Model save
Browse files- README.md +74 -0
- all_results.json +9 -0
- generation_config.json +6 -0
- train_results.json +9 -0
- trainer_state.json +1453 -0
README.md
ADDED
|
@@ -0,0 +1,74 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
license: apache-2.0
|
| 4 |
+
base_model: mistralai/Mistral-7B-Instruct-v0.2
|
| 5 |
+
tags:
|
| 6 |
+
- trl
|
| 7 |
+
- simpo
|
| 8 |
+
- generated_from_trainer
|
| 9 |
+
model-index:
|
| 10 |
+
- name: simpo-exps
|
| 11 |
+
results: []
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 15 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 16 |
+
|
| 17 |
+
# simpo-exps
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [mistralai/Mistral-7B-Instruct-v0.2](https://huggingface.co/mistralai/Mistral-7B-Instruct-v0.2) on an unknown dataset.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 0.7909
|
| 22 |
+
- Rewards/chosen: -1.6908
|
| 23 |
+
- Rewards/rejected: -1.8447
|
| 24 |
+
- Rewards/accuracies: 0.6024
|
| 25 |
+
- Rewards/margins: 0.1538
|
| 26 |
+
- Logps/rejected: -0.7379
|
| 27 |
+
- Logps/chosen: -0.6763
|
| 28 |
+
- Logits/rejected: -2.7573
|
| 29 |
+
- Logits/chosen: -2.7593
|
| 30 |
+
|
| 31 |
+
## Model description
|
| 32 |
+
|
| 33 |
+
More information needed
|
| 34 |
+
|
| 35 |
+
## Intended uses & limitations
|
| 36 |
+
|
| 37 |
+
More information needed
|
| 38 |
+
|
| 39 |
+
## Training and evaluation data
|
| 40 |
+
|
| 41 |
+
More information needed
|
| 42 |
+
|
| 43 |
+
## Training procedure
|
| 44 |
+
|
| 45 |
+
### Training hyperparameters
|
| 46 |
+
|
| 47 |
+
The following hyperparameters were used during training:
|
| 48 |
+
- learning_rate: 5e-07
|
| 49 |
+
- train_batch_size: 2
|
| 50 |
+
- eval_batch_size: 4
|
| 51 |
+
- seed: 42
|
| 52 |
+
- distributed_type: multi-GPU
|
| 53 |
+
- num_devices: 4
|
| 54 |
+
- gradient_accumulation_steps: 16
|
| 55 |
+
- total_train_batch_size: 128
|
| 56 |
+
- total_eval_batch_size: 16
|
| 57 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 58 |
+
- lr_scheduler_type: cosine
|
| 59 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 60 |
+
- num_epochs: 1
|
| 61 |
+
|
| 62 |
+
### Training results
|
| 63 |
+
|
| 64 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 65 |
+
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 66 |
+
| 0.7494 | 0.8573 | 400 | 0.7909 | -1.6908 | -1.8447 | 0.6024 | 0.1538 | -0.7379 | -0.6763 | -2.7573 | -2.7593 |
|
| 67 |
+
|
| 68 |
+
|
| 69 |
+
### Framework versions
|
| 70 |
+
|
| 71 |
+
- Transformers 4.44.2
|
| 72 |
+
- Pytorch 2.2.2+cu121
|
| 73 |
+
- Datasets 2.18.0
|
| 74 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9987943737441393,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.7914171927476645,
|
| 5 |
+
"train_runtime": 8402.7408,
|
| 6 |
+
"train_samples": 59720,
|
| 7 |
+
"train_samples_per_second": 7.107,
|
| 8 |
+
"train_steps_per_second": 0.055
|
| 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": "4.44.2"
|
| 6 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9987943737441393,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.7914171927476645,
|
| 5 |
+
"train_runtime": 8402.7408,
|
| 6 |
+
"train_samples": 59720,
|
| 7 |
+
"train_samples_per_second": 7.107,
|
| 8 |
+
"train_steps_per_second": 0.055
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1453 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9987943737441393,
|
| 5 |
+
"eval_steps": 400,
|
| 6 |
+
"global_step": 466,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.010716677829872739,
|
| 13 |
+
"grad_norm": 27.862872452731896,
|
| 14 |
+
"learning_rate": 5.3191489361702123e-08,
|
| 15 |
+
"logits/chosen": -2.2512857913970947,
|
| 16 |
+
"logits/rejected": -2.264533519744873,
|
| 17 |
+
"logps/chosen": -0.5397141575813293,
|
| 18 |
+
"logps/rejected": -0.540291428565979,
|
| 19 |
+
"loss": 0.8648,
|
| 20 |
+
"rewards/accuracies": 0.4749999940395355,
|
| 21 |
+
"rewards/chosen": -1.3492851257324219,
|
| 22 |
+
"rewards/margins": 0.0014432668685913086,
|
| 23 |
+
"rewards/rejected": -1.3507286310195923,
|
| 24 |
+
"step": 5
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.021433355659745478,
|
| 28 |
+
"grad_norm": 35.31926398689821,
|
| 29 |
+
"learning_rate": 1.0638297872340425e-07,
|
| 30 |
+
"logits/chosen": -2.308763027191162,
|
| 31 |
+
"logits/rejected": -2.2997522354125977,
|
| 32 |
+
"logps/chosen": -0.5613104701042175,
|
| 33 |
+
"logps/rejected": -0.6120178699493408,
|
| 34 |
+
"loss": 0.8465,
|
| 35 |
+
"rewards/accuracies": 0.512499988079071,
|
| 36 |
+
"rewards/chosen": -1.4032763242721558,
|
| 37 |
+
"rewards/margins": 0.1267683506011963,
|
| 38 |
+
"rewards/rejected": -1.5300445556640625,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.032150033489618215,
|
| 43 |
+
"grad_norm": 22.32409182123902,
|
| 44 |
+
"learning_rate": 1.5957446808510638e-07,
|
| 45 |
+
"logits/chosen": -2.272329330444336,
|
| 46 |
+
"logits/rejected": -2.265493392944336,
|
| 47 |
+
"logps/chosen": -0.535502552986145,
|
| 48 |
+
"logps/rejected": -0.5462267994880676,
|
| 49 |
+
"loss": 0.8411,
|
| 50 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 51 |
+
"rewards/chosen": -1.3387562036514282,
|
| 52 |
+
"rewards/margins": 0.026810690760612488,
|
| 53 |
+
"rewards/rejected": -1.3655668497085571,
|
| 54 |
+
"step": 15
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.042866711319490956,
|
| 58 |
+
"grad_norm": 24.85973432029644,
|
| 59 |
+
"learning_rate": 2.127659574468085e-07,
|
| 60 |
+
"logits/chosen": -2.275291919708252,
|
| 61 |
+
"logits/rejected": -2.273228168487549,
|
| 62 |
+
"logps/chosen": -0.5528542399406433,
|
| 63 |
+
"logps/rejected": -0.5709605813026428,
|
| 64 |
+
"loss": 0.8452,
|
| 65 |
+
"rewards/accuracies": 0.512499988079071,
|
| 66 |
+
"rewards/chosen": -1.3821356296539307,
|
| 67 |
+
"rewards/margins": 0.04526572674512863,
|
| 68 |
+
"rewards/rejected": -1.4274013042449951,
|
| 69 |
+
"step": 20
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.0535833891493637,
|
| 73 |
+
"grad_norm": 66.37399964080045,
|
| 74 |
+
"learning_rate": 2.659574468085106e-07,
|
| 75 |
+
"logits/chosen": -2.227910041809082,
|
| 76 |
+
"logits/rejected": -2.2429375648498535,
|
| 77 |
+
"logps/chosen": -0.5119736790657043,
|
| 78 |
+
"logps/rejected": -0.5303855538368225,
|
| 79 |
+
"loss": 0.8369,
|
| 80 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 81 |
+
"rewards/chosen": -1.279934287071228,
|
| 82 |
+
"rewards/margins": 0.046029649674892426,
|
| 83 |
+
"rewards/rejected": -1.3259638547897339,
|
| 84 |
+
"step": 25
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.06430006697923643,
|
| 88 |
+
"grad_norm": 19.386202836307778,
|
| 89 |
+
"learning_rate": 3.1914893617021275e-07,
|
| 90 |
+
"logits/chosen": -2.3197097778320312,
|
| 91 |
+
"logits/rejected": -2.33433198928833,
|
| 92 |
+
"logps/chosen": -0.5134044885635376,
|
| 93 |
+
"logps/rejected": -0.5359271764755249,
|
| 94 |
+
"loss": 0.8261,
|
| 95 |
+
"rewards/accuracies": 0.5,
|
| 96 |
+
"rewards/chosen": -1.2835114002227783,
|
| 97 |
+
"rewards/margins": 0.056306738406419754,
|
| 98 |
+
"rewards/rejected": -1.339818000793457,
|
| 99 |
+
"step": 30
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.07501674480910918,
|
| 103 |
+
"grad_norm": 20.601240257458635,
|
| 104 |
+
"learning_rate": 3.7234042553191484e-07,
|
| 105 |
+
"logits/chosen": -2.3346030712127686,
|
| 106 |
+
"logits/rejected": -2.3339004516601562,
|
| 107 |
+
"logps/chosen": -0.45776882767677307,
|
| 108 |
+
"logps/rejected": -0.4768803119659424,
|
| 109 |
+
"loss": 0.8405,
|
| 110 |
+
"rewards/accuracies": 0.518750011920929,
|
| 111 |
+
"rewards/chosen": -1.144421935081482,
|
| 112 |
+
"rewards/margins": 0.0477786585688591,
|
| 113 |
+
"rewards/rejected": -1.192200779914856,
|
| 114 |
+
"step": 35
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.08573342263898191,
|
| 118 |
+
"grad_norm": 24.697322476213458,
|
| 119 |
+
"learning_rate": 4.25531914893617e-07,
|
| 120 |
+
"logits/chosen": -2.255758047103882,
|
| 121 |
+
"logits/rejected": -2.2820348739624023,
|
| 122 |
+
"logps/chosen": -0.5251036882400513,
|
| 123 |
+
"logps/rejected": -0.5213593244552612,
|
| 124 |
+
"loss": 0.8474,
|
| 125 |
+
"rewards/accuracies": 0.4749999940395355,
|
| 126 |
+
"rewards/chosen": -1.312759280204773,
|
| 127 |
+
"rewards/margins": -0.009361037984490395,
|
| 128 |
+
"rewards/rejected": -1.3033982515335083,
|
| 129 |
+
"step": 40
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.09645010046885466,
|
| 133 |
+
"grad_norm": 19.817830752667202,
|
| 134 |
+
"learning_rate": 4.787234042553192e-07,
|
| 135 |
+
"logits/chosen": -2.3332431316375732,
|
| 136 |
+
"logits/rejected": -2.3259410858154297,
|
| 137 |
+
"logps/chosen": -0.4812852442264557,
|
| 138 |
+
"logps/rejected": -0.48691949248313904,
|
| 139 |
+
"loss": 0.8221,
|
| 140 |
+
"rewards/accuracies": 0.543749988079071,
|
| 141 |
+
"rewards/chosen": -1.2032132148742676,
|
| 142 |
+
"rewards/margins": 0.014085566624999046,
|
| 143 |
+
"rewards/rejected": -1.2172987461090088,
|
| 144 |
+
"step": 45
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.1071667782987274,
|
| 148 |
+
"grad_norm": 19.66835689612492,
|
| 149 |
+
"learning_rate": 4.999367579541646e-07,
|
| 150 |
+
"logits/chosen": -2.2813732624053955,
|
| 151 |
+
"logits/rejected": -2.271610975265503,
|
| 152 |
+
"logps/chosen": -0.45226341485977173,
|
| 153 |
+
"logps/rejected": -0.46558040380477905,
|
| 154 |
+
"loss": 0.8299,
|
| 155 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 156 |
+
"rewards/chosen": -1.1306583881378174,
|
| 157 |
+
"rewards/margins": 0.03329245001077652,
|
| 158 |
+
"rewards/rejected": -1.1639509201049805,
|
| 159 |
+
"step": 50
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.11788345612860013,
|
| 163 |
+
"grad_norm": 18.423253141965084,
|
| 164 |
+
"learning_rate": 4.99550394651012e-07,
|
| 165 |
+
"logits/chosen": -2.239288806915283,
|
| 166 |
+
"logits/rejected": -2.2441604137420654,
|
| 167 |
+
"logps/chosen": -0.4442782998085022,
|
| 168 |
+
"logps/rejected": -0.46781450510025024,
|
| 169 |
+
"loss": 0.8258,
|
| 170 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 171 |
+
"rewards/chosen": -1.110695719718933,
|
| 172 |
+
"rewards/margins": 0.05884043127298355,
|
| 173 |
+
"rewards/rejected": -1.1695362329483032,
|
| 174 |
+
"step": 55
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.12860013395847286,
|
| 178 |
+
"grad_norm": 19.559581246861793,
|
| 179 |
+
"learning_rate": 4.988133448023071e-07,
|
| 180 |
+
"logits/chosen": -2.3177483081817627,
|
| 181 |
+
"logits/rejected": -2.3303709030151367,
|
| 182 |
+
"logps/chosen": -0.4191853106021881,
|
| 183 |
+
"logps/rejected": -0.41467294096946716,
|
| 184 |
+
"loss": 0.8268,
|
| 185 |
+
"rewards/accuracies": 0.4124999940395355,
|
| 186 |
+
"rewards/chosen": -1.0479633808135986,
|
| 187 |
+
"rewards/margins": -0.01128093060106039,
|
| 188 |
+
"rewards/rejected": -1.036682367324829,
|
| 189 |
+
"step": 60
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.13931681178834562,
|
| 193 |
+
"grad_norm": 24.533483296865708,
|
| 194 |
+
"learning_rate": 4.977266441646379e-07,
|
| 195 |
+
"logits/chosen": -2.283264636993408,
|
| 196 |
+
"logits/rejected": -2.285848617553711,
|
| 197 |
+
"logps/chosen": -0.4495847225189209,
|
| 198 |
+
"logps/rejected": -0.46883049607276917,
|
| 199 |
+
"loss": 0.8155,
|
| 200 |
+
"rewards/accuracies": 0.4937500059604645,
|
| 201 |
+
"rewards/chosen": -1.1239618062973022,
|
| 202 |
+
"rewards/margins": 0.04811444133520126,
|
| 203 |
+
"rewards/rejected": -1.1720762252807617,
|
| 204 |
+
"step": 65
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.15003348961821836,
|
| 208 |
+
"grad_norm": 18.186755862107237,
|
| 209 |
+
"learning_rate": 4.962918198495654e-07,
|
| 210 |
+
"logits/chosen": -2.331956624984741,
|
| 211 |
+
"logits/rejected": -2.3319480419158936,
|
| 212 |
+
"logps/chosen": -0.46049419045448303,
|
| 213 |
+
"logps/rejected": -0.48135805130004883,
|
| 214 |
+
"loss": 0.8145,
|
| 215 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 216 |
+
"rewards/chosen": -1.1512354612350464,
|
| 217 |
+
"rewards/margins": 0.05215970426797867,
|
| 218 |
+
"rewards/rejected": -1.203395128250122,
|
| 219 |
+
"step": 70
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.1607501674480911,
|
| 223 |
+
"grad_norm": 41.3300960666502,
|
| 224 |
+
"learning_rate": 4.945108881776135e-07,
|
| 225 |
+
"logits/chosen": -2.3779914379119873,
|
| 226 |
+
"logits/rejected": -2.375589370727539,
|
| 227 |
+
"logps/chosen": -0.41506296396255493,
|
| 228 |
+
"logps/rejected": -0.435122013092041,
|
| 229 |
+
"loss": 0.8147,
|
| 230 |
+
"rewards/accuracies": 0.543749988079071,
|
| 231 |
+
"rewards/chosen": -1.0376574993133545,
|
| 232 |
+
"rewards/margins": 0.050147540867328644,
|
| 233 |
+
"rewards/rejected": -1.087804913520813,
|
| 234 |
+
"step": 75
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.17146684527796383,
|
| 238 |
+
"grad_norm": 19.713931611802217,
|
| 239 |
+
"learning_rate": 4.923863518447879e-07,
|
| 240 |
+
"logits/chosen": -2.3455538749694824,
|
| 241 |
+
"logits/rejected": -2.3334898948669434,
|
| 242 |
+
"logps/chosen": -0.44175705313682556,
|
| 243 |
+
"logps/rejected": -0.4695035517215729,
|
| 244 |
+
"loss": 0.8223,
|
| 245 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 246 |
+
"rewards/chosen": -1.104392647743225,
|
| 247 |
+
"rewards/margins": 0.0693662017583847,
|
| 248 |
+
"rewards/rejected": -1.173758864402771,
|
| 249 |
+
"step": 80
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.18218352310783656,
|
| 253 |
+
"grad_norm": 20.212157179302782,
|
| 254 |
+
"learning_rate": 4.899211964056035e-07,
|
| 255 |
+
"logits/chosen": -2.296379566192627,
|
| 256 |
+
"logits/rejected": -2.317187786102295,
|
| 257 |
+
"logps/chosen": -0.41793909668922424,
|
| 258 |
+
"logps/rejected": -0.4339120388031006,
|
| 259 |
+
"loss": 0.8234,
|
| 260 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 261 |
+
"rewards/chosen": -1.0448477268218994,
|
| 262 |
+
"rewards/margins": 0.03993237763643265,
|
| 263 |
+
"rewards/rejected": -1.084780216217041,
|
| 264 |
+
"step": 85
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.19290020093770932,
|
| 268 |
+
"grad_norm": 19.946761983669234,
|
| 269 |
+
"learning_rate": 4.871188860775643e-07,
|
| 270 |
+
"logits/chosen": -2.217319965362549,
|
| 271 |
+
"logits/rejected": -2.2343027591705322,
|
| 272 |
+
"logps/chosen": -0.4441414475440979,
|
| 273 |
+
"logps/rejected": -0.463876336812973,
|
| 274 |
+
"loss": 0.8217,
|
| 275 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 276 |
+
"rewards/chosen": -1.110353708267212,
|
| 277 |
+
"rewards/margins": 0.049337320029735565,
|
| 278 |
+
"rewards/rejected": -1.1596909761428833,
|
| 279 |
+
"step": 90
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.20361687876758205,
|
| 283 |
+
"grad_norm": 22.43054077775309,
|
| 284 |
+
"learning_rate": 4.839833588729897e-07,
|
| 285 |
+
"logits/chosen": -2.3150296211242676,
|
| 286 |
+
"logits/rejected": -2.305851936340332,
|
| 287 |
+
"logps/chosen": -0.4246097207069397,
|
| 288 |
+
"logps/rejected": -0.4527285695075989,
|
| 289 |
+
"loss": 0.8119,
|
| 290 |
+
"rewards/accuracies": 0.512499988079071,
|
| 291 |
+
"rewards/chosen": -1.0615241527557373,
|
| 292 |
+
"rewards/margins": 0.07029719650745392,
|
| 293 |
+
"rewards/rejected": -1.1318213939666748,
|
| 294 |
+
"step": 95
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.2143335565974548,
|
| 298 |
+
"grad_norm": 22.6806216251904,
|
| 299 |
+
"learning_rate": 4.80519021065031e-07,
|
| 300 |
+
"logits/chosen": -2.2826828956604004,
|
| 301 |
+
"logits/rejected": -2.275033473968506,
|
| 302 |
+
"logps/chosen": -0.41437339782714844,
|
| 303 |
+
"logps/rejected": -0.4373084604740143,
|
| 304 |
+
"loss": 0.8065,
|
| 305 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 306 |
+
"rewards/chosen": -1.0359336137771606,
|
| 307 |
+
"rewards/margins": 0.05733749270439148,
|
| 308 |
+
"rewards/rejected": -1.093271017074585,
|
| 309 |
+
"step": 100
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.22505023442732752,
|
| 313 |
+
"grad_norm": 19.45194393140105,
|
| 314 |
+
"learning_rate": 4.767307409956528e-07,
|
| 315 |
+
"logits/chosen": -2.2803359031677246,
|
| 316 |
+
"logits/rejected": -2.2837014198303223,
|
| 317 |
+
"logps/chosen": -0.4491514563560486,
|
| 318 |
+
"logps/rejected": -0.4561590254306793,
|
| 319 |
+
"loss": 0.8252,
|
| 320 |
+
"rewards/accuracies": 0.48124998807907104,
|
| 321 |
+
"rewards/chosen": -1.1228786706924438,
|
| 322 |
+
"rewards/margins": 0.01751893199980259,
|
| 323 |
+
"rewards/rejected": -1.1403974294662476,
|
| 324 |
+
"step": 105
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.23576691225720026,
|
| 328 |
+
"grad_norm": 21.617685020101888,
|
| 329 |
+
"learning_rate": 4.726238422342805e-07,
|
| 330 |
+
"logits/chosen": -2.263587474822998,
|
| 331 |
+
"logits/rejected": -2.231074094772339,
|
| 332 |
+
"logps/chosen": -0.456240713596344,
|
| 333 |
+
"logps/rejected": -0.47300928831100464,
|
| 334 |
+
"loss": 0.8075,
|
| 335 |
+
"rewards/accuracies": 0.543749988079071,
|
| 336 |
+
"rewards/chosen": -1.1406018733978271,
|
| 337 |
+
"rewards/margins": 0.04192143306136131,
|
| 338 |
+
"rewards/rejected": -1.1825231313705444,
|
| 339 |
+
"step": 110
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.24648359008707302,
|
| 343 |
+
"grad_norm": 24.689073829083224,
|
| 344 |
+
"learning_rate": 4.6820409609673027e-07,
|
| 345 |
+
"logits/chosen": -2.3198399543762207,
|
| 346 |
+
"logits/rejected": -2.3183414936065674,
|
| 347 |
+
"logps/chosen": -0.4174441397190094,
|
| 348 |
+
"logps/rejected": -0.4455544054508209,
|
| 349 |
+
"loss": 0.8138,
|
| 350 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 351 |
+
"rewards/chosen": -1.0436103343963623,
|
| 352 |
+
"rewards/margins": 0.07027561962604523,
|
| 353 |
+
"rewards/rejected": -1.1138859987258911,
|
| 354 |
+
"step": 115
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.2572002679169457,
|
| 358 |
+
"grad_norm": 21.210704415851712,
|
| 359 |
+
"learning_rate": 4.6347771353493173e-07,
|
| 360 |
+
"logits/chosen": -2.3105266094207764,
|
| 361 |
+
"logits/rejected": -2.3077518939971924,
|
| 362 |
+
"logps/chosen": -0.4340137839317322,
|
| 363 |
+
"logps/rejected": -0.4530583322048187,
|
| 364 |
+
"loss": 0.8106,
|
| 365 |
+
"rewards/accuracies": 0.5,
|
| 366 |
+
"rewards/chosen": -1.0850344896316528,
|
| 367 |
+
"rewards/margins": 0.047611434012651443,
|
| 368 |
+
"rewards/rejected": -1.132645845413208,
|
| 369 |
+
"step": 120
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.2679169457468185,
|
| 373 |
+
"grad_norm": 28.92966643864981,
|
| 374 |
+
"learning_rate": 4.5845133640884217e-07,
|
| 375 |
+
"logits/chosen": -2.2830965518951416,
|
| 376 |
+
"logits/rejected": -2.3132407665252686,
|
| 377 |
+
"logps/chosen": -0.45075589418411255,
|
| 378 |
+
"logps/rejected": -0.5100811719894409,
|
| 379 |
+
"loss": 0.7989,
|
| 380 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 381 |
+
"rewards/chosen": -1.1268895864486694,
|
| 382 |
+
"rewards/margins": 0.14831334352493286,
|
| 383 |
+
"rewards/rejected": -1.275202989578247,
|
| 384 |
+
"step": 125
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.27863362357669125,
|
| 388 |
+
"grad_norm": 21.983830009995575,
|
| 389 |
+
"learning_rate": 4.531320281528169e-07,
|
| 390 |
+
"logits/chosen": -2.354926109313965,
|
| 391 |
+
"logits/rejected": -2.346691131591797,
|
| 392 |
+
"logps/chosen": -0.4212542474269867,
|
| 393 |
+
"logps/rejected": -0.46237748861312866,
|
| 394 |
+
"loss": 0.8063,
|
| 395 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 396 |
+
"rewards/chosen": -1.053135633468628,
|
| 397 |
+
"rewards/margins": 0.10280808061361313,
|
| 398 |
+
"rewards/rejected": -1.1559436321258545,
|
| 399 |
+
"step": 130
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.289350301406564,
|
| 403 |
+
"grad_norm": 22.96804359323175,
|
| 404 |
+
"learning_rate": 4.4752726384955293e-07,
|
| 405 |
+
"logits/chosen": -2.366454601287842,
|
| 406 |
+
"logits/rejected": -2.3833343982696533,
|
| 407 |
+
"logps/chosen": -0.4526381492614746,
|
| 408 |
+
"logps/rejected": -0.5151073336601257,
|
| 409 |
+
"loss": 0.7816,
|
| 410 |
+
"rewards/accuracies": 0.606249988079071,
|
| 411 |
+
"rewards/chosen": -1.1315953731536865,
|
| 412 |
+
"rewards/margins": 0.15617303550243378,
|
| 413 |
+
"rewards/rejected": -1.2877686023712158,
|
| 414 |
+
"step": 135
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.3000669792364367,
|
| 418 |
+
"grad_norm": 30.57012215821005,
|
| 419 |
+
"learning_rate": 4.4164491972555294e-07,
|
| 420 |
+
"logits/chosen": -2.4294421672821045,
|
| 421 |
+
"logits/rejected": -2.4487221240997314,
|
| 422 |
+
"logps/chosen": -0.44983214139938354,
|
| 423 |
+
"logps/rejected": -0.49694761633872986,
|
| 424 |
+
"loss": 0.8084,
|
| 425 |
+
"rewards/accuracies": 0.606249988079071,
|
| 426 |
+
"rewards/chosen": -1.1245803833007812,
|
| 427 |
+
"rewards/margins": 0.11778867244720459,
|
| 428 |
+
"rewards/rejected": -1.2423690557479858,
|
| 429 |
+
"step": 140
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.31078365706630945,
|
| 433 |
+
"grad_norm": 21.180411511344364,
|
| 434 |
+
"learning_rate": 4.3549326208287393e-07,
|
| 435 |
+
"logits/chosen": -2.530730962753296,
|
| 436 |
+
"logits/rejected": -2.5269761085510254,
|
| 437 |
+
"logps/chosen": -0.42974090576171875,
|
| 438 |
+
"logps/rejected": -0.4626144766807556,
|
| 439 |
+
"loss": 0.7883,
|
| 440 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 441 |
+
"rewards/chosen": -1.0743523836135864,
|
| 442 |
+
"rewards/margins": 0.08218394219875336,
|
| 443 |
+
"rewards/rejected": -1.1565362215042114,
|
| 444 |
+
"step": 145
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.3215003348961822,
|
| 448 |
+
"grad_norm": 29.400301381227408,
|
| 449 |
+
"learning_rate": 4.2908093568271187e-07,
|
| 450 |
+
"logits/chosen": -2.5069005489349365,
|
| 451 |
+
"logits/rejected": -2.4892706871032715,
|
| 452 |
+
"logps/chosen": -0.4535086750984192,
|
| 453 |
+
"logps/rejected": -0.4898821711540222,
|
| 454 |
+
"loss": 0.797,
|
| 455 |
+
"rewards/accuracies": 0.606249988079071,
|
| 456 |
+
"rewards/chosen": -1.1337716579437256,
|
| 457 |
+
"rewards/margins": 0.09093374013900757,
|
| 458 |
+
"rewards/rejected": -1.224705457687378,
|
| 459 |
+
"step": 150
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.3322170127260549,
|
| 463 |
+
"grad_norm": 28.140109365748334,
|
| 464 |
+
"learning_rate": 4.2241695159714843e-07,
|
| 465 |
+
"logits/chosen": -2.483699321746826,
|
| 466 |
+
"logits/rejected": -2.4972500801086426,
|
| 467 |
+
"logps/chosen": -0.4816765785217285,
|
| 468 |
+
"logps/rejected": -0.5146931409835815,
|
| 469 |
+
"loss": 0.8022,
|
| 470 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 471 |
+
"rewards/chosen": -1.2041914463043213,
|
| 472 |
+
"rewards/margins": 0.08254122734069824,
|
| 473 |
+
"rewards/rejected": -1.2867326736450195,
|
| 474 |
+
"step": 155
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.34293369055592765,
|
| 478 |
+
"grad_norm": 25.486526483358187,
|
| 479 |
+
"learning_rate": 4.1551067454613074e-07,
|
| 480 |
+
"logits/chosen": -2.505037784576416,
|
| 481 |
+
"logits/rejected": -2.5097804069519043,
|
| 482 |
+
"logps/chosen": -0.448524534702301,
|
| 483 |
+
"logps/rejected": -0.46093541383743286,
|
| 484 |
+
"loss": 0.8047,
|
| 485 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 486 |
+
"rewards/chosen": -1.1213113069534302,
|
| 487 |
+
"rewards/margins": 0.031027212738990784,
|
| 488 |
+
"rewards/rejected": -1.1523386240005493,
|
| 489 |
+
"step": 160
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.3536503683858004,
|
| 493 |
+
"grad_norm": 18.985578073968945,
|
| 494 |
+
"learning_rate": 4.0837180973747876e-07,
|
| 495 |
+
"logits/chosen": -2.515350103378296,
|
| 496 |
+
"logits/rejected": -2.528233528137207,
|
| 497 |
+
"logps/chosen": -0.43952083587646484,
|
| 498 |
+
"logps/rejected": -0.4735996127128601,
|
| 499 |
+
"loss": 0.7917,
|
| 500 |
+
"rewards/accuracies": 0.606249988079071,
|
| 501 |
+
"rewards/chosen": -1.0988019704818726,
|
| 502 |
+
"rewards/margins": 0.08519701659679413,
|
| 503 |
+
"rewards/rejected": -1.1839991807937622,
|
| 504 |
+
"step": 165
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 0.3643670462156731,
|
| 508 |
+
"grad_norm": 25.120865745392436,
|
| 509 |
+
"learning_rate": 4.0101038922841455e-07,
|
| 510 |
+
"logits/chosen": -2.576842784881592,
|
| 511 |
+
"logits/rejected": -2.5906457901000977,
|
| 512 |
+
"logps/chosen": -0.4653005599975586,
|
| 513 |
+
"logps/rejected": -0.4848781228065491,
|
| 514 |
+
"loss": 0.8051,
|
| 515 |
+
"rewards/accuracies": 0.53125,
|
| 516 |
+
"rewards/chosen": -1.163251280784607,
|
| 517 |
+
"rewards/margins": 0.0489438995718956,
|
| 518 |
+
"rewards/rejected": -1.2121951580047607,
|
| 519 |
+
"step": 170
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.3750837240455459,
|
| 523 |
+
"grad_norm": 22.847481742575464,
|
| 524 |
+
"learning_rate": 3.934367578277783e-07,
|
| 525 |
+
"logits/chosen": -2.530890941619873,
|
| 526 |
+
"logits/rejected": -2.531864643096924,
|
| 527 |
+
"logps/chosen": -0.45446857810020447,
|
| 528 |
+
"logps/rejected": -0.5132210850715637,
|
| 529 |
+
"loss": 0.7911,
|
| 530 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 531 |
+
"rewards/chosen": -1.1361713409423828,
|
| 532 |
+
"rewards/margins": 0.14688125252723694,
|
| 533 |
+
"rewards/rejected": -1.283052682876587,
|
| 534 |
+
"step": 175
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.38580040187541864,
|
| 538 |
+
"grad_norm": 26.0995283209312,
|
| 539 |
+
"learning_rate": 3.856615585587434e-07,
|
| 540 |
+
"logits/chosen": -2.599412441253662,
|
| 541 |
+
"logits/rejected": -2.6105477809906006,
|
| 542 |
+
"logps/chosen": -0.4383458197116852,
|
| 543 |
+
"logps/rejected": -0.4895172715187073,
|
| 544 |
+
"loss": 0.7885,
|
| 545 |
+
"rewards/accuracies": 0.675000011920929,
|
| 546 |
+
"rewards/chosen": -1.0958645343780518,
|
| 547 |
+
"rewards/margins": 0.12792867422103882,
|
| 548 |
+
"rewards/rejected": -1.2237932682037354,
|
| 549 |
+
"step": 180
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.3965170797052914,
|
| 553 |
+
"grad_norm": 29.1301293940354,
|
| 554 |
+
"learning_rate": 3.776957177024581e-07,
|
| 555 |
+
"logits/chosen": -2.597064256668091,
|
| 556 |
+
"logits/rejected": -2.584998607635498,
|
| 557 |
+
"logps/chosen": -0.46641644835472107,
|
| 558 |
+
"logps/rejected": -0.5318672060966492,
|
| 559 |
+
"loss": 0.7907,
|
| 560 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 561 |
+
"rewards/chosen": -1.1660410165786743,
|
| 562 |
+
"rewards/margins": 0.16362698376178741,
|
| 563 |
+
"rewards/rejected": -1.3296680450439453,
|
| 564 |
+
"step": 185
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.4072337575351641,
|
| 568 |
+
"grad_norm": 26.246700781484808,
|
| 569 |
+
"learning_rate": 3.69550429443633e-07,
|
| 570 |
+
"logits/chosen": -2.6202404499053955,
|
| 571 |
+
"logits/rejected": -2.6158204078674316,
|
| 572 |
+
"logps/chosen": -0.4914141595363617,
|
| 573 |
+
"logps/rejected": -0.4930477738380432,
|
| 574 |
+
"loss": 0.7991,
|
| 575 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 576 |
+
"rewards/chosen": -1.2285354137420654,
|
| 577 |
+
"rewards/margins": 0.004083993844687939,
|
| 578 |
+
"rewards/rejected": -1.2326194047927856,
|
| 579 |
+
"step": 190
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.41795043536503684,
|
| 583 |
+
"grad_norm": 37.655825101171395,
|
| 584 |
+
"learning_rate": 3.612371401396502e-07,
|
| 585 |
+
"logits/chosen": -2.7039732933044434,
|
| 586 |
+
"logits/rejected": -2.6810288429260254,
|
| 587 |
+
"logps/chosen": -0.47953447699546814,
|
| 588 |
+
"logps/rejected": -0.5316765904426575,
|
| 589 |
+
"loss": 0.7925,
|
| 590 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 591 |
+
"rewards/chosen": -1.198836088180542,
|
| 592 |
+
"rewards/margins": 0.13035525381565094,
|
| 593 |
+
"rewards/rejected": -1.3291913270950317,
|
| 594 |
+
"step": 195
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 0.4286671131949096,
|
| 598 |
+
"grad_norm": 22.671830849613123,
|
| 599 |
+
"learning_rate": 3.527675322353011e-07,
|
| 600 |
+
"logits/chosen": -2.711343288421631,
|
| 601 |
+
"logits/rejected": -2.724891185760498,
|
| 602 |
+
"logps/chosen": -0.4652802348136902,
|
| 603 |
+
"logps/rejected": -0.52509605884552,
|
| 604 |
+
"loss": 0.7948,
|
| 605 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 606 |
+
"rewards/chosen": -1.1632006168365479,
|
| 607 |
+
"rewards/margins": 0.14953944087028503,
|
| 608 |
+
"rewards/rejected": -1.3127400875091553,
|
| 609 |
+
"step": 200
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.4393837910247823,
|
| 613 |
+
"grad_norm": 30.58821587730182,
|
| 614 |
+
"learning_rate": 3.4415350784575693e-07,
|
| 615 |
+
"logits/chosen": -2.7035841941833496,
|
| 616 |
+
"logits/rejected": -2.694777250289917,
|
| 617 |
+
"logps/chosen": -0.5165461897850037,
|
| 618 |
+
"logps/rejected": -0.573054850101471,
|
| 619 |
+
"loss": 0.7735,
|
| 620 |
+
"rewards/accuracies": 0.581250011920929,
|
| 621 |
+
"rewards/chosen": -1.291365623474121,
|
| 622 |
+
"rewards/margins": 0.1412714421749115,
|
| 623 |
+
"rewards/rejected": -1.432637095451355,
|
| 624 |
+
"step": 205
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 0.45010046885465504,
|
| 628 |
+
"grad_norm": 31.58801707836199,
|
| 629 |
+
"learning_rate": 3.3540717203084224e-07,
|
| 630 |
+
"logits/chosen": -2.7756335735321045,
|
| 631 |
+
"logits/rejected": -2.751556873321533,
|
| 632 |
+
"logps/chosen": -0.5132556557655334,
|
| 633 |
+
"logps/rejected": -0.5941559076309204,
|
| 634 |
+
"loss": 0.786,
|
| 635 |
+
"rewards/accuracies": 0.637499988079071,
|
| 636 |
+
"rewards/chosen": -1.2831389904022217,
|
| 637 |
+
"rewards/margins": 0.20225080847740173,
|
| 638 |
+
"rewards/rejected": -1.4853899478912354,
|
| 639 |
+
"step": 210
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.4608171466845278,
|
| 643 |
+
"grad_norm": 41.456925783555256,
|
| 644 |
+
"learning_rate": 3.265408157841153e-07,
|
| 645 |
+
"logits/chosen": -2.7932379245758057,
|
| 646 |
+
"logits/rejected": -2.7999000549316406,
|
| 647 |
+
"logps/chosen": -0.47902554273605347,
|
| 648 |
+
"logps/rejected": -0.5101593732833862,
|
| 649 |
+
"loss": 0.8028,
|
| 650 |
+
"rewards/accuracies": 0.5625,
|
| 651 |
+
"rewards/chosen": -1.197563886642456,
|
| 652 |
+
"rewards/margins": 0.07783453166484833,
|
| 653 |
+
"rewards/rejected": -1.2753984928131104,
|
| 654 |
+
"step": 215
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.4715338245144005,
|
| 658 |
+
"grad_norm": 31.036341061185052,
|
| 659 |
+
"learning_rate": 3.1756689876066187e-07,
|
| 660 |
+
"logits/chosen": -2.755951404571533,
|
| 661 |
+
"logits/rejected": -2.772839069366455,
|
| 662 |
+
"logps/chosen": -0.5013461112976074,
|
| 663 |
+
"logps/rejected": -0.5391478538513184,
|
| 664 |
+
"loss": 0.7933,
|
| 665 |
+
"rewards/accuracies": 0.5625,
|
| 666 |
+
"rewards/chosen": -1.2533652782440186,
|
| 667 |
+
"rewards/margins": 0.09450428932905197,
|
| 668 |
+
"rewards/rejected": -1.3478695154190063,
|
| 669 |
+
"step": 220
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.4822505023442733,
|
| 673 |
+
"grad_norm": 26.21868089092331,
|
| 674 |
+
"learning_rate": 3.0849803176787184e-07,
|
| 675 |
+
"logits/chosen": -2.7325472831726074,
|
| 676 |
+
"logits/rejected": -2.7435898780822754,
|
| 677 |
+
"logps/chosen": -0.5336536765098572,
|
| 678 |
+
"logps/rejected": -0.5821245908737183,
|
| 679 |
+
"loss": 0.7835,
|
| 680 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 681 |
+
"rewards/chosen": -1.3341342210769653,
|
| 682 |
+
"rewards/margins": 0.12117739021778107,
|
| 683 |
+
"rewards/rejected": -1.4553115367889404,
|
| 684 |
+
"step": 225
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 0.49296718017414604,
|
| 688 |
+
"grad_norm": 32.03132066446478,
|
| 689 |
+
"learning_rate": 2.993469590438065e-07,
|
| 690 |
+
"logits/chosen": -2.6822681427001953,
|
| 691 |
+
"logits/rejected": -2.7262749671936035,
|
| 692 |
+
"logps/chosen": -0.5299819111824036,
|
| 693 |
+
"logps/rejected": -0.5460516810417175,
|
| 694 |
+
"loss": 0.8012,
|
| 695 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 696 |
+
"rewards/chosen": -1.3249547481536865,
|
| 697 |
+
"rewards/margins": 0.04017447307705879,
|
| 698 |
+
"rewards/rejected": -1.3651292324066162,
|
| 699 |
+
"step": 230
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.5036838580040187,
|
| 703 |
+
"grad_norm": 29.43042758117792,
|
| 704 |
+
"learning_rate": 2.901265403480597e-07,
|
| 705 |
+
"logits/chosen": -2.7072153091430664,
|
| 706 |
+
"logits/rejected": -2.7308104038238525,
|
| 707 |
+
"logps/chosen": -0.5455118417739868,
|
| 708 |
+
"logps/rejected": -0.5925866961479187,
|
| 709 |
+
"loss": 0.7764,
|
| 710 |
+
"rewards/accuracies": 0.5625,
|
| 711 |
+
"rewards/chosen": -1.3637795448303223,
|
| 712 |
+
"rewards/margins": 0.11768718808889389,
|
| 713 |
+
"rewards/rejected": -1.4814666509628296,
|
| 714 |
+
"step": 235
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 0.5144005358338914,
|
| 718 |
+
"grad_norm": 29.703263320129235,
|
| 719 |
+
"learning_rate": 2.808497328902782e-07,
|
| 720 |
+
"logits/chosen": -2.6819677352905273,
|
| 721 |
+
"logits/rejected": -2.6889100074768066,
|
| 722 |
+
"logps/chosen": -0.5209198594093323,
|
| 723 |
+
"logps/rejected": -0.5574553608894348,
|
| 724 |
+
"loss": 0.8044,
|
| 725 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 726 |
+
"rewards/chosen": -1.3022996187210083,
|
| 727 |
+
"rewards/margins": 0.09133879840373993,
|
| 728 |
+
"rewards/rejected": -1.3936383724212646,
|
| 729 |
+
"step": 240
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.5251172136637642,
|
| 733 |
+
"grad_norm": 37.052180806405566,
|
| 734 |
+
"learning_rate": 2.715295731217396e-07,
|
| 735 |
+
"logits/chosen": -2.7002854347229004,
|
| 736 |
+
"logits/rejected": -2.70177960395813,
|
| 737 |
+
"logps/chosen": -0.5708497762680054,
|
| 738 |
+
"logps/rejected": -0.5927925109863281,
|
| 739 |
+
"loss": 0.8003,
|
| 740 |
+
"rewards/accuracies": 0.5,
|
| 741 |
+
"rewards/chosen": -1.4271246194839478,
|
| 742 |
+
"rewards/margins": 0.05485675856471062,
|
| 743 |
+
"rewards/rejected": -1.4819812774658203,
|
| 744 |
+
"step": 245
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.535833891493637,
|
| 748 |
+
"grad_norm": 32.762361065320945,
|
| 749 |
+
"learning_rate": 2.621791584155725e-07,
|
| 750 |
+
"logits/chosen": -2.6251308917999268,
|
| 751 |
+
"logits/rejected": -2.614744186401367,
|
| 752 |
+
"logps/chosen": -0.5160342454910278,
|
| 753 |
+
"logps/rejected": -0.6052494645118713,
|
| 754 |
+
"loss": 0.7709,
|
| 755 |
+
"rewards/accuracies": 0.625,
|
| 756 |
+
"rewards/chosen": -1.2900855541229248,
|
| 757 |
+
"rewards/margins": 0.2230379283428192,
|
| 758 |
+
"rewards/rejected": -1.5131235122680664,
|
| 759 |
+
"step": 250
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.5465505693235098,
|
| 763 |
+
"grad_norm": 31.109425612718024,
|
| 764 |
+
"learning_rate": 2.52811628661367e-07,
|
| 765 |
+
"logits/chosen": -2.7665226459503174,
|
| 766 |
+
"logits/rejected": -2.7633566856384277,
|
| 767 |
+
"logps/chosen": -0.5172406435012817,
|
| 768 |
+
"logps/rejected": -0.55857914686203,
|
| 769 |
+
"loss": 0.7997,
|
| 770 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 771 |
+
"rewards/chosen": -1.2931016683578491,
|
| 772 |
+
"rewards/margins": 0.10334619134664536,
|
| 773 |
+
"rewards/rejected": -1.3964478969573975,
|
| 774 |
+
"step": 255
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 0.5572672471533825,
|
| 778 |
+
"grad_norm": 34.78606161386547,
|
| 779 |
+
"learning_rate": 2.434401478000353e-07,
|
| 780 |
+
"logits/chosen": -2.760376214981079,
|
| 781 |
+
"logits/rejected": -2.7761454582214355,
|
| 782 |
+
"logps/chosen": -0.5151620507240295,
|
| 783 |
+
"logps/rejected": -0.5613347887992859,
|
| 784 |
+
"loss": 0.778,
|
| 785 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 786 |
+
"rewards/chosen": -1.287904977798462,
|
| 787 |
+
"rewards/margins": 0.11543188244104385,
|
| 788 |
+
"rewards/rejected": -1.4033368825912476,
|
| 789 |
+
"step": 260
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.5679839249832552,
|
| 793 |
+
"grad_norm": 26.46498554148562,
|
| 794 |
+
"learning_rate": 2.340778853248759e-07,
|
| 795 |
+
"logits/chosen": -2.7599408626556396,
|
| 796 |
+
"logits/rejected": -2.7651569843292236,
|
| 797 |
+
"logps/chosen": -0.5734432935714722,
|
| 798 |
+
"logps/rejected": -0.6467745900154114,
|
| 799 |
+
"loss": 0.7688,
|
| 800 |
+
"rewards/accuracies": 0.59375,
|
| 801 |
+
"rewards/chosen": -1.4336082935333252,
|
| 802 |
+
"rewards/margins": 0.18332824110984802,
|
| 803 |
+
"rewards/rejected": -1.616936445236206,
|
| 804 |
+
"step": 265
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 0.578700602813128,
|
| 808 |
+
"grad_norm": 34.04025832229061,
|
| 809 |
+
"learning_rate": 2.247379977748339e-07,
|
| 810 |
+
"logits/chosen": -2.7183938026428223,
|
| 811 |
+
"logits/rejected": -2.75644588470459,
|
| 812 |
+
"logps/chosen": -0.5665550231933594,
|
| 813 |
+
"logps/rejected": -0.6176514625549316,
|
| 814 |
+
"loss": 0.7952,
|
| 815 |
+
"rewards/accuracies": 0.518750011920929,
|
| 816 |
+
"rewards/chosen": -1.4163875579833984,
|
| 817 |
+
"rewards/margins": 0.12774114310741425,
|
| 818 |
+
"rewards/rejected": -1.544128656387329,
|
| 819 |
+
"step": 270
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 0.5894172806430007,
|
| 823 |
+
"grad_norm": 31.258853663890047,
|
| 824 |
+
"learning_rate": 2.1543361024596573e-07,
|
| 825 |
+
"logits/chosen": -2.7444400787353516,
|
| 826 |
+
"logits/rejected": -2.745250940322876,
|
| 827 |
+
"logps/chosen": -0.5861095190048218,
|
| 828 |
+
"logps/rejected": -0.6748340725898743,
|
| 829 |
+
"loss": 0.758,
|
| 830 |
+
"rewards/accuracies": 0.59375,
|
| 831 |
+
"rewards/chosen": -1.4652737379074097,
|
| 832 |
+
"rewards/margins": 0.22181133925914764,
|
| 833 |
+
"rewards/rejected": -1.6870851516723633,
|
| 834 |
+
"step": 275
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 0.6001339584728734,
|
| 838 |
+
"grad_norm": 46.09656975376248,
|
| 839 |
+
"learning_rate": 2.0617779794709082e-07,
|
| 840 |
+
"logits/chosen": -2.724043607711792,
|
| 841 |
+
"logits/rejected": -2.74163818359375,
|
| 842 |
+
"logps/chosen": -0.5856515169143677,
|
| 843 |
+
"logps/rejected": -0.643699049949646,
|
| 844 |
+
"loss": 0.7812,
|
| 845 |
+
"rewards/accuracies": 0.606249988079071,
|
| 846 |
+
"rewards/chosen": -1.464128851890564,
|
| 847 |
+
"rewards/margins": 0.14511898159980774,
|
| 848 |
+
"rewards/rejected": -1.6092478036880493,
|
| 849 |
+
"step": 280
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.6108506363027462,
|
| 853 |
+
"grad_norm": 43.94493014255228,
|
| 854 |
+
"learning_rate": 1.9698356782554687e-07,
|
| 855 |
+
"logits/chosen": -2.6912693977355957,
|
| 856 |
+
"logits/rejected": -2.707552194595337,
|
| 857 |
+
"logps/chosen": -0.5880975723266602,
|
| 858 |
+
"logps/rejected": -0.6368181705474854,
|
| 859 |
+
"loss": 0.7642,
|
| 860 |
+
"rewards/accuracies": 0.59375,
|
| 861 |
+
"rewards/chosen": -1.4702438116073608,
|
| 862 |
+
"rewards/margins": 0.12180168926715851,
|
| 863 |
+
"rewards/rejected": -1.592045545578003,
|
| 864 |
+
"step": 285
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.6215673141326189,
|
| 868 |
+
"grad_norm": 49.663136616239086,
|
| 869 |
+
"learning_rate": 1.878638402888727e-07,
|
| 870 |
+
"logits/chosen": -2.7193799018859863,
|
| 871 |
+
"logits/rejected": -2.722045421600342,
|
| 872 |
+
"logps/chosen": -0.6236746311187744,
|
| 873 |
+
"logps/rejected": -0.6843193173408508,
|
| 874 |
+
"loss": 0.785,
|
| 875 |
+
"rewards/accuracies": 0.574999988079071,
|
| 876 |
+
"rewards/chosen": -1.559186577796936,
|
| 877 |
+
"rewards/margins": 0.15161168575286865,
|
| 878 |
+
"rewards/rejected": -1.7107985019683838,
|
| 879 |
+
"step": 290
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 0.6322839919624916,
|
| 883 |
+
"grad_norm": 34.23744852441083,
|
| 884 |
+
"learning_rate": 1.7883143104810223e-07,
|
| 885 |
+
"logits/chosen": -2.6754257678985596,
|
| 886 |
+
"logits/rejected": -2.673750877380371,
|
| 887 |
+
"logps/chosen": -0.6068626046180725,
|
| 888 |
+
"logps/rejected": -0.7437673807144165,
|
| 889 |
+
"loss": 0.754,
|
| 890 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 891 |
+
"rewards/chosen": -1.5171564817428589,
|
| 892 |
+
"rewards/margins": 0.3422619104385376,
|
| 893 |
+
"rewards/rejected": -1.859418511390686,
|
| 894 |
+
"step": 295
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 0.6430006697923644,
|
| 898 |
+
"grad_norm": 33.8142047010383,
|
| 899 |
+
"learning_rate": 1.6989903310818548e-07,
|
| 900 |
+
"logits/chosen": -2.714789867401123,
|
| 901 |
+
"logits/rejected": -2.710334062576294,
|
| 902 |
+
"logps/chosen": -0.6445562243461609,
|
| 903 |
+
"logps/rejected": -0.7172508239746094,
|
| 904 |
+
"loss": 0.7768,
|
| 905 |
+
"rewards/accuracies": 0.581250011920929,
|
| 906 |
+
"rewards/chosen": -1.611390471458435,
|
| 907 |
+
"rewards/margins": 0.1817365437746048,
|
| 908 |
+
"rewards/rejected": -1.7931270599365234,
|
| 909 |
+
"step": 300
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 0.6537173476222371,
|
| 913 |
+
"grad_norm": 89.8750741255919,
|
| 914 |
+
"learning_rate": 1.610791989308466e-07,
|
| 915 |
+
"logits/chosen": -2.693514347076416,
|
| 916 |
+
"logits/rejected": -2.6736016273498535,
|
| 917 |
+
"logps/chosen": -0.6287732720375061,
|
| 918 |
+
"logps/rejected": -0.701014518737793,
|
| 919 |
+
"loss": 0.7907,
|
| 920 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 921 |
+
"rewards/chosen": -1.5719332695007324,
|
| 922 |
+
"rewards/margins": 0.1806029975414276,
|
| 923 |
+
"rewards/rejected": -1.7525361776351929,
|
| 924 |
+
"step": 305
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 0.6644340254521098,
|
| 928 |
+
"grad_norm": 49.27896755369507,
|
| 929 |
+
"learning_rate": 1.523843227949418e-07,
|
| 930 |
+
"logits/chosen": -2.791170597076416,
|
| 931 |
+
"logits/rejected": -2.7779464721679688,
|
| 932 |
+
"logps/chosen": -0.6191990971565247,
|
| 933 |
+
"logps/rejected": -0.690567672252655,
|
| 934 |
+
"loss": 0.7726,
|
| 935 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 936 |
+
"rewards/chosen": -1.5479977130889893,
|
| 937 |
+
"rewards/margins": 0.17842140793800354,
|
| 938 |
+
"rewards/rejected": -1.72641921043396,
|
| 939 |
+
"step": 310
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 0.6751507032819826,
|
| 943 |
+
"grad_norm": 52.532294308547925,
|
| 944 |
+
"learning_rate": 1.4382662337910867e-07,
|
| 945 |
+
"logits/chosen": -2.7822985649108887,
|
| 946 |
+
"logits/rejected": -2.8041725158691406,
|
| 947 |
+
"logps/chosen": -0.546491801738739,
|
| 948 |
+
"logps/rejected": -0.6088187098503113,
|
| 949 |
+
"loss": 0.7633,
|
| 950 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 951 |
+
"rewards/chosen": -1.3662294149398804,
|
| 952 |
+
"rewards/margins": 0.15581734478473663,
|
| 953 |
+
"rewards/rejected": -1.5220468044281006,
|
| 954 |
+
"step": 315
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.6858673811118553,
|
| 958 |
+
"grad_norm": 31.304061633836678,
|
| 959 |
+
"learning_rate": 1.3541812659118104e-07,
|
| 960 |
+
"logits/chosen": -2.784820556640625,
|
| 961 |
+
"logits/rejected": -2.790811777114868,
|
| 962 |
+
"logps/chosen": -0.6415971517562866,
|
| 963 |
+
"logps/rejected": -0.6807955503463745,
|
| 964 |
+
"loss": 0.7832,
|
| 965 |
+
"rewards/accuracies": 0.581250011920929,
|
| 966 |
+
"rewards/chosen": -1.6039928197860718,
|
| 967 |
+
"rewards/margins": 0.09799607843160629,
|
| 968 |
+
"rewards/rejected": -1.701988935470581,
|
| 969 |
+
"step": 320
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.696584058941728,
|
| 973 |
+
"grad_norm": 42.827372642417416,
|
| 974 |
+
"learning_rate": 1.2717064866850047e-07,
|
| 975 |
+
"logits/chosen": -2.8045849800109863,
|
| 976 |
+
"logits/rejected": -2.7954859733581543,
|
| 977 |
+
"logps/chosen": -0.6314974427223206,
|
| 978 |
+
"logps/rejected": -0.7085036635398865,
|
| 979 |
+
"loss": 0.7874,
|
| 980 |
+
"rewards/accuracies": 0.581250011920929,
|
| 981 |
+
"rewards/chosen": -1.5787436962127686,
|
| 982 |
+
"rewards/margins": 0.19251549243927002,
|
| 983 |
+
"rewards/rejected": -1.771259069442749,
|
| 984 |
+
"step": 325
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 0.7073007367716008,
|
| 988 |
+
"grad_norm": 35.26979827058928,
|
| 989 |
+
"learning_rate": 1.1909577957287117e-07,
|
| 990 |
+
"logits/chosen": -2.8660318851470947,
|
| 991 |
+
"logits/rejected": -2.870734691619873,
|
| 992 |
+
"logps/chosen": -0.5516513586044312,
|
| 993 |
+
"logps/rejected": -0.605838418006897,
|
| 994 |
+
"loss": 0.7555,
|
| 995 |
+
"rewards/accuracies": 0.625,
|
| 996 |
+
"rewards/chosen": -1.379128336906433,
|
| 997 |
+
"rewards/margins": 0.13546761870384216,
|
| 998 |
+
"rewards/rejected": -1.5145959854125977,
|
| 999 |
+
"step": 330
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 0.7180174146014735,
|
| 1003 |
+
"grad_norm": 33.88925580800264,
|
| 1004 |
+
"learning_rate": 1.1120486670349441e-07,
|
| 1005 |
+
"logits/chosen": -2.82328724861145,
|
| 1006 |
+
"logits/rejected": -2.847529649734497,
|
| 1007 |
+
"logps/chosen": -0.6149914860725403,
|
| 1008 |
+
"logps/rejected": -0.6911052465438843,
|
| 1009 |
+
"loss": 0.7548,
|
| 1010 |
+
"rewards/accuracies": 0.643750011920929,
|
| 1011 |
+
"rewards/chosen": -1.5374788045883179,
|
| 1012 |
+
"rewards/margins": 0.19028423726558685,
|
| 1013 |
+
"rewards/rejected": -1.7277631759643555,
|
| 1014 |
+
"step": 335
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 0.7287340924313462,
|
| 1018 |
+
"grad_norm": 31.71698090485735,
|
| 1019 |
+
"learning_rate": 1.0350899895076948e-07,
|
| 1020 |
+
"logits/chosen": -2.790008068084717,
|
| 1021 |
+
"logits/rejected": -2.7927775382995605,
|
| 1022 |
+
"logps/chosen": -0.6008199453353882,
|
| 1023 |
+
"logps/rejected": -0.6924527287483215,
|
| 1024 |
+
"loss": 0.7686,
|
| 1025 |
+
"rewards/accuracies": 0.612500011920929,
|
| 1026 |
+
"rewards/chosen": -1.5020498037338257,
|
| 1027 |
+
"rewards/margins": 0.2290818691253662,
|
| 1028 |
+
"rewards/rejected": -1.7311317920684814,
|
| 1029 |
+
"step": 340
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 0.739450770261219,
|
| 1033 |
+
"grad_norm": 36.716335914223706,
|
| 1034 |
+
"learning_rate": 9.601899111337069e-08,
|
| 1035 |
+
"logits/chosen": -2.885432481765747,
|
| 1036 |
+
"logits/rejected": -2.8697662353515625,
|
| 1037 |
+
"logps/chosen": -0.6102738976478577,
|
| 1038 |
+
"logps/rejected": -0.6805952191352844,
|
| 1039 |
+
"loss": 0.7819,
|
| 1040 |
+
"rewards/accuracies": 0.65625,
|
| 1041 |
+
"rewards/chosen": -1.5256847143173218,
|
| 1042 |
+
"rewards/margins": 0.17580333352088928,
|
| 1043 |
+
"rewards/rejected": -1.7014878988265991,
|
| 1044 |
+
"step": 345
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 0.7501674480910918,
|
| 1048 |
+
"grad_norm": 33.17344514912859,
|
| 1049 |
+
"learning_rate": 8.874536870049765e-08,
|
| 1050 |
+
"logits/chosen": -2.865382432937622,
|
| 1051 |
+
"logits/rejected": -2.88032865524292,
|
| 1052 |
+
"logps/chosen": -0.6345846056938171,
|
| 1053 |
+
"logps/rejected": -0.7329587936401367,
|
| 1054 |
+
"loss": 0.7599,
|
| 1055 |
+
"rewards/accuracies": 0.606249988079071,
|
| 1056 |
+
"rewards/chosen": -1.5864615440368652,
|
| 1057 |
+
"rewards/margins": 0.24593544006347656,
|
| 1058 |
+
"rewards/rejected": -1.8323971033096313,
|
| 1059 |
+
"step": 350
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.7608841259209645,
|
| 1063 |
+
"grad_norm": 35.10962301829223,
|
| 1064 |
+
"learning_rate": 8.169835314065663e-08,
|
| 1065 |
+
"logits/chosen": -2.787919521331787,
|
| 1066 |
+
"logits/rejected": -2.7910847663879395,
|
| 1067 |
+
"logps/chosen": -0.6197460889816284,
|
| 1068 |
+
"logps/rejected": -0.6998850107192993,
|
| 1069 |
+
"loss": 0.7762,
|
| 1070 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 1071 |
+
"rewards/chosen": -1.5493652820587158,
|
| 1072 |
+
"rewards/margins": 0.20034721493721008,
|
| 1073 |
+
"rewards/rejected": -1.749712586402893,
|
| 1074 |
+
"step": 355
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.7716008037508373,
|
| 1078 |
+
"grad_norm": 34.02497610424062,
|
| 1079 |
+
"learning_rate": 7.488784741775816e-08,
|
| 1080 |
+
"logits/chosen": -2.8549702167510986,
|
| 1081 |
+
"logits/rejected": -2.8631627559661865,
|
| 1082 |
+
"logps/chosen": -0.6374293565750122,
|
| 1083 |
+
"logps/rejected": -0.7279438972473145,
|
| 1084 |
+
"loss": 0.7685,
|
| 1085 |
+
"rewards/accuracies": 0.606249988079071,
|
| 1086 |
+
"rewards/chosen": -1.5935734510421753,
|
| 1087 |
+
"rewards/margins": 0.2262863665819168,
|
| 1088 |
+
"rewards/rejected": -1.8198598623275757,
|
| 1089 |
+
"step": 360
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 0.78231748158071,
|
| 1093 |
+
"grad_norm": 34.499708849490524,
|
| 1094 |
+
"learning_rate": 6.832342215471715e-08,
|
| 1095 |
+
"logits/chosen": -2.90082049369812,
|
| 1096 |
+
"logits/rejected": -2.8912737369537354,
|
| 1097 |
+
"logps/chosen": -0.6256529688835144,
|
| 1098 |
+
"logps/rejected": -0.6695230603218079,
|
| 1099 |
+
"loss": 0.765,
|
| 1100 |
+
"rewards/accuracies": 0.637499988079071,
|
| 1101 |
+
"rewards/chosen": -1.5641324520111084,
|
| 1102 |
+
"rewards/margins": 0.10967500507831573,
|
| 1103 |
+
"rewards/rejected": -1.6738075017929077,
|
| 1104 |
+
"step": 365
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 0.7930341594105828,
|
| 1108 |
+
"grad_norm": 40.533056688484706,
|
| 1109 |
+
"learning_rate": 6.201430216410985e-08,
|
| 1110 |
+
"logits/chosen": -2.840193510055542,
|
| 1111 |
+
"logits/rejected": -2.845524549484253,
|
| 1112 |
+
"logps/chosen": -0.638153076171875,
|
| 1113 |
+
"logps/rejected": -0.6873365044593811,
|
| 1114 |
+
"loss": 0.7559,
|
| 1115 |
+
"rewards/accuracies": 0.543749988079071,
|
| 1116 |
+
"rewards/chosen": -1.5953826904296875,
|
| 1117 |
+
"rewards/margins": 0.12295844405889511,
|
| 1118 |
+
"rewards/rejected": -1.7183412313461304,
|
| 1119 |
+
"step": 370
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 0.8037508372404555,
|
| 1123 |
+
"grad_norm": 61.33843154278605,
|
| 1124 |
+
"learning_rate": 5.596935348479029e-08,
|
| 1125 |
+
"logits/chosen": -2.8891007900238037,
|
| 1126 |
+
"logits/rejected": -2.9143340587615967,
|
| 1127 |
+
"logps/chosen": -0.6350052952766418,
|
| 1128 |
+
"logps/rejected": -0.680306613445282,
|
| 1129 |
+
"loss": 0.7693,
|
| 1130 |
+
"rewards/accuracies": 0.606249988079071,
|
| 1131 |
+
"rewards/chosen": -1.5875132083892822,
|
| 1132 |
+
"rewards/margins": 0.11325322091579437,
|
| 1133 |
+
"rewards/rejected": -1.7007665634155273,
|
| 1134 |
+
"step": 375
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 0.8144675150703282,
|
| 1138 |
+
"grad_norm": 50.258970661764174,
|
| 1139 |
+
"learning_rate": 5.01970709226808e-08,
|
| 1140 |
+
"logits/chosen": -2.8725762367248535,
|
| 1141 |
+
"logits/rejected": -2.8825366497039795,
|
| 1142 |
+
"logps/chosen": -0.6515634059906006,
|
| 1143 |
+
"logps/rejected": -0.7237032651901245,
|
| 1144 |
+
"loss": 0.7826,
|
| 1145 |
+
"rewards/accuracies": 0.606249988079071,
|
| 1146 |
+
"rewards/chosen": -1.6289085149765015,
|
| 1147 |
+
"rewards/margins": 0.1803499013185501,
|
| 1148 |
+
"rewards/rejected": -1.809258222579956,
|
| 1149 |
+
"step": 380
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 0.825184192900201,
|
| 1153 |
+
"grad_norm": 42.96591025859426,
|
| 1154 |
+
"learning_rate": 4.4705566113246635e-08,
|
| 1155 |
+
"logits/chosen": -2.857102870941162,
|
| 1156 |
+
"logits/rejected": -2.857473850250244,
|
| 1157 |
+
"logps/chosen": -0.641956090927124,
|
| 1158 |
+
"logps/rejected": -0.7416104078292847,
|
| 1159 |
+
"loss": 0.7604,
|
| 1160 |
+
"rewards/accuracies": 0.643750011920929,
|
| 1161 |
+
"rewards/chosen": -1.6048901081085205,
|
| 1162 |
+
"rewards/margins": 0.2491360902786255,
|
| 1163 |
+
"rewards/rejected": -1.854026198387146,
|
| 1164 |
+
"step": 385
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.8359008707300737,
|
| 1168 |
+
"grad_norm": 33.97745224085272,
|
| 1169 |
+
"learning_rate": 3.950255612243042e-08,
|
| 1170 |
+
"logits/chosen": -2.8420250415802,
|
| 1171 |
+
"logits/rejected": -2.840378522872925,
|
| 1172 |
+
"logps/chosen": -0.6417650580406189,
|
| 1173 |
+
"logps/rejected": -0.7058727145195007,
|
| 1174 |
+
"loss": 0.7771,
|
| 1175 |
+
"rewards/accuracies": 0.574999988079071,
|
| 1176 |
+
"rewards/chosen": -1.6044127941131592,
|
| 1177 |
+
"rewards/margins": 0.16026917099952698,
|
| 1178 |
+
"rewards/rejected": -1.7646818161010742,
|
| 1179 |
+
"step": 390
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.8466175485599464,
|
| 1183 |
+
"grad_norm": 41.64546946228987,
|
| 1184 |
+
"learning_rate": 3.459535260206384e-08,
|
| 1185 |
+
"logits/chosen": -2.836892604827881,
|
| 1186 |
+
"logits/rejected": -2.8546109199523926,
|
| 1187 |
+
"logps/chosen": -0.7084933519363403,
|
| 1188 |
+
"logps/rejected": -0.8302263021469116,
|
| 1189 |
+
"loss": 0.7662,
|
| 1190 |
+
"rewards/accuracies": 0.625,
|
| 1191 |
+
"rewards/chosen": -1.771233320236206,
|
| 1192 |
+
"rewards/margins": 0.3043324947357178,
|
| 1193 |
+
"rewards/rejected": -2.075565814971924,
|
| 1194 |
+
"step": 395
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 0.8573342263898192,
|
| 1198 |
+
"grad_norm": 38.05882539644461,
|
| 1199 |
+
"learning_rate": 2.999085151499764e-08,
|
| 1200 |
+
"logits/chosen": -2.8660452365875244,
|
| 1201 |
+
"logits/rejected": -2.886165142059326,
|
| 1202 |
+
"logps/chosen": -0.6565605401992798,
|
| 1203 |
+
"logps/rejected": -0.758074164390564,
|
| 1204 |
+
"loss": 0.7494,
|
| 1205 |
+
"rewards/accuracies": 0.643750011920929,
|
| 1206 |
+
"rewards/chosen": -1.6414012908935547,
|
| 1207 |
+
"rewards/margins": 0.25378426909446716,
|
| 1208 |
+
"rewards/rejected": -1.8951854705810547,
|
| 1209 |
+
"step": 400
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 0.8573342263898192,
|
| 1213 |
+
"eval_logits/chosen": -2.7592597007751465,
|
| 1214 |
+
"eval_logits/rejected": -2.7572638988494873,
|
| 1215 |
+
"eval_logps/chosen": -0.676339328289032,
|
| 1216 |
+
"eval_logps/rejected": -0.7378701567649841,
|
| 1217 |
+
"eval_loss": 0.7909048795700073,
|
| 1218 |
+
"eval_rewards/accuracies": 0.602393627166748,
|
| 1219 |
+
"eval_rewards/chosen": -1.690848469734192,
|
| 1220 |
+
"eval_rewards/margins": 0.15382711589336395,
|
| 1221 |
+
"eval_rewards/rejected": -1.8446754217147827,
|
| 1222 |
+
"eval_runtime": 102.5,
|
| 1223 |
+
"eval_samples_per_second": 29.21,
|
| 1224 |
+
"eval_steps_per_second": 1.834,
|
| 1225 |
+
"step": 400
|
| 1226 |
+
},
|
| 1227 |
+
{
|
| 1228 |
+
"epoch": 0.8680509042196919,
|
| 1229 |
+
"grad_norm": 36.86643037307591,
|
| 1230 |
+
"learning_rate": 2.5695523444387747e-08,
|
| 1231 |
+
"logits/chosen": -2.9111428260803223,
|
| 1232 |
+
"logits/rejected": -2.910399913787842,
|
| 1233 |
+
"logps/chosen": -0.6198450326919556,
|
| 1234 |
+
"logps/rejected": -0.7135489583015442,
|
| 1235 |
+
"loss": 0.7679,
|
| 1236 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1237 |
+
"rewards/chosen": -1.5496125221252441,
|
| 1238 |
+
"rewards/margins": 0.23425981402397156,
|
| 1239 |
+
"rewards/rejected": -1.7838722467422485,
|
| 1240 |
+
"step": 405
|
| 1241 |
+
},
|
| 1242 |
+
{
|
| 1243 |
+
"epoch": 0.8787675820495646,
|
| 1244 |
+
"grad_norm": 42.73305785852333,
|
| 1245 |
+
"learning_rate": 2.1715404500756875e-08,
|
| 1246 |
+
"logits/chosen": -2.8384289741516113,
|
| 1247 |
+
"logits/rejected": -2.8592209815979004,
|
| 1248 |
+
"logps/chosen": -0.648736834526062,
|
| 1249 |
+
"logps/rejected": -0.7135356664657593,
|
| 1250 |
+
"loss": 0.7489,
|
| 1251 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 1252 |
+
"rewards/chosen": -1.6218420267105103,
|
| 1253 |
+
"rewards/margins": 0.16199705004692078,
|
| 1254 |
+
"rewards/rejected": -1.7838389873504639,
|
| 1255 |
+
"step": 410
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 0.8894842598794374,
|
| 1259 |
+
"grad_norm": 40.990552105099376,
|
| 1260 |
+
"learning_rate": 1.8056087839608546e-08,
|
| 1261 |
+
"logits/chosen": -2.8605241775512695,
|
| 1262 |
+
"logits/rejected": -2.8745930194854736,
|
| 1263 |
+
"logps/chosen": -0.6644041538238525,
|
| 1264 |
+
"logps/rejected": -0.7614117860794067,
|
| 1265 |
+
"loss": 0.7642,
|
| 1266 |
+
"rewards/accuracies": 0.668749988079071,
|
| 1267 |
+
"rewards/chosen": -1.661010503768921,
|
| 1268 |
+
"rewards/margins": 0.2425190508365631,
|
| 1269 |
+
"rewards/rejected": -1.9035295248031616,
|
| 1270 |
+
"step": 415
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 0.9002009377093101,
|
| 1274 |
+
"grad_norm": 43.0964450321845,
|
| 1275 |
+
"learning_rate": 1.4722715801513935e-08,
|
| 1276 |
+
"logits/chosen": -2.884080171585083,
|
| 1277 |
+
"logits/rejected": -2.8876559734344482,
|
| 1278 |
+
"logps/chosen": -0.6492477655410767,
|
| 1279 |
+
"logps/rejected": -0.7468755841255188,
|
| 1280 |
+
"loss": 0.757,
|
| 1281 |
+
"rewards/accuracies": 0.637499988079071,
|
| 1282 |
+
"rewards/chosen": -1.6231193542480469,
|
| 1283 |
+
"rewards/margins": 0.2440694123506546,
|
| 1284 |
+
"rewards/rejected": -1.8671890497207642,
|
| 1285 |
+
"step": 420
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 0.9109176155391828,
|
| 1289 |
+
"grad_norm": 35.62063577969192,
|
| 1290 |
+
"learning_rate": 1.1719972685717395e-08,
|
| 1291 |
+
"logits/chosen": -2.870063304901123,
|
| 1292 |
+
"logits/rejected": -2.8697421550750732,
|
| 1293 |
+
"logps/chosen": -0.6383547186851501,
|
| 1294 |
+
"logps/rejected": -0.7067494988441467,
|
| 1295 |
+
"loss": 0.7706,
|
| 1296 |
+
"rewards/accuracies": 0.581250011920929,
|
| 1297 |
+
"rewards/chosen": -1.5958869457244873,
|
| 1298 |
+
"rewards/margins": 0.1709868460893631,
|
| 1299 |
+
"rewards/rejected": -1.7668737173080444,
|
| 1300 |
+
"step": 425
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"epoch": 0.9216342933690556,
|
| 1304 |
+
"grad_norm": 43.124891991653534,
|
| 1305 |
+
"learning_rate": 9.05207816741485e-09,
|
| 1306 |
+
"logits/chosen": -2.836172580718994,
|
| 1307 |
+
"logits/rejected": -2.8423163890838623,
|
| 1308 |
+
"logps/chosen": -0.6527460813522339,
|
| 1309 |
+
"logps/rejected": -0.730507493019104,
|
| 1310 |
+
"loss": 0.7819,
|
| 1311 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 1312 |
+
"rewards/chosen": -1.63186514377594,
|
| 1313 |
+
"rewards/margins": 0.19440340995788574,
|
| 1314 |
+
"rewards/rejected": -1.8262684345245361,
|
| 1315 |
+
"step": 430
|
| 1316 |
+
},
|
| 1317 |
+
{
|
| 1318 |
+
"epoch": 0.9323509711989283,
|
| 1319 |
+
"grad_norm": 36.86465436137469,
|
| 1320 |
+
"learning_rate": 6.722781367956481e-09,
|
| 1321 |
+
"logits/chosen": -2.782316207885742,
|
| 1322 |
+
"logits/rejected": -2.7888705730438232,
|
| 1323 |
+
"logps/chosen": -0.6667356491088867,
|
| 1324 |
+
"logps/rejected": -0.7845855355262756,
|
| 1325 |
+
"loss": 0.7675,
|
| 1326 |
+
"rewards/accuracies": 0.637499988079071,
|
| 1327 |
+
"rewards/chosen": -1.6668390035629272,
|
| 1328 |
+
"rewards/margins": 0.2946247458457947,
|
| 1329 |
+
"rewards/rejected": -1.9614636898040771,
|
| 1330 |
+
"step": 435
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"epoch": 0.943067649028801,
|
| 1334 |
+
"grad_norm": 47.527494016819155,
|
| 1335 |
+
"learning_rate": 4.735355586305861e-09,
|
| 1336 |
+
"logits/chosen": -2.836718797683716,
|
| 1337 |
+
"logits/rejected": -2.8620688915252686,
|
| 1338 |
+
"logps/chosen": -0.6587570309638977,
|
| 1339 |
+
"logps/rejected": -0.7252873778343201,
|
| 1340 |
+
"loss": 0.7738,
|
| 1341 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 1342 |
+
"rewards/chosen": -1.6468925476074219,
|
| 1343 |
+
"rewards/margins": 0.16632597148418427,
|
| 1344 |
+
"rewards/rejected": -1.8132184743881226,
|
| 1345 |
+
"step": 440
|
| 1346 |
+
},
|
| 1347 |
+
{
|
| 1348 |
+
"epoch": 0.9537843268586738,
|
| 1349 |
+
"grad_norm": 40.42551390475137,
|
| 1350 |
+
"learning_rate": 3.0925936991598454e-09,
|
| 1351 |
+
"logits/chosen": -2.839256763458252,
|
| 1352 |
+
"logits/rejected": -2.8521199226379395,
|
| 1353 |
+
"logps/chosen": -0.6958639025688171,
|
| 1354 |
+
"logps/rejected": -0.7931674718856812,
|
| 1355 |
+
"loss": 0.7647,
|
| 1356 |
+
"rewards/accuracies": 0.637499988079071,
|
| 1357 |
+
"rewards/chosen": -1.7396596670150757,
|
| 1358 |
+
"rewards/margins": 0.243258997797966,
|
| 1359 |
+
"rewards/rejected": -1.9829187393188477,
|
| 1360 |
+
"step": 445
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 0.9645010046885466,
|
| 1364 |
+
"grad_norm": 42.24998694242168,
|
| 1365 |
+
"learning_rate": 1.7968042361930225e-09,
|
| 1366 |
+
"logits/chosen": -2.821035623550415,
|
| 1367 |
+
"logits/rejected": -2.843494415283203,
|
| 1368 |
+
"logps/chosen": -0.6622462868690491,
|
| 1369 |
+
"logps/rejected": -0.7363108396530151,
|
| 1370 |
+
"loss": 0.7558,
|
| 1371 |
+
"rewards/accuracies": 0.612500011920929,
|
| 1372 |
+
"rewards/chosen": -1.6556155681610107,
|
| 1373 |
+
"rewards/margins": 0.1851612776517868,
|
| 1374 |
+
"rewards/rejected": -1.8407771587371826,
|
| 1375 |
+
"step": 450
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 0.9752176825184193,
|
| 1379 |
+
"grad_norm": 58.74520439596834,
|
| 1380 |
+
"learning_rate": 8.498081359423759e-10,
|
| 1381 |
+
"logits/chosen": -2.8467049598693848,
|
| 1382 |
+
"logits/rejected": -2.8465356826782227,
|
| 1383 |
+
"logps/chosen": -0.6843534708023071,
|
| 1384 |
+
"logps/rejected": -0.7486766576766968,
|
| 1385 |
+
"loss": 0.7873,
|
| 1386 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 1387 |
+
"rewards/chosen": -1.7108838558197021,
|
| 1388 |
+
"rewards/margins": 0.1608080118894577,
|
| 1389 |
+
"rewards/rejected": -1.8716917037963867,
|
| 1390 |
+
"step": 455
|
| 1391 |
+
},
|
| 1392 |
+
{
|
| 1393 |
+
"epoch": 0.9859343603482921,
|
| 1394 |
+
"grad_norm": 41.029699707171545,
|
| 1395 |
+
"learning_rate": 2.5293618689056774e-10,
|
| 1396 |
+
"logits/chosen": -2.8341171741485596,
|
| 1397 |
+
"logits/rejected": -2.8409810066223145,
|
| 1398 |
+
"logps/chosen": -0.6728862524032593,
|
| 1399 |
+
"logps/rejected": -0.7803421020507812,
|
| 1400 |
+
"loss": 0.7617,
|
| 1401 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 1402 |
+
"rewards/chosen": -1.6822154521942139,
|
| 1403 |
+
"rewards/margins": 0.2686397135257721,
|
| 1404 |
+
"rewards/rejected": -1.9508554935455322,
|
| 1405 |
+
"step": 460
|
| 1406 |
+
},
|
| 1407 |
+
{
|
| 1408 |
+
"epoch": 0.9966510381781648,
|
| 1409 |
+
"grad_norm": 36.64296263080341,
|
| 1410 |
+
"learning_rate": 7.02715734424153e-12,
|
| 1411 |
+
"logits/chosen": -2.8702194690704346,
|
| 1412 |
+
"logits/rejected": -2.8706626892089844,
|
| 1413 |
+
"logps/chosen": -0.6983669996261597,
|
| 1414 |
+
"logps/rejected": -0.7460958361625671,
|
| 1415 |
+
"loss": 0.7864,
|
| 1416 |
+
"rewards/accuracies": 0.5625,
|
| 1417 |
+
"rewards/chosen": -1.745917558670044,
|
| 1418 |
+
"rewards/margins": 0.11932210624217987,
|
| 1419 |
+
"rewards/rejected": -1.8652395009994507,
|
| 1420 |
+
"step": 465
|
| 1421 |
+
},
|
| 1422 |
+
{
|
| 1423 |
+
"epoch": 0.9987943737441393,
|
| 1424 |
+
"step": 466,
|
| 1425 |
+
"total_flos": 0.0,
|
| 1426 |
+
"train_loss": 0.7914171927476645,
|
| 1427 |
+
"train_runtime": 8402.7408,
|
| 1428 |
+
"train_samples_per_second": 7.107,
|
| 1429 |
+
"train_steps_per_second": 0.055
|
| 1430 |
+
}
|
| 1431 |
+
],
|
| 1432 |
+
"logging_steps": 5,
|
| 1433 |
+
"max_steps": 466,
|
| 1434 |
+
"num_input_tokens_seen": 0,
|
| 1435 |
+
"num_train_epochs": 1,
|
| 1436 |
+
"save_steps": 1000000,
|
| 1437 |
+
"stateful_callbacks": {
|
| 1438 |
+
"TrainerControl": {
|
| 1439 |
+
"args": {
|
| 1440 |
+
"should_epoch_stop": false,
|
| 1441 |
+
"should_evaluate": false,
|
| 1442 |
+
"should_log": false,
|
| 1443 |
+
"should_save": true,
|
| 1444 |
+
"should_training_stop": true
|
| 1445 |
+
},
|
| 1446 |
+
"attributes": {}
|
| 1447 |
+
}
|
| 1448 |
+
},
|
| 1449 |
+
"total_flos": 0.0,
|
| 1450 |
+
"train_batch_size": 2,
|
| 1451 |
+
"trial_name": null,
|
| 1452 |
+
"trial_params": null
|
| 1453 |
+
}
|