
"logits/chosen": -18.530935287475586,
|
| 981 |
+
"logits/rejected": -18.39280891418457,
|
| 982 |
+
"logps/chosen": -511.75115966796875,
|
| 983 |
+
"logps/rejected": -486.78564453125,
|
| 984 |
+
"loss": 0.6949,
|
| 985 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 986 |
+
"rewards/chosen": 0.006620324216783047,
|
| 987 |
+
"rewards/margins": 0.0015118361916393042,
|
| 988 |
+
"rewards/rejected": 0.005108488257974386,
|
| 989 |
+
"step": 590
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 1.795063575168287,
|
| 993 |
+
"grad_norm": 13.75,
|
| 994 |
+
"learning_rate": 2.0792395900926155e-07,
|
| 995 |
+
"logits/chosen": -18.366960525512695,
|
| 996 |
+
"logits/rejected": -18.27281379699707,
|
| 997 |
+
"logps/chosen": -495.75445556640625,
|
| 998 |
+
"logps/rejected": -479.6105041503906,
|
| 999 |
+
"loss": 0.6952,
|
| 1000 |
+
"rewards/accuracies": 0.4703125059604645,
|
| 1001 |
+
"rewards/chosen": -0.01034991629421711,
|
| 1002 |
+
"rewards/margins": 0.0005709372344426811,
|
| 1003 |
+
"rewards/rejected": -0.010920852422714233,
|
| 1004 |
+
"step": 600
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 1.795063575168287,
|
| 1008 |
+
"eval_logits/chosen": -18.48212432861328,
|
| 1009 |
+
"eval_logits/rejected": -18.363605499267578,
|
| 1010 |
+
"eval_logps/chosen": -494.263916015625,
|
| 1011 |
+
"eval_logps/rejected": -478.8390808105469,
|
| 1012 |
+
"eval_loss": 0.695127010345459,
|
| 1013 |
+
"eval_rewards/accuracies": 0.4882287085056305,
|
| 1014 |
+
"eval_rewards/chosen": -0.0046835425309836864,
|
| 1015 |
+
"eval_rewards/margins": 0.0010564102558419108,
|
| 1016 |
+
"eval_rewards/rejected": -0.005739952437579632,
|
| 1017 |
+
"eval_runtime": 842.133,
|
| 1018 |
+
"eval_samples_per_second": 6.351,
|
| 1019 |
+
"eval_steps_per_second": 0.794,
|
| 1020 |
+
"step": 600
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 1.824981301421092,
|
| 1024 |
+
"grad_norm": 13.3125,
|
| 1025 |
+
"learning_rate": 1.9935866046667134e-07,
|
| 1026 |
+
"logits/chosen": -18.486186981201172,
|
| 1027 |
+
"logits/rejected": -18.327022552490234,
|
| 1028 |
+
"logps/chosen": -518.8204956054688,
|
| 1029 |
+
"logps/rejected": -491.86737060546875,
|
| 1030 |
+
"loss": 0.6928,
|
| 1031 |
+
"rewards/accuracies": 0.5078125,
|
| 1032 |
+
"rewards/chosen": -0.0042625004425644875,
|
| 1033 |
+
"rewards/margins": 0.00539636705070734,
|
| 1034 |
+
"rewards/rejected": -0.009658867493271828,
|
| 1035 |
+
"step": 610
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 1.8548990276738968,
|
| 1039 |
+
"grad_norm": 13.0625,
|
| 1040 |
+
"learning_rate": 1.9085492372948678e-07,
|
| 1041 |
+
"logits/chosen": -18.573169708251953,
|
| 1042 |
+
"logits/rejected": -18.4843692779541,
|
| 1043 |
+
"logps/chosen": -523.7990112304688,
|
| 1044 |
+
"logps/rejected": -502.0945739746094,
|
| 1045 |
+
"loss": 0.6918,
|
| 1046 |
+
"rewards/accuracies": 0.504687488079071,
|
| 1047 |
+
"rewards/chosen": -0.007377514149993658,
|
| 1048 |
+
"rewards/margins": 0.00755069125443697,
|
| 1049 |
+
"rewards/rejected": -0.01492820493876934,
|
| 1050 |
+
"step": 620
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 1.8848167539267016,
|
| 1054 |
+
"grad_norm": 13.4375,
|
| 1055 |
+
"learning_rate": 1.824230863083477e-07,
|
| 1056 |
+
"logits/chosen": -18.435693740844727,
|
| 1057 |
+
"logits/rejected": -18.278911590576172,
|
| 1058 |
+
"logps/chosen": -500.37103271484375,
|
| 1059 |
+
"logps/rejected": -478.5856018066406,
|
| 1060 |
+
"loss": 0.6983,
|
| 1061 |
+
"rewards/accuracies": 0.46406251192092896,
|
| 1062 |
+
"rewards/chosen": -0.014031234197318554,
|
| 1063 |
+
"rewards/margins": -0.005335616413503885,
|
| 1064 |
+
"rewards/rejected": -0.008695616386830807,
|
| 1065 |
+
"step": 630
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 1.9147344801795063,
|
| 1069 |
+
"grad_norm": 11.75,
|
| 1070 |
+
"learning_rate": 1.7407339830997154e-07,
|
| 1071 |
+
"logits/chosen": -18.601871490478516,
|
| 1072 |
+
"logits/rejected": -18.49062156677246,
|
| 1073 |
+
"logps/chosen": -509.70428466796875,
|
| 1074 |
+
"logps/rejected": -490.6293029785156,
|
| 1075 |
+
"loss": 0.6979,
|
| 1076 |
+
"rewards/accuracies": 0.4515624940395355,
|
| 1077 |
+
"rewards/chosen": -0.005590735003352165,
|
| 1078 |
+
"rewards/margins": -0.00480449665337801,
|
| 1079 |
+
"rewards/rejected": -0.0007862389320507646,
|
| 1080 |
+
"step": 640
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 1.9446522064323113,
|
| 1084 |
+
"grad_norm": 14.0,
|
| 1085 |
+
"learning_rate": 1.658160099766799e-07,
|
| 1086 |
+
"logits/chosen": -18.551902770996094,
|
| 1087 |
+
"logits/rejected": -18.38374900817871,
|
| 1088 |
+
"logps/chosen": -494.0267639160156,
|
| 1089 |
+
"logps/rejected": -469.78143310546875,
|
| 1090 |
+
"loss": 0.6993,
|
| 1091 |
+
"rewards/accuracies": 0.4515624940395355,
|
| 1092 |
+
"rewards/chosen": -0.014440399594604969,
|
| 1093 |
+
"rewards/margins": -0.007908724248409271,
|
| 1094 |
+
"rewards/rejected": -0.006531675346195698,
|
| 1095 |
+
"step": 650
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 1.974569932685116,
|
| 1099 |
+
"grad_norm": 13.4375,
|
| 1100 |
+
"learning_rate": 1.576609593473246e-07,
|
| 1101 |
+
"logits/chosen": -18.249267578125,
|
| 1102 |
+
"logits/rejected": -18.19643783569336,
|
| 1103 |
+
"logps/chosen": -497.38641357421875,
|
| 1104 |
+
"logps/rejected": -477.0294494628906,
|
| 1105 |
+
"loss": 0.695,
|
| 1106 |
+
"rewards/accuracies": 0.44999998807907104,
|
| 1107 |
+
"rewards/chosen": -0.010250954888761044,
|
| 1108 |
+
"rewards/margins": 0.0014018019428476691,
|
| 1109 |
+
"rewards/rejected": -0.011652758345007896,
|
| 1110 |
+
"step": 660
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 2.004487658937921,
|
| 1114 |
+
"grad_norm": 13.125,
|
| 1115 |
+
"learning_rate": 1.4961816005461222e-07,
|
| 1116 |
+
"logits/chosen": -18.5499324798584,
|
| 1117 |
+
"logits/rejected": -18.45195198059082,
|
| 1118 |
+
"logps/chosen": -503.68096923828125,
|
| 1119 |
+
"logps/rejected": -480.076904296875,
|
| 1120 |
+
"loss": 0.695,
|
| 1121 |
+
"rewards/accuracies": 0.49531251192092896,
|
| 1122 |
+
"rewards/chosen": -0.003980870358645916,
|
| 1123 |
+
"rewards/margins": 0.0012054868275299668,
|
| 1124 |
+
"rewards/rejected": -0.005186358001083136,
|
| 1125 |
+
"step": 670
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 2.0344053851907256,
|
| 1129 |
+
"grad_norm": 11.9375,
|
| 1130 |
+
"learning_rate": 1.416973892736638e-07,
|
| 1131 |
+
"logits/chosen": -18.451631546020508,
|
| 1132 |
+
"logits/rejected": -18.396869659423828,
|
| 1133 |
+
"logps/chosen": -498.55267333984375,
|
| 1134 |
+
"logps/rejected": -484.3282165527344,
|
| 1135 |
+
"loss": 0.6977,
|
| 1136 |
+
"rewards/accuracies": 0.4625000059604645,
|
| 1137 |
+
"rewards/chosen": -0.014167798683047295,
|
| 1138 |
+
"rewards/margins": -0.00439287768676877,
|
| 1139 |
+
"rewards/rejected": -0.009774921461939812,
|
| 1140 |
+
"step": 680
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 2.0643231114435303,
|
| 1144 |
+
"grad_norm": 12.875,
|
| 1145 |
+
"learning_rate": 1.339082758364569e-07,
|
| 1146 |
+
"logits/chosen": -18.43014907836914,
|
| 1147 |
+
"logits/rejected": -18.314260482788086,
|
| 1148 |
+
"logps/chosen": -495.68145751953125,
|
| 1149 |
+
"logps/rejected": -485.66632080078125,
|
| 1150 |
+
"loss": 0.6979,
|
| 1151 |
+
"rewards/accuracies": 0.4625000059604645,
|
| 1152 |
+
"rewards/chosen": -0.017589876428246498,
|
| 1153 |
+
"rewards/margins": -0.0047332653775811195,
|
| 1154 |
+
"rewards/rejected": -0.012856610119342804,
|
| 1155 |
+
"step": 690
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 2.094240837696335,
|
| 1159 |
+
"grad_norm": 12.75,
|
| 1160 |
+
"learning_rate": 1.2626028852660026e-07,
|
| 1161 |
+
"logits/chosen": -18.41107749938965,
|
| 1162 |
+
"logits/rejected": -18.227169036865234,
|
| 1163 |
+
"logps/chosen": -505.9676818847656,
|
| 1164 |
+
"logps/rejected": -484.62548828125,
|
| 1165 |
+
"loss": 0.6947,
|
| 1166 |
+
"rewards/accuracies": 0.4765625,
|
| 1167 |
+
"rewards/chosen": -0.014858139678835869,
|
| 1168 |
+
"rewards/margins": 0.0021128151565790176,
|
| 1169 |
+
"rewards/rejected": -0.016970956698060036,
|
| 1170 |
+
"step": 700
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 2.094240837696335,
|
| 1174 |
+
"eval_logits/chosen": -18.482025146484375,
|
| 1175 |
+
"eval_logits/rejected": -18.363367080688477,
|
| 1176 |
+
"eval_logps/chosen": -494.2701416015625,
|
| 1177 |
+
"eval_logps/rejected": -478.8379211425781,
|
| 1178 |
+
"eval_loss": 0.6954838633537292,
|
| 1179 |
+
"eval_rewards/accuracies": 0.4822496175765991,
|
| 1180 |
+
"eval_rewards/chosen": -0.005303152371197939,
|
| 1181 |
+
"eval_rewards/margins": 0.0003158066247124225,
|
| 1182 |
+
"eval_rewards/rejected": -0.005618959199637175,
|
| 1183 |
+
"eval_runtime": 841.92,
|
| 1184 |
+
"eval_samples_per_second": 6.352,
|
| 1185 |
+
"eval_steps_per_second": 0.795,
|
| 1186 |
+
"step": 700
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 2.12415856394914,
|
| 1190 |
+
"grad_norm": 13.375,
|
| 1191 |
+
"learning_rate": 1.1876272456867081e-07,
|
| 1192 |
+
"logits/chosen": -18.531566619873047,
|
| 1193 |
+
"logits/rejected": -18.326671600341797,
|
| 1194 |
+
"logps/chosen": -510.41888427734375,
|
| 1195 |
+
"logps/rejected": -488.3369140625,
|
| 1196 |
+
"loss": 0.6989,
|
| 1197 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 1198 |
+
"rewards/chosen": -0.008112497627735138,
|
| 1199 |
+
"rewards/margins": -0.006051483564078808,
|
| 1200 |
+
"rewards/rejected": -0.002061013597995043,
|
| 1201 |
+
"step": 710
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 2.1540762902019446,
|
| 1205 |
+
"grad_norm": 13.0,
|
| 1206 |
+
"learning_rate": 1.114246983261046e-07,
|
| 1207 |
+
"logits/chosen": -18.460107803344727,
|
| 1208 |
+
"logits/rejected": -18.281509399414062,
|
| 1209 |
+
"logps/chosen": -493.8604431152344,
|
| 1210 |
+
"logps/rejected": -470.369140625,
|
| 1211 |
+
"loss": 0.697,
|
| 1212 |
+
"rewards/accuracies": 0.46406251192092896,
|
| 1213 |
+
"rewards/chosen": -0.0026031527668237686,
|
| 1214 |
+
"rewards/margins": -0.0029606991447508335,
|
| 1215 |
+
"rewards/rejected": 0.0003575454466044903,
|
| 1216 |
+
"step": 720
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 2.1839940164547493,
|
| 1220 |
+
"grad_norm": 14.625,
|
| 1221 |
+
"learning_rate": 1.0425513022138202e-07,
|
| 1222 |
+
"logits/chosen": -18.540891647338867,
|
| 1223 |
+
"logits/rejected": -18.398468017578125,
|
| 1224 |
+
"logps/chosen": -490.9407653808594,
|
| 1225 |
+
"logps/rejected": -473.2347717285156,
|
| 1226 |
+
"loss": 0.693,
|
| 1227 |
+
"rewards/accuracies": 0.4859375059604645,
|
| 1228 |
+
"rewards/chosen": -0.0007933437591418624,
|
| 1229 |
+
"rewards/margins": 0.005172435659915209,
|
| 1230 |
+
"rewards/rejected": -0.005965779535472393,
|
| 1231 |
+
"step": 730
|
| 1232 |
+
},
|
| 1233 |
+
{
|
| 1234 |
+
"epoch": 2.213911742707554,
|
| 1235 |
+
"grad_norm": 13.4375,
|
| 1236 |
+
"learning_rate": 9.726273589197565e-08,
|
| 1237 |
+
"logits/chosen": -18.563766479492188,
|
| 1238 |
+
"logits/rejected": -18.471820831298828,
|
| 1239 |
+
"logps/chosen": -508.5835876464844,
|
| 1240 |
+
"logps/rejected": -494.5013122558594,
|
| 1241 |
+
"loss": 0.6915,
|
| 1242 |
+
"rewards/accuracies": 0.5,
|
| 1243 |
+
"rewards/chosen": -0.00014840513176750392,
|
| 1244 |
+
"rewards/margins": 0.008764137513935566,
|
| 1245 |
+
"rewards/rejected": -0.008912542834877968,
|
| 1246 |
+
"step": 740
|
| 1247 |
+
},
|
| 1248 |
+
{
|
| 1249 |
+
"epoch": 2.243829468960359,
|
| 1250 |
+
"grad_norm": 16.25,
|
| 1251 |
+
"learning_rate": 9.045601559524434e-08,
|
| 1252 |
+
"logits/chosen": -18.197242736816406,
|
| 1253 |
+
"logits/rejected": -18.150014877319336,
|
| 1254 |
+
"logps/chosen": -496.7591247558594,
|
| 1255 |
+
"logps/rejected": -475.01141357421875,
|
| 1256 |
+
"loss": 0.6983,
|
| 1257 |
+
"rewards/accuracies": 0.4609375,
|
| 1258 |
+
"rewards/chosen": -0.005830574780702591,
|
| 1259 |
+
"rewards/margins": -0.005549018736928701,
|
| 1260 |
+
"rewards/rejected": -0.0002815568877849728,
|
| 1261 |
+
"step": 750
|
| 1262 |
+
},
|
| 1263 |
+
{
|
| 1264 |
+
"epoch": 2.273747195213164,
|
| 1265 |
+
"grad_norm": 12.8125,
|
| 1266 |
+
"learning_rate": 8.384324387515202e-08,
|
| 1267 |
+
"logits/chosen": -18.513484954833984,
|
| 1268 |
+
"logits/rejected": -18.47811508178711,
|
| 1269 |
+
"logps/chosen": -509.2792053222656,
|
| 1270 |
+
"logps/rejected": -504.4214782714844,
|
| 1271 |
+
"loss": 0.6973,
|
| 1272 |
+
"rewards/accuracies": 0.4703125059604645,
|
| 1273 |
+
"rewards/chosen": -0.0049002873711287975,
|
| 1274 |
+
"rewards/margins": -0.0031452118419110775,
|
| 1275 |
+
"rewards/rejected": -0.001755074947141111,
|
| 1276 |
+
"step": 760
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 2.303664921465969,
|
| 1280 |
+
"grad_norm": 13.6875,
|
| 1281 |
+
"learning_rate": 7.743245950337365e-08,
|
| 1282 |
+
"logits/chosen": -18.391809463500977,
|
| 1283 |
+
"logits/rejected": -18.290664672851562,
|
| 1284 |
+
"logps/chosen": -499.37353515625,
|
| 1285 |
+
"logps/rejected": -488.48388671875,
|
| 1286 |
+
"loss": 0.6934,
|
| 1287 |
+
"rewards/accuracies": 0.503125011920929,
|
| 1288 |
+
"rewards/chosen": -0.013569754548370838,
|
| 1289 |
+
"rewards/margins": 0.004729210399091244,
|
| 1290 |
+
"rewards/rejected": -0.018298964947462082,
|
| 1291 |
+
"step": 770
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 2.3335826477187736,
|
| 1295 |
+
"grad_norm": 11.9375,
|
| 1296 |
+
"learning_rate": 7.123145570701638e-08,
|
| 1297 |
+
"logits/chosen": -18.322669982910156,
|
| 1298 |
+
"logits/rejected": -18.09329605102539,
|
| 1299 |
+
"logps/chosen": -498.12548828125,
|
| 1300 |
+
"logps/rejected": -465.91497802734375,
|
| 1301 |
+
"loss": 0.6925,
|
| 1302 |
+
"rewards/accuracies": 0.4703125059604645,
|
| 1303 |
+
"rewards/chosen": -0.011946515180170536,
|
| 1304 |
+
"rewards/margins": 0.006278673652559519,
|
| 1305 |
+
"rewards/rejected": -0.018225187435746193,
|
| 1306 |
+
"step": 780
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 2.3635003739715783,
|
| 1310 |
+
"grad_norm": 12.0,
|
| 1311 |
+
"learning_rate": 6.524777069483525e-08,
|
| 1312 |
+
"logits/chosen": -18.426311492919922,
|
| 1313 |
+
"logits/rejected": -18.294950485229492,
|
| 1314 |
+
"logps/chosen": -502.55810546875,
|
| 1315 |
+
"logps/rejected": -476.4676818847656,
|
| 1316 |
+
"loss": 0.7011,
|
| 1317 |
+
"rewards/accuracies": 0.44062501192092896,
|
| 1318 |
+
"rewards/chosen": -0.02296162210404873,
|
| 1319 |
+
"rewards/margins": -0.01101845782250166,
|
| 1320 |
+
"rewards/rejected": -0.01194316241890192,
|
| 1321 |
+
"step": 790
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"epoch": 2.393418100224383,
|
| 1325 |
+
"grad_norm": 14.75,
|
| 1326 |
+
"learning_rate": 5.948867849345979e-08,
|
| 1327 |
+
"logits/chosen": -18.55495834350586,
|
| 1328 |
+
"logits/rejected": -18.406208038330078,
|
| 1329 |
+
"logps/chosen": -493.659423828125,
|
| 1330 |
+
"logps/rejected": -478.211181640625,
|
| 1331 |
+
"loss": 0.6995,
|
| 1332 |
+
"rewards/accuracies": 0.45625001192092896,
|
| 1333 |
+
"rewards/chosen": -0.009105198085308075,
|
| 1334 |
+
"rewards/margins": -0.007736638188362122,
|
| 1335 |
+
"rewards/rejected": -0.001368560828268528,
|
| 1336 |
+
"step": 800
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 2.393418100224383,
|
| 1340 |
+
"eval_logits/chosen": -18.481842041015625,
|
| 1341 |
+
"eval_logits/rejected": -18.36322021484375,
|
| 1342 |
+
"eval_logps/chosen": -494.27740478515625,
|
| 1343 |
+
"eval_logps/rejected": -478.857421875,
|
| 1344 |
+
"eval_loss": 0.6948285102844238,
|
| 1345 |
+
"eval_rewards/accuracies": 0.49177876114845276,
|
| 1346 |
+
"eval_rewards/chosen": -0.006035235244780779,
|
| 1347 |
+
"eval_rewards/margins": 0.0015357759548351169,
|
| 1348 |
+
"eval_rewards/rejected": -0.007571011781692505,
|
| 1349 |
+
"eval_runtime": 842.3774,
|
| 1350 |
+
"eval_samples_per_second": 6.349,
|
| 1351 |
+
"eval_steps_per_second": 0.794,
|
| 1352 |
+
"step": 800
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 2.423335826477188,
|
| 1356 |
+
"grad_norm": 15.5,
|
| 1357 |
+
"learning_rate": 5.396118010477274e-08,
|
| 1358 |
+
"logits/chosen": -18.395492553710938,
|
| 1359 |
+
"logits/rejected": -18.27509117126465,
|
| 1360 |
+
"logps/chosen": -502.3111877441406,
|
| 1361 |
+
"logps/rejected": -475.66973876953125,
|
| 1362 |
+
"loss": 0.6989,
|
| 1363 |
+
"rewards/accuracies": 0.4453125,
|
| 1364 |
+
"rewards/chosen": -0.01698307693004608,
|
| 1365 |
+
"rewards/margins": -0.006895692553371191,
|
| 1366 |
+
"rewards/rejected": -0.010087382979691029,
|
| 1367 |
+
"step": 810
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 2.4532535527299926,
|
| 1371 |
+
"grad_norm": 15.0625,
|
| 1372 |
+
"learning_rate": 4.867199499518873e-08,
|
| 1373 |
+
"logits/chosen": -18.451091766357422,
|
| 1374 |
+
"logits/rejected": -18.34830093383789,
|
| 1375 |
+
"logps/chosen": -500.35595703125,
|
| 1376 |
+
"logps/rejected": -482.3310546875,
|
| 1377 |
+
"loss": 0.6971,
|
| 1378 |
+
"rewards/accuracies": 0.44999998807907104,
|
| 1379 |
+
"rewards/chosen": -0.006859651766717434,
|
| 1380 |
+
"rewards/margins": -0.0022591417655348778,
|
| 1381 |
+
"rewards/rejected": -0.004600510932505131,
|
| 1382 |
+
"step": 820
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 2.4831712789827973,
|
| 1386 |
+
"grad_norm": 13.875,
|
| 1387 |
+
"learning_rate": 4.362755292717993e-08,
|
| 1388 |
+
"logits/chosen": -18.62723731994629,
|
| 1389 |
+
"logits/rejected": -18.555469512939453,
|
| 1390 |
+
"logps/chosen": -508.4722595214844,
|
| 1391 |
+
"logps/rejected": -480.99932861328125,
|
| 1392 |
+
"loss": 0.6959,
|
| 1393 |
+
"rewards/accuracies": 0.4515624940395355,
|
| 1394 |
+
"rewards/chosen": 7.3916649853345e-05,
|
| 1395 |
+
"rewards/margins": -0.0006395616801455617,
|
| 1396 |
+
"rewards/rejected": 0.0007134780171327293,
|
| 1397 |
+
"step": 830
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 2.513089005235602,
|
| 1401 |
+
"grad_norm": 15.1875,
|
| 1402 |
+
"learning_rate": 3.883398614297831e-08,
|
| 1403 |
+
"logits/chosen": -18.562936782836914,
|
| 1404 |
+
"logits/rejected": -18.407032012939453,
|
| 1405 |
+
"logps/chosen": -518.3756103515625,
|
| 1406 |
+
"logps/rejected": -503.34063720703125,
|
| 1407 |
+
"loss": 0.6958,
|
| 1408 |
+
"rewards/accuracies": 0.4671874940395355,
|
| 1409 |
+
"rewards/chosen": -0.007723680697381496,
|
| 1410 |
+
"rewards/margins": -2.291230521223042e-05,
|
| 1411 |
+
"rewards/rejected": -0.007700768765062094,
|
| 1412 |
+
"step": 840
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 2.543006731488407,
|
| 1416 |
+
"grad_norm": 13.375,
|
| 1417 |
+
"learning_rate": 3.4297121909956485e-08,
|
| 1418 |
+
"logits/chosen": -18.537033081054688,
|
| 1419 |
+
"logits/rejected": -18.43063735961914,
|
| 1420 |
+
"logps/chosen": -490.99017333984375,
|
| 1421 |
+
"logps/rejected": -469.899658203125,
|
| 1422 |
+
"loss": 0.6941,
|
| 1423 |
+
"rewards/accuracies": 0.45625001192092896,
|
| 1424 |
+
"rewards/chosen": -0.006467898841947317,
|
| 1425 |
+
"rewards/margins": 0.003015016671270132,
|
| 1426 |
+
"rewards/rejected": -0.00948291551321745,
|
| 1427 |
+
"step": 850
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 2.5729244577412116,
|
| 1431 |
+
"grad_norm": 14.0,
|
| 1432 |
+
"learning_rate": 3.002247543674921e-08,
|
| 1433 |
+
"logits/chosen": -18.188644409179688,
|
| 1434 |
+
"logits/rejected": -18.097511291503906,
|
| 1435 |
+
"logps/chosen": -497.510986328125,
|
| 1436 |
+
"logps/rejected": -473.9842224121094,
|
| 1437 |
+
"loss": 0.6961,
|
| 1438 |
+
"rewards/accuracies": 0.46875,
|
| 1439 |
+
"rewards/chosen": -0.011478578671813011,
|
| 1440 |
+
"rewards/margins": -0.0014710351824760437,
|
| 1441 |
+
"rewards/rejected": -0.010007544420659542,
|
| 1442 |
+
"step": 860
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 2.6028421839940163,
|
| 1446 |
+
"grad_norm": 14.4375,
|
| 1447 |
+
"learning_rate": 2.6015243168726607e-08,
|
| 1448 |
+
"logits/chosen": -18.49826431274414,
|
| 1449 |
+
"logits/rejected": -18.359790802001953,
|
| 1450 |
+
"logps/chosen": -513.165771484375,
|
| 1451 |
+
"logps/rejected": -497.18487548828125,
|
| 1452 |
+
"loss": 0.6947,
|
| 1453 |
+
"rewards/accuracies": 0.4906249940395355,
|
| 1454 |
+
"rewards/chosen": -0.007126118056476116,
|
| 1455 |
+
"rewards/margins": 0.002322929911315441,
|
| 1456 |
+
"rewards/rejected": -0.009449047967791557,
|
| 1457 |
+
"step": 870
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 2.632759910246821,
|
| 1461 |
+
"grad_norm": 12.4375,
|
| 1462 |
+
"learning_rate": 2.228029647097057e-08,
|
| 1463 |
+
"logits/chosen": -18.447689056396484,
|
| 1464 |
+
"logits/rejected": -18.275293350219727,
|
| 1465 |
+
"logps/chosen": -497.46392822265625,
|
| 1466 |
+
"logps/rejected": -472.00531005859375,
|
| 1467 |
+
"loss": 0.6973,
|
| 1468 |
+
"rewards/accuracies": 0.4765625,
|
| 1469 |
+
"rewards/chosen": -0.00794267375022173,
|
| 1470 |
+
"rewards/margins": -0.0032855805475264788,
|
| 1471 |
+
"rewards/rejected": -0.004657094366848469,
|
| 1472 |
+
"step": 880
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 2.662677636499626,
|
| 1476 |
+
"grad_norm": 12.6875,
|
| 1477 |
+
"learning_rate": 1.882217570643216e-08,
|
| 1478 |
+
"logits/chosen": -18.6273193359375,
|
| 1479 |
+
"logits/rejected": -18.57034683227539,
|
| 1480 |
+
"logps/chosen": -493.40191650390625,
|
| 1481 |
+
"logps/rejected": -481.95062255859375,
|
| 1482 |
+
"loss": 0.6963,
|
| 1483 |
+
"rewards/accuracies": 0.49687498807907104,
|
| 1484 |
+
"rewards/chosen": -0.010606282390654087,
|
| 1485 |
+
"rewards/margins": -0.0014992398209869862,
|
| 1486 |
+
"rewards/rejected": -0.009107043035328388,
|
| 1487 |
+
"step": 890
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 2.6925953627524306,
|
| 1491 |
+
"grad_norm": 13.5625,
|
| 1492 |
+
"learning_rate": 1.5645084716469776e-08,
|
| 1493 |
+
"logits/chosen": -18.280813217163086,
|
| 1494 |
+
"logits/rejected": -18.22847557067871,
|
| 1495 |
+
"logps/chosen": -503.5279846191406,
|
| 1496 |
+
"logps/rejected": -490.97320556640625,
|
| 1497 |
+
"loss": 0.6932,
|
| 1498 |
+
"rewards/accuracies": 0.49687498807907104,
|
| 1499 |
+
"rewards/chosen": -0.007966242730617523,
|
| 1500 |
+
"rewards/margins": 0.004499543458223343,
|
| 1501 |
+
"rewards/rejected": -0.012465784326195717,
|
| 1502 |
+
"step": 900
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 2.6925953627524306,
|
| 1506 |
+
"eval_logits/chosen": -18.48170280456543,
|
| 1507 |
+
"eval_logits/rejected": -18.363323211669922,
|
| 1508 |
+
"eval_logps/chosen": -494.2970275878906,
|
| 1509 |
+
"eval_logps/rejected": -478.86920166015625,
|
| 1510 |
+
"eval_loss": 0.6951792240142822,
|
| 1511 |
+
"eval_rewards/accuracies": 0.48374438285827637,
|
| 1512 |
+
"eval_rewards/chosen": -0.007996306754648685,
|
| 1513 |
+
"eval_rewards/margins": 0.0007518637576140463,
|
| 1514 |
+
"eval_rewards/rejected": -0.008748171851038933,
|
| 1515 |
+
"eval_runtime": 842.6667,
|
| 1516 |
+
"eval_samples_per_second": 6.347,
|
| 1517 |
+
"eval_steps_per_second": 0.794,
|
| 1518 |
+
"step": 900
|
| 1519 |
+
},
|
| 1520 |
+
{
|
| 1521 |
+
"epoch": 2.7225130890052354,
|
| 1522 |
+
"grad_norm": 13.375,
|
| 1523 |
+
"learning_rate": 1.2752885710477584e-08,
|
| 1524 |
+
"logits/chosen": -18.436344146728516,
|
| 1525 |
+
"logits/rejected": -18.28668975830078,
|
| 1526 |
+
"logps/chosen": -502.52911376953125,
|
| 1527 |
+
"logps/rejected": -478.9175720214844,
|
| 1528 |
+
"loss": 0.6917,
|
| 1529 |
+
"rewards/accuracies": 0.47968751192092896,
|
| 1530 |
+
"rewards/chosen": -0.0051972828805446625,
|
| 1531 |
+
"rewards/margins": 0.007670181337743998,
|
| 1532 |
+
"rewards/rejected": -0.012867462821304798,
|
| 1533 |
+
"step": 910
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 2.75243081525804,
|
| 1537 |
+
"grad_norm": 14.5625,
|
| 1538 |
+
"learning_rate": 1.014909457081664e-08,
|
| 1539 |
+
"logits/chosen": -18.285552978515625,
|
| 1540 |
+
"logits/rejected": -18.188400268554688,
|
| 1541 |
+
"logps/chosen": -495.7474670410156,
|
| 1542 |
+
"logps/rejected": -482.32086181640625,
|
| 1543 |
+
"loss": 0.7018,
|
| 1544 |
+
"rewards/accuracies": 0.4281249940395355,
|
| 1545 |
+
"rewards/chosen": -0.014425873756408691,
|
| 1546 |
+
"rewards/margins": -0.012442706152796745,
|
| 1547 |
+
"rewards/rejected": -0.0019831680692732334,
|
| 1548 |
+
"step": 920
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 2.7823485415108453,
|
| 1552 |
+
"grad_norm": 13.625,
|
| 1553 |
+
"learning_rate": 7.836876578755758e-09,
|
| 1554 |
+
"logits/chosen": -18.36880874633789,
|
| 1555 |
+
"logits/rejected": -18.273122787475586,
|
| 1556 |
+
"logps/chosen": -505.2574157714844,
|
| 1557 |
+
"logps/rejected": -478.4847717285156,
|
| 1558 |
+
"loss": 0.6956,
|
| 1559 |
+
"rewards/accuracies": 0.4859375059604645,
|
| 1560 |
+
"rewards/chosen": -0.005088160280138254,
|
| 1561 |
+
"rewards/margins": 0.00010951366130029783,
|
| 1562 |
+
"rewards/rejected": -0.005197672639042139,
|
| 1563 |
+
"step": 930
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 2.81226626776365,
|
| 1567 |
+
"grad_norm": 13.125,
|
| 1568 |
+
"learning_rate": 5.819042566618831e-09,
|
| 1569 |
+
"logits/chosen": -18.473098754882812,
|
| 1570 |
+
"logits/rejected": -18.300846099853516,
|
| 1571 |
+
"logps/chosen": -503.1376953125,
|
| 1572 |
+
"logps/rejected": -475.238037109375,
|
| 1573 |
+
"loss": 0.6943,
|
| 1574 |
+
"rewards/accuracies": 0.4749999940395355,
|
| 1575 |
+
"rewards/chosen": 0.0007304133614525199,
|
| 1576 |
+
"rewards/margins": 0.002083559287711978,
|
| 1577 |
+
"rewards/rejected": -0.00135314604267478,
|
| 1578 |
+
"step": 940
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 2.842183994016455,
|
| 1582 |
+
"grad_norm": 12.625,
|
| 1583 |
+
"learning_rate": 4.098045500815106e-09,
|
| 1584 |
+
"logits/chosen": -18.308658599853516,
|
| 1585 |
+
"logits/rejected": -18.19416046142578,
|
| 1586 |
+
"logps/chosen": -495.00604248046875,
|
| 1587 |
+
"logps/rejected": -475.799072265625,
|
| 1588 |
+
"loss": 0.696,
|
| 1589 |
+
"rewards/accuracies": 0.44843751192092896,
|
| 1590 |
+
"rewards/chosen": -0.004273262806236744,
|
| 1591 |
+
"rewards/margins": -0.0009155833977274597,
|
| 1592 |
+
"rewards/rejected": -0.0033576800487935543,
|
| 1593 |
+
"step": 950
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 2.8721017202692596,
|
| 1597 |
+
"grad_norm": 14.5625,
|
| 1598 |
+
"learning_rate": 2.6759774999071673e-09,
|
| 1599 |
+
"logits/chosen": -18.59539794921875,
|
| 1600 |
+
"logits/rejected": -18.472515106201172,
|
| 1601 |
+
"logps/chosen": -514.5612182617188,
|
| 1602 |
+
"logps/rejected": -492.31890869140625,
|
| 1603 |
+
"loss": 0.6964,
|
| 1604 |
+
"rewards/accuracies": 0.4515624940395355,
|
| 1605 |
+
"rewards/chosen": -0.0067207301035523415,
|
| 1606 |
+
"rewards/margins": -0.0017061748076230288,
|
| 1607 |
+
"rewards/rejected": -0.005014555994421244,
|
| 1608 |
+
"step": 960
|
| 1609 |
+
},
|
| 1610 |
+
{
|
| 1611 |
+
"epoch": 2.9020194465220643,
|
| 1612 |
+
"grad_norm": 11.625,
|
| 1613 |
+
"learning_rate": 1.5545672913407914e-09,
|
| 1614 |
+
"logits/chosen": -18.352832794189453,
|
| 1615 |
+
"logits/rejected": -18.20647621154785,
|
| 1616 |
+
"logps/chosen": -501.56317138671875,
|
| 1617 |
+
"logps/rejected": -487.9432678222656,
|
| 1618 |
+
"loss": 0.6959,
|
| 1619 |
+
"rewards/accuracies": 0.5,
|
| 1620 |
+
"rewards/chosen": -0.013429331593215466,
|
| 1621 |
+
"rewards/margins": -0.0001061473521986045,
|
| 1622 |
+
"rewards/rejected": -0.013323183171451092,
|
| 1623 |
+
"step": 970
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 2.931937172774869,
|
| 1627 |
+
"grad_norm": 19.5,
|
| 1628 |
+
"learning_rate": 7.351781099292353e-10,
|
| 1629 |
+
"logits/chosen": -18.442501068115234,
|
| 1630 |
+
"logits/rejected": -18.37118148803711,
|
| 1631 |
+
"logps/chosen": -501.3379821777344,
|
| 1632 |
+
"logps/rejected": -496.15924072265625,
|
| 1633 |
+
"loss": 0.7002,
|
| 1634 |
+
"rewards/accuracies": 0.45625001192092896,
|
| 1635 |
+
"rewards/chosen": -0.01971256360411644,
|
| 1636 |
+
"rewards/margins": -0.00949421338737011,
|
| 1637 |
+
"rewards/rejected": -0.01021835207939148,
|
| 1638 |
+
"step": 980
|
| 1639 |
+
},
|
| 1640 |
+
{
|
| 1641 |
+
"epoch": 2.961854899027674,
|
| 1642 |
+
"grad_norm": 11.5,
|
| 1643 |
+
"learning_rate": 2.1880604064578035e-10,
|
| 1644 |
+
"logits/chosen": -18.448497772216797,
|
| 1645 |
+
"logits/rejected": -18.216806411743164,
|
| 1646 |
+
"logps/chosen": -508.0194396972656,
|
| 1647 |
+
"logps/rejected": -466.27301025390625,
|
| 1648 |
+
"loss": 0.6949,
|
| 1649 |
+
"rewards/accuracies": 0.484375,
|
| 1650 |
+
"rewards/chosen": -0.0018872346263378859,
|
| 1651 |
+
"rewards/margins": 0.0016591917956247926,
|
| 1652 |
+
"rewards/rejected": -0.003546426072716713,
|
| 1653 |
+
"step": 990
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 2.9917726252804786,
|
| 1657 |
+
"grad_norm": 19.5,
|
| 1658 |
+
"learning_rate": 6.07880773983993e-12,
|
| 1659 |
+
"logits/chosen": -18.198387145996094,
|
| 1660 |
+
"logits/rejected": -18.139768600463867,
|
| 1661 |
+
"logps/chosen": -483.9991149902344,
|
| 1662 |
+
"logps/rejected": -473.9156188964844,
|
| 1663 |
+
"loss": 0.6964,
|
| 1664 |
+
"rewards/accuracies": 0.45781248807907104,
|
| 1665 |
+
"rewards/chosen": -0.009544356726109982,
|
| 1666 |
+
"rewards/margins": -0.0014166312757879496,
|
| 1667 |
+
"rewards/rejected": -0.00812772661447525,
|
| 1668 |
+
"step": 1000
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 2.9917726252804786,
|
| 1672 |
+
"eval_logits/chosen": -18.481870651245117,
|
| 1673 |
+
"eval_logits/rejected": -18.36333465576172,
|
| 1674 |
+
"eval_logps/chosen": -494.2957763671875,
|
| 1675 |
+
"eval_logps/rejected": -478.8612060546875,
|
| 1676 |
+
"eval_loss": 0.6955399513244629,
|
| 1677 |
+
"eval_rewards/accuracies": 0.48131540417671204,
|
| 1678 |
+
"eval_rewards/chosen": -0.007866346277296543,
|
| 1679 |
+
"eval_rewards/margins": 8.37442566989921e-05,
|
| 1680 |
+
"eval_rewards/rejected": -0.007950090803205967,
|
| 1681 |
+
"eval_runtime": 843.4786,
|
| 1682 |
+
"eval_samples_per_second": 6.34,
|
| 1683 |
+
"eval_steps_per_second": 0.793,
|
| 1684 |
+
"step": 1000
|
| 1685 |
+
},
|
| 1686 |
+
{
|
| 1687 |
+
"epoch": 2.99775617053104,
|
| 1688 |
+
"step": 1002,
|
| 1689 |
+
"total_flos": 0.0,
|
| 1690 |
+
"train_loss": 0.6953648589328377,
|
| 1691 |
+
"train_runtime": 32693.7742,
|
| 1692 |
+
"train_samples_per_second": 1.963,
|
| 1693 |
+
"train_steps_per_second": 0.031
|
| 1694 |
+
}
|
| 1695 |
+
],
|
| 1696 |
+
"logging_steps": 10,
|
| 1697 |
+
"max_steps": 1002,
|
| 1698 |
+
"num_input_tokens_seen": 0,
|
| 1699 |
+
"num_train_epochs": 3,
|
| 1700 |
+
"save_steps": 100,
|
| 1701 |
+
"stateful_callbacks": {
|
| 1702 |
+
"TrainerControl": {
|
| 1703 |
+
"args": {
|
| 1704 |
+
"should_epoch_stop": false,
|
| 1705 |
+
"should_evaluate": false,
|
| 1706 |
+
"should_log": false,
|
| 1707 |
+
"should_save": true,
|
| 1708 |
+
"should_training_stop": false
|
| 1709 |
+
},
|
| 1710 |
+
"attributes": {}
|
| 1711 |
+
}
|
| 1712 |
+
},
|
| 1713 |
+
"total_flos": 0.0,
|
| 1714 |
+
"train_batch_size": 8,
|
| 1715 |
+
"trial_name": null,
|
| 1716 |
+
"trial_params": null
|
| 1717 |
+
}
|