Model save
Browse files- README.md +74 -0
- all_results.json +9 -0
- generation_config.json +12 -0
- train_results.json +9 -0
- trainer_state.json +1423 -0
README.md
ADDED
|
@@ -0,0 +1,74 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
license: llama3
|
| 4 |
+
base_model: meta-llama/Meta-Llama-3-8B-Instruct
|
| 5 |
+
tags:
|
| 6 |
+
- trl
|
| 7 |
+
- simpo
|
| 8 |
+
- generated_from_trainer
|
| 9 |
+
model-index:
|
| 10 |
+
- name: llama3instruct_-qfUNL-10-0_3-1e-6-1_best
|
| 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 |
+
# llama3instruct_-qfUNL-10-0_3-1e-6-1_best
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [meta-llama/Meta-Llama-3-8B-Instruct](https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct) on an unknown dataset.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 1.8816
|
| 22 |
+
- Rewards/chosen: -5.4374
|
| 23 |
+
- Rewards/rejected: -7.2714
|
| 24 |
+
- Rewards/accuracies: 0.7711
|
| 25 |
+
- Rewards/margins: 1.8339
|
| 26 |
+
- Logps/rejected: -0.7271
|
| 27 |
+
- Logps/chosen: -0.5437
|
| 28 |
+
- Logits/rejected: -1.3070
|
| 29 |
+
- Logits/chosen: -1.3448
|
| 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: 1e-06
|
| 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: 1.0
|
| 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 |
+
| 1.8354 | 0.8743 | 400 | 1.8816 | -5.4374 | -7.2714 | 0.7711 | 1.8339 | -0.7271 | -0.5437 | -1.3070 | -1.3448 |
|
| 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": 0.9989071038251366,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 2.2740620700744363,
|
| 5 |
+
"train_runtime": 5957.0658,
|
| 6 |
+
"train_samples": 58558,
|
| 7 |
+
"train_samples_per_second": 9.83,
|
| 8 |
+
"train_steps_per_second": 0.077
|
| 9 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 128000,
|
| 3 |
+
"do_sample": true,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
128001,
|
| 6 |
+
128009
|
| 7 |
+
],
|
| 8 |
+
"max_length": 4096,
|
| 9 |
+
"temperature": 0.6,
|
| 10 |
+
"top_p": 0.9,
|
| 11 |
+
"transformers_version": "4.44.2"
|
| 12 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9989071038251366,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 2.2740620700744363,
|
| 5 |
+
"train_runtime": 5957.0658,
|
| 6 |
+
"train_samples": 58558,
|
| 7 |
+
"train_samples_per_second": 9.83,
|
| 8 |
+
"train_steps_per_second": 0.077
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1423 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9989071038251366,
|
| 5 |
+
"eval_steps": 400,
|
| 6 |
+
"global_step": 457,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.01092896174863388,
|
| 13 |
+
"grad_norm": 334.04762282844746,
|
| 14 |
+
"learning_rate": 1.0869565217391303e-07,
|
| 15 |
+
"logits/chosen": -1.0183193683624268,
|
| 16 |
+
"logits/rejected": -1.0128684043884277,
|
| 17 |
+
"logps/chosen": -0.2805940508842468,
|
| 18 |
+
"logps/rejected": -0.28598836064338684,
|
| 19 |
+
"loss": 3.2497,
|
| 20 |
+
"rewards/accuracies": 0.53125,
|
| 21 |
+
"rewards/chosen": -2.805940628051758,
|
| 22 |
+
"rewards/margins": 0.05394287034869194,
|
| 23 |
+
"rewards/rejected": -2.8598835468292236,
|
| 24 |
+
"step": 5
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.02185792349726776,
|
| 28 |
+
"grad_norm": 180.71056791479407,
|
| 29 |
+
"learning_rate": 2.1739130434782607e-07,
|
| 30 |
+
"logits/chosen": -1.0511164665222168,
|
| 31 |
+
"logits/rejected": -1.0003410577774048,
|
| 32 |
+
"logps/chosen": -0.2573581039905548,
|
| 33 |
+
"logps/rejected": -0.27061566710472107,
|
| 34 |
+
"loss": 3.257,
|
| 35 |
+
"rewards/accuracies": 0.53125,
|
| 36 |
+
"rewards/chosen": -2.5735812187194824,
|
| 37 |
+
"rewards/margins": 0.13257600367069244,
|
| 38 |
+
"rewards/rejected": -2.7061567306518555,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.03278688524590164,
|
| 43 |
+
"grad_norm": 223.3194766304879,
|
| 44 |
+
"learning_rate": 3.260869565217391e-07,
|
| 45 |
+
"logits/chosen": -0.9949806928634644,
|
| 46 |
+
"logits/rejected": -0.9507344961166382,
|
| 47 |
+
"logps/chosen": -0.26757049560546875,
|
| 48 |
+
"logps/rejected": -0.2733513414859772,
|
| 49 |
+
"loss": 3.2029,
|
| 50 |
+
"rewards/accuracies": 0.518750011920929,
|
| 51 |
+
"rewards/chosen": -2.6757049560546875,
|
| 52 |
+
"rewards/margins": 0.05780839920043945,
|
| 53 |
+
"rewards/rejected": -2.733513355255127,
|
| 54 |
+
"step": 15
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.04371584699453552,
|
| 58 |
+
"grad_norm": 208.56574432084847,
|
| 59 |
+
"learning_rate": 4.3478260869565214e-07,
|
| 60 |
+
"logits/chosen": -0.9485995173454285,
|
| 61 |
+
"logits/rejected": -0.8978453874588013,
|
| 62 |
+
"logps/chosen": -0.2727873921394348,
|
| 63 |
+
"logps/rejected": -0.28461021184921265,
|
| 64 |
+
"loss": 3.2536,
|
| 65 |
+
"rewards/accuracies": 0.550000011920929,
|
| 66 |
+
"rewards/chosen": -2.7278740406036377,
|
| 67 |
+
"rewards/margins": 0.11822839826345444,
|
| 68 |
+
"rewards/rejected": -2.846102237701416,
|
| 69 |
+
"step": 20
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.0546448087431694,
|
| 73 |
+
"grad_norm": 273.9157814428763,
|
| 74 |
+
"learning_rate": 5.434782608695652e-07,
|
| 75 |
+
"logits/chosen": -0.9404302835464478,
|
| 76 |
+
"logits/rejected": -0.8673273324966431,
|
| 77 |
+
"logps/chosen": -0.2763015627861023,
|
| 78 |
+
"logps/rejected": -0.29439201951026917,
|
| 79 |
+
"loss": 3.1849,
|
| 80 |
+
"rewards/accuracies": 0.581250011920929,
|
| 81 |
+
"rewards/chosen": -2.7630152702331543,
|
| 82 |
+
"rewards/margins": 0.18090477585792542,
|
| 83 |
+
"rewards/rejected": -2.943920135498047,
|
| 84 |
+
"step": 25
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.06557377049180328,
|
| 88 |
+
"grad_norm": 252.26801160371912,
|
| 89 |
+
"learning_rate": 6.521739130434782e-07,
|
| 90 |
+
"logits/chosen": -1.0567680597305298,
|
| 91 |
+
"logits/rejected": -0.9916139841079712,
|
| 92 |
+
"logps/chosen": -0.26832228899002075,
|
| 93 |
+
"logps/rejected": -0.2841094732284546,
|
| 94 |
+
"loss": 3.2649,
|
| 95 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 96 |
+
"rewards/chosen": -2.683222770690918,
|
| 97 |
+
"rewards/margins": 0.15787188708782196,
|
| 98 |
+
"rewards/rejected": -2.841094493865967,
|
| 99 |
+
"step": 30
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.07650273224043716,
|
| 103 |
+
"grad_norm": 120.68425546239968,
|
| 104 |
+
"learning_rate": 7.608695652173913e-07,
|
| 105 |
+
"logits/chosen": -1.0046221017837524,
|
| 106 |
+
"logits/rejected": -0.9383124113082886,
|
| 107 |
+
"logps/chosen": -0.2608323395252228,
|
| 108 |
+
"logps/rejected": -0.27886277437210083,
|
| 109 |
+
"loss": 3.1693,
|
| 110 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 111 |
+
"rewards/chosen": -2.608323335647583,
|
| 112 |
+
"rewards/margins": 0.1803044080734253,
|
| 113 |
+
"rewards/rejected": -2.788627862930298,
|
| 114 |
+
"step": 35
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.08743169398907104,
|
| 118 |
+
"grad_norm": 146.20382704567854,
|
| 119 |
+
"learning_rate": 8.695652173913043e-07,
|
| 120 |
+
"logits/chosen": -0.9549115896224976,
|
| 121 |
+
"logits/rejected": -0.8933135271072388,
|
| 122 |
+
"logps/chosen": -0.28391939401626587,
|
| 123 |
+
"logps/rejected": -0.29877811670303345,
|
| 124 |
+
"loss": 3.2899,
|
| 125 |
+
"rewards/accuracies": 0.518750011920929,
|
| 126 |
+
"rewards/chosen": -2.83919358253479,
|
| 127 |
+
"rewards/margins": 0.14858722686767578,
|
| 128 |
+
"rewards/rejected": -2.987781047821045,
|
| 129 |
+
"step": 40
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.09836065573770492,
|
| 133 |
+
"grad_norm": 93.83055902381597,
|
| 134 |
+
"learning_rate": 9.782608695652173e-07,
|
| 135 |
+
"logits/chosen": -1.0223612785339355,
|
| 136 |
+
"logits/rejected": -0.9385920763015747,
|
| 137 |
+
"logps/chosen": -0.2871010899543762,
|
| 138 |
+
"logps/rejected": -0.3065423369407654,
|
| 139 |
+
"loss": 3.0292,
|
| 140 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 141 |
+
"rewards/chosen": -2.8710105419158936,
|
| 142 |
+
"rewards/margins": 0.19441291689872742,
|
| 143 |
+
"rewards/rejected": -3.0654234886169434,
|
| 144 |
+
"step": 45
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.1092896174863388,
|
| 148 |
+
"grad_norm": 281.2686273771578,
|
| 149 |
+
"learning_rate": 9.997663088532014e-07,
|
| 150 |
+
"logits/chosen": -0.9519433975219727,
|
| 151 |
+
"logits/rejected": -0.8702558279037476,
|
| 152 |
+
"logps/chosen": -0.28290504217147827,
|
| 153 |
+
"logps/rejected": -0.29115915298461914,
|
| 154 |
+
"loss": 3.1626,
|
| 155 |
+
"rewards/accuracies": 0.48750001192092896,
|
| 156 |
+
"rewards/chosen": -2.829050302505493,
|
| 157 |
+
"rewards/margins": 0.08254138380289078,
|
| 158 |
+
"rewards/rejected": -2.9115920066833496,
|
| 159 |
+
"step": 50
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.12021857923497267,
|
| 163 |
+
"grad_norm": 121.70775424511473,
|
| 164 |
+
"learning_rate": 9.98817312944725e-07,
|
| 165 |
+
"logits/chosen": -0.9872657060623169,
|
| 166 |
+
"logits/rejected": -0.8688336610794067,
|
| 167 |
+
"logps/chosen": -0.2818766236305237,
|
| 168 |
+
"logps/rejected": -0.3157486021518707,
|
| 169 |
+
"loss": 3.0197,
|
| 170 |
+
"rewards/accuracies": 0.543749988079071,
|
| 171 |
+
"rewards/chosen": -2.8187661170959473,
|
| 172 |
+
"rewards/margins": 0.3387202322483063,
|
| 173 |
+
"rewards/rejected": -3.1574864387512207,
|
| 174 |
+
"step": 55
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.13114754098360656,
|
| 178 |
+
"grad_norm": 86.612813297471,
|
| 179 |
+
"learning_rate": 9.971397915250336e-07,
|
| 180 |
+
"logits/chosen": -1.0298466682434082,
|
| 181 |
+
"logits/rejected": -0.9812299609184265,
|
| 182 |
+
"logps/chosen": -0.2719958424568176,
|
| 183 |
+
"logps/rejected": -0.31191515922546387,
|
| 184 |
+
"loss": 2.8695,
|
| 185 |
+
"rewards/accuracies": 0.59375,
|
| 186 |
+
"rewards/chosen": -2.719958782196045,
|
| 187 |
+
"rewards/margins": 0.399192750453949,
|
| 188 |
+
"rewards/rejected": -3.1191515922546387,
|
| 189 |
+
"step": 60
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.14207650273224043,
|
| 193 |
+
"grad_norm": 161.66395420923772,
|
| 194 |
+
"learning_rate": 9.94736194623663e-07,
|
| 195 |
+
"logits/chosen": -0.9994015693664551,
|
| 196 |
+
"logits/rejected": -0.9284494519233704,
|
| 197 |
+
"logps/chosen": -0.31242281198501587,
|
| 198 |
+
"logps/rejected": -0.34273454546928406,
|
| 199 |
+
"loss": 3.0819,
|
| 200 |
+
"rewards/accuracies": 0.53125,
|
| 201 |
+
"rewards/chosen": -3.124228000640869,
|
| 202 |
+
"rewards/margins": 0.30311769247055054,
|
| 203 |
+
"rewards/rejected": -3.4273457527160645,
|
| 204 |
+
"step": 65
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.15300546448087432,
|
| 208 |
+
"grad_norm": 335.25942540938917,
|
| 209 |
+
"learning_rate": 9.916100327075037e-07,
|
| 210 |
+
"logits/chosen": -0.9326862096786499,
|
| 211 |
+
"logits/rejected": -0.9134162664413452,
|
| 212 |
+
"logps/chosen": -0.3001170754432678,
|
| 213 |
+
"logps/rejected": -0.3256034851074219,
|
| 214 |
+
"loss": 2.8894,
|
| 215 |
+
"rewards/accuracies": 0.581250011920929,
|
| 216 |
+
"rewards/chosen": -3.0011706352233887,
|
| 217 |
+
"rewards/margins": 0.2548643946647644,
|
| 218 |
+
"rewards/rejected": -3.2560348510742188,
|
| 219 |
+
"step": 70
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.16393442622950818,
|
| 223 |
+
"grad_norm": 79.67001422366646,
|
| 224 |
+
"learning_rate": 9.877658715537428e-07,
|
| 225 |
+
"logits/chosen": -0.9303417205810547,
|
| 226 |
+
"logits/rejected": -0.9135540127754211,
|
| 227 |
+
"logps/chosen": -0.32335883378982544,
|
| 228 |
+
"logps/rejected": -0.35447338223457336,
|
| 229 |
+
"loss": 2.9985,
|
| 230 |
+
"rewards/accuracies": 0.625,
|
| 231 |
+
"rewards/chosen": -3.233588695526123,
|
| 232 |
+
"rewards/margins": 0.3111455738544464,
|
| 233 |
+
"rewards/rejected": -3.544733762741089,
|
| 234 |
+
"step": 75
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.17486338797814208,
|
| 238 |
+
"grad_norm": 83.46864509697639,
|
| 239 |
+
"learning_rate": 9.832093255815216e-07,
|
| 240 |
+
"logits/chosen": -0.942563533782959,
|
| 241 |
+
"logits/rejected": -0.8771953582763672,
|
| 242 |
+
"logps/chosen": -0.31826624274253845,
|
| 243 |
+
"logps/rejected": -0.3393171727657318,
|
| 244 |
+
"loss": 2.9663,
|
| 245 |
+
"rewards/accuracies": 0.606249988079071,
|
| 246 |
+
"rewards/chosen": -3.1826624870300293,
|
| 247 |
+
"rewards/margins": 0.21050938963890076,
|
| 248 |
+
"rewards/rejected": -3.393172025680542,
|
| 249 |
+
"step": 80
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.18579234972677597,
|
| 253 |
+
"grad_norm": 119.60039352898374,
|
| 254 |
+
"learning_rate": 9.779470496520441e-07,
|
| 255 |
+
"logits/chosen": -0.9433110952377319,
|
| 256 |
+
"logits/rejected": -0.8911749124526978,
|
| 257 |
+
"logps/chosen": -0.31765812635421753,
|
| 258 |
+
"logps/rejected": -0.3729430139064789,
|
| 259 |
+
"loss": 2.8342,
|
| 260 |
+
"rewards/accuracies": 0.625,
|
| 261 |
+
"rewards/chosen": -3.176581621170044,
|
| 262 |
+
"rewards/margins": 0.5528485178947449,
|
| 263 |
+
"rewards/rejected": -3.7294304370880127,
|
| 264 |
+
"step": 85
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.19672131147540983,
|
| 268 |
+
"grad_norm": 96.2051979280833,
|
| 269 |
+
"learning_rate": 9.719867293491144e-07,
|
| 270 |
+
"logits/chosen": -1.0151774883270264,
|
| 271 |
+
"logits/rejected": -0.9306642413139343,
|
| 272 |
+
"logps/chosen": -0.3450705409049988,
|
| 273 |
+
"logps/rejected": -0.38217830657958984,
|
| 274 |
+
"loss": 2.9117,
|
| 275 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 276 |
+
"rewards/chosen": -3.4507057666778564,
|
| 277 |
+
"rewards/margins": 0.37107738852500916,
|
| 278 |
+
"rewards/rejected": -3.8217825889587402,
|
| 279 |
+
"step": 90
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.20765027322404372,
|
| 283 |
+
"grad_norm": 90.0031374445934,
|
| 284 |
+
"learning_rate": 9.653370697542987e-07,
|
| 285 |
+
"logits/chosen": -0.9307557344436646,
|
| 286 |
+
"logits/rejected": -0.9305530786514282,
|
| 287 |
+
"logps/chosen": -0.34383976459503174,
|
| 288 |
+
"logps/rejected": -0.3695612847805023,
|
| 289 |
+
"loss": 2.7261,
|
| 290 |
+
"rewards/accuracies": 0.581250011920929,
|
| 291 |
+
"rewards/chosen": -3.4383978843688965,
|
| 292 |
+
"rewards/margins": 0.2572151720523834,
|
| 293 |
+
"rewards/rejected": -3.695613145828247,
|
| 294 |
+
"step": 95
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.2185792349726776,
|
| 298 |
+
"grad_norm": 106.07682375515387,
|
| 299 |
+
"learning_rate": 9.580077827331037e-07,
|
| 300 |
+
"logits/chosen": -0.9448221921920776,
|
| 301 |
+
"logits/rejected": -0.8973434567451477,
|
| 302 |
+
"logps/chosen": -0.38050413131713867,
|
| 303 |
+
"logps/rejected": -0.4358042776584625,
|
| 304 |
+
"loss": 2.7135,
|
| 305 |
+
"rewards/accuracies": 0.637499988079071,
|
| 306 |
+
"rewards/chosen": -3.805041551589966,
|
| 307 |
+
"rewards/margins": 0.5530012249946594,
|
| 308 |
+
"rewards/rejected": -4.3580427169799805,
|
| 309 |
+
"step": 100
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.22950819672131148,
|
| 313 |
+
"grad_norm": 124.64829049828732,
|
| 314 |
+
"learning_rate": 9.500095727507419e-07,
|
| 315 |
+
"logits/chosen": -0.9876478910446167,
|
| 316 |
+
"logits/rejected": -0.9548322558403015,
|
| 317 |
+
"logps/chosen": -0.3621714115142822,
|
| 318 |
+
"logps/rejected": -0.4053111672401428,
|
| 319 |
+
"loss": 2.8156,
|
| 320 |
+
"rewards/accuracies": 0.625,
|
| 321 |
+
"rewards/chosen": -3.6217141151428223,
|
| 322 |
+
"rewards/margins": 0.43139734864234924,
|
| 323 |
+
"rewards/rejected": -4.0531110763549805,
|
| 324 |
+
"step": 105
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.24043715846994534,
|
| 328 |
+
"grad_norm": 109.72879050820204,
|
| 329 |
+
"learning_rate": 9.413541212382004e-07,
|
| 330 |
+
"logits/chosen": -0.9878107905387878,
|
| 331 |
+
"logits/rejected": -0.9694309234619141,
|
| 332 |
+
"logps/chosen": -0.37187474966049194,
|
| 333 |
+
"logps/rejected": -0.4513128697872162,
|
| 334 |
+
"loss": 2.6674,
|
| 335 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 336 |
+
"rewards/chosen": -3.718747615814209,
|
| 337 |
+
"rewards/margins": 0.7943812608718872,
|
| 338 |
+
"rewards/rejected": -4.513128280639648,
|
| 339 |
+
"step": 110
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.25136612021857924,
|
| 343 |
+
"grad_norm": 89.69262563371302,
|
| 344 |
+
"learning_rate": 9.320540695314438e-07,
|
| 345 |
+
"logits/chosen": -0.9974454045295715,
|
| 346 |
+
"logits/rejected": -0.9548517465591431,
|
| 347 |
+
"logps/chosen": -0.3737660050392151,
|
| 348 |
+
"logps/rejected": -0.47232741117477417,
|
| 349 |
+
"loss": 2.5733,
|
| 350 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 351 |
+
"rewards/chosen": -3.7376599311828613,
|
| 352 |
+
"rewards/margins": 0.9856147766113281,
|
| 353 |
+
"rewards/rejected": -4.7232747077941895,
|
| 354 |
+
"step": 115
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.26229508196721313,
|
| 358 |
+
"grad_norm": 82.76010396371875,
|
| 359 |
+
"learning_rate": 9.221230004086721e-07,
|
| 360 |
+
"logits/chosen": -1.0242812633514404,
|
| 361 |
+
"logits/rejected": -0.949226975440979,
|
| 362 |
+
"logps/chosen": -0.3816962242126465,
|
| 363 |
+
"logps/rejected": -0.4385160803794861,
|
| 364 |
+
"loss": 2.6795,
|
| 365 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 366 |
+
"rewards/chosen": -3.816962480545044,
|
| 367 |
+
"rewards/margins": 0.5681988000869751,
|
| 368 |
+
"rewards/rejected": -4.38516092300415,
|
| 369 |
+
"step": 120
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.273224043715847,
|
| 373 |
+
"grad_norm": 186.60007457771528,
|
| 374 |
+
"learning_rate": 9.11575418252596e-07,
|
| 375 |
+
"logits/chosen": -0.9328588247299194,
|
| 376 |
+
"logits/rejected": -0.8940401077270508,
|
| 377 |
+
"logps/chosen": -0.395950585603714,
|
| 378 |
+
"logps/rejected": -0.46849578619003296,
|
| 379 |
+
"loss": 2.5289,
|
| 380 |
+
"rewards/accuracies": 0.6875,
|
| 381 |
+
"rewards/chosen": -3.9595062732696533,
|
| 382 |
+
"rewards/margins": 0.7254521250724792,
|
| 383 |
+
"rewards/rejected": -4.684957981109619,
|
| 384 |
+
"step": 125
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.28415300546448086,
|
| 388 |
+
"grad_norm": 91.04741453080123,
|
| 389 |
+
"learning_rate": 9.004267278667031e-07,
|
| 390 |
+
"logits/chosen": -0.9401098489761353,
|
| 391 |
+
"logits/rejected": -0.9308618307113647,
|
| 392 |
+
"logps/chosen": -0.41634416580200195,
|
| 393 |
+
"logps/rejected": -0.5375351905822754,
|
| 394 |
+
"loss": 2.5277,
|
| 395 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 396 |
+
"rewards/chosen": -4.1634416580200195,
|
| 397 |
+
"rewards/margins": 1.2119100093841553,
|
| 398 |
+
"rewards/rejected": -5.375351905822754,
|
| 399 |
+
"step": 130
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.29508196721311475,
|
| 403 |
+
"grad_norm": 72.34007677315606,
|
| 404 |
+
"learning_rate": 8.886932119764565e-07,
|
| 405 |
+
"logits/chosen": -0.9937323331832886,
|
| 406 |
+
"logits/rejected": -0.9037588834762573,
|
| 407 |
+
"logps/chosen": -0.39853519201278687,
|
| 408 |
+
"logps/rejected": -0.49872303009033203,
|
| 409 |
+
"loss": 2.4319,
|
| 410 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 411 |
+
"rewards/chosen": -3.9853522777557373,
|
| 412 |
+
"rewards/margins": 1.0018781423568726,
|
| 413 |
+
"rewards/rejected": -4.987229824066162,
|
| 414 |
+
"step": 135
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.30601092896174864,
|
| 418 |
+
"grad_norm": 83.52757366186898,
|
| 419 |
+
"learning_rate": 8.763920074482809e-07,
|
| 420 |
+
"logits/chosen": -0.9973970651626587,
|
| 421 |
+
"logits/rejected": -0.9379005432128906,
|
| 422 |
+
"logps/chosen": -0.42847341299057007,
|
| 423 |
+
"logps/rejected": -0.560250461101532,
|
| 424 |
+
"loss": 2.2317,
|
| 425 |
+
"rewards/accuracies": 0.706250011920929,
|
| 426 |
+
"rewards/chosen": -4.28473424911499,
|
| 427 |
+
"rewards/margins": 1.3177704811096191,
|
| 428 |
+
"rewards/rejected": -5.602505207061768,
|
| 429 |
+
"step": 140
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.31693989071038253,
|
| 433 |
+
"grad_norm": 106.73407366387536,
|
| 434 |
+
"learning_rate": 8.635410802610723e-07,
|
| 435 |
+
"logits/chosen": -0.9793792963027954,
|
| 436 |
+
"logits/rejected": -0.9569779634475708,
|
| 437 |
+
"logps/chosen": -0.4017811417579651,
|
| 438 |
+
"logps/rejected": -0.4612775444984436,
|
| 439 |
+
"loss": 2.3643,
|
| 440 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 441 |
+
"rewards/chosen": -4.017810821533203,
|
| 442 |
+
"rewards/margins": 0.5949639081954956,
|
| 443 |
+
"rewards/rejected": -4.612774848937988,
|
| 444 |
+
"step": 145
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.32786885245901637,
|
| 448 |
+
"grad_norm": 60.51485768923878,
|
| 449 |
+
"learning_rate": 8.501591992667849e-07,
|
| 450 |
+
"logits/chosen": -1.044575572013855,
|
| 451 |
+
"logits/rejected": -1.0059934854507446,
|
| 452 |
+
"logps/chosen": -0.4378751218318939,
|
| 453 |
+
"logps/rejected": -0.5975905656814575,
|
| 454 |
+
"loss": 2.2151,
|
| 455 |
+
"rewards/accuracies": 0.737500011920929,
|
| 456 |
+
"rewards/chosen": -4.378750801086426,
|
| 457 |
+
"rewards/margins": 1.5971543788909912,
|
| 458 |
+
"rewards/rejected": -5.975905418395996,
|
| 459 |
+
"step": 150
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.33879781420765026,
|
| 463 |
+
"grad_norm": 68.73684458948155,
|
| 464 |
+
"learning_rate": 8.362659087784152e-07,
|
| 465 |
+
"logits/chosen": -0.9929186105728149,
|
| 466 |
+
"logits/rejected": -0.9317314028739929,
|
| 467 |
+
"logps/chosen": -0.4304039478302002,
|
| 468 |
+
"logps/rejected": -0.5334851741790771,
|
| 469 |
+
"loss": 2.3197,
|
| 470 |
+
"rewards/accuracies": 0.675000011920929,
|
| 471 |
+
"rewards/chosen": -4.304039478302002,
|
| 472 |
+
"rewards/margins": 1.0308125019073486,
|
| 473 |
+
"rewards/rejected": -5.3348517417907715,
|
| 474 |
+
"step": 155
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.34972677595628415,
|
| 478 |
+
"grad_norm": 90.32236437312078,
|
| 479 |
+
"learning_rate": 8.218815000254231e-07,
|
| 480 |
+
"logits/chosen": -1.0294911861419678,
|
| 481 |
+
"logits/rejected": -0.9657951593399048,
|
| 482 |
+
"logps/chosen": -0.4845092296600342,
|
| 483 |
+
"logps/rejected": -0.566204309463501,
|
| 484 |
+
"loss": 2.3671,
|
| 485 |
+
"rewards/accuracies": 0.6875,
|
| 486 |
+
"rewards/chosen": -4.845091819763184,
|
| 487 |
+
"rewards/margins": 0.81695157289505,
|
| 488 |
+
"rewards/rejected": -5.662043571472168,
|
| 489 |
+
"step": 160
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.36065573770491804,
|
| 493 |
+
"grad_norm": 83.55321323451464,
|
| 494 |
+
"learning_rate": 8.07026981518276e-07,
|
| 495 |
+
"logits/chosen": -1.0140776634216309,
|
| 496 |
+
"logits/rejected": -0.9537431597709656,
|
| 497 |
+
"logps/chosen": -0.46461719274520874,
|
| 498 |
+
"logps/rejected": -0.5585328340530396,
|
| 499 |
+
"loss": 2.3255,
|
| 500 |
+
"rewards/accuracies": 0.737500011920929,
|
| 501 |
+
"rewards/chosen": -4.646172046661377,
|
| 502 |
+
"rewards/margins": 0.9391565322875977,
|
| 503 |
+
"rewards/rejected": -5.585328578948975,
|
| 504 |
+
"step": 165
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 0.37158469945355194,
|
| 508 |
+
"grad_norm": 91.94544798993923,
|
| 509 |
+
"learning_rate": 7.917240483654e-07,
|
| 510 |
+
"logits/chosen": -1.0053484439849854,
|
| 511 |
+
"logits/rejected": -0.9363968968391418,
|
| 512 |
+
"logps/chosen": -0.4600796699523926,
|
| 513 |
+
"logps/rejected": -0.5375440716743469,
|
| 514 |
+
"loss": 2.416,
|
| 515 |
+
"rewards/accuracies": 0.6875,
|
| 516 |
+
"rewards/chosen": -4.600796699523926,
|
| 517 |
+
"rewards/margins": 0.7746434211730957,
|
| 518 |
+
"rewards/rejected": -5.3754401206970215,
|
| 519 |
+
"step": 170
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.3825136612021858,
|
| 523 |
+
"grad_norm": 91.07280013110446,
|
| 524 |
+
"learning_rate": 7.759950505873521e-07,
|
| 525 |
+
"logits/chosen": -1.0666639804840088,
|
| 526 |
+
"logits/rejected": -1.029950737953186,
|
| 527 |
+
"logps/chosen": -0.4774397313594818,
|
| 528 |
+
"logps/rejected": -0.5452654957771301,
|
| 529 |
+
"loss": 2.2653,
|
| 530 |
+
"rewards/accuracies": 0.643750011920929,
|
| 531 |
+
"rewards/chosen": -4.774397850036621,
|
| 532 |
+
"rewards/margins": 0.6782568097114563,
|
| 533 |
+
"rewards/rejected": -5.452654838562012,
|
| 534 |
+
"step": 175
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.39344262295081966,
|
| 538 |
+
"grad_norm": 85.29762605205215,
|
| 539 |
+
"learning_rate": 7.598629604744872e-07,
|
| 540 |
+
"logits/chosen": -1.0595664978027344,
|
| 541 |
+
"logits/rejected": -1.0487810373306274,
|
| 542 |
+
"logps/chosen": -0.4570770263671875,
|
| 543 |
+
"logps/rejected": -0.5979724526405334,
|
| 544 |
+
"loss": 2.1524,
|
| 545 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 546 |
+
"rewards/chosen": -4.570770263671875,
|
| 547 |
+
"rewards/margins": 1.4089539051055908,
|
| 548 |
+
"rewards/rejected": -5.979724407196045,
|
| 549 |
+
"step": 180
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.40437158469945356,
|
| 553 |
+
"grad_norm": 67.2351189385839,
|
| 554 |
+
"learning_rate": 7.433513390357989e-07,
|
| 555 |
+
"logits/chosen": -1.0875113010406494,
|
| 556 |
+
"logits/rejected": -1.0918848514556885,
|
| 557 |
+
"logps/chosen": -0.4790703356266022,
|
| 558 |
+
"logps/rejected": -0.6097405552864075,
|
| 559 |
+
"loss": 2.164,
|
| 560 |
+
"rewards/accuracies": 0.793749988079071,
|
| 561 |
+
"rewards/chosen": -4.790703773498535,
|
| 562 |
+
"rewards/margins": 1.3067026138305664,
|
| 563 |
+
"rewards/rejected": -6.097405433654785,
|
| 564 |
+
"step": 185
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.41530054644808745,
|
| 568 |
+
"grad_norm": 106.92479621624057,
|
| 569 |
+
"learning_rate": 7.264843015879321e-07,
|
| 570 |
+
"logits/chosen": -1.0997188091278076,
|
| 571 |
+
"logits/rejected": -1.0514479875564575,
|
| 572 |
+
"logps/chosen": -0.467475026845932,
|
| 573 |
+
"logps/rejected": -0.6209039092063904,
|
| 574 |
+
"loss": 2.2151,
|
| 575 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 576 |
+
"rewards/chosen": -4.674751281738281,
|
| 577 |
+
"rewards/margins": 1.5342884063720703,
|
| 578 |
+
"rewards/rejected": -6.209038734436035,
|
| 579 |
+
"step": 190
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.4262295081967213,
|
| 583 |
+
"grad_norm": 79.4429028708432,
|
| 584 |
+
"learning_rate": 7.092864825346266e-07,
|
| 585 |
+
"logits/chosen": -1.12069833278656,
|
| 586 |
+
"logits/rejected": -1.0915439128875732,
|
| 587 |
+
"logps/chosen": -0.5538396239280701,
|
| 588 |
+
"logps/rejected": -0.7245546579360962,
|
| 589 |
+
"loss": 2.2765,
|
| 590 |
+
"rewards/accuracies": 0.737500011920929,
|
| 591 |
+
"rewards/chosen": -5.538395881652832,
|
| 592 |
+
"rewards/margins": 1.7071504592895508,
|
| 593 |
+
"rewards/rejected": -7.245546817779541,
|
| 594 |
+
"step": 195
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 0.4371584699453552,
|
| 598 |
+
"grad_norm": 79.46289075402596,
|
| 599 |
+
"learning_rate": 6.917829993880302e-07,
|
| 600 |
+
"logits/chosen": -1.1029081344604492,
|
| 601 |
+
"logits/rejected": -1.0124149322509766,
|
| 602 |
+
"logps/chosen": -0.5200821161270142,
|
| 603 |
+
"logps/rejected": -0.6488970518112183,
|
| 604 |
+
"loss": 2.1257,
|
| 605 |
+
"rewards/accuracies": 0.762499988079071,
|
| 606 |
+
"rewards/chosen": -5.200821399688721,
|
| 607 |
+
"rewards/margins": 1.288149118423462,
|
| 608 |
+
"rewards/rejected": -6.4889702796936035,
|
| 609 |
+
"step": 200
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.44808743169398907,
|
| 613 |
+
"grad_norm": 67.1583967725435,
|
| 614 |
+
"learning_rate": 6.739994160844309e-07,
|
| 615 |
+
"logits/chosen": -1.0795233249664307,
|
| 616 |
+
"logits/rejected": -1.0885263681411743,
|
| 617 |
+
"logps/chosen": -0.48235517740249634,
|
| 618 |
+
"logps/rejected": -0.6249179244041443,
|
| 619 |
+
"loss": 2.0323,
|
| 620 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 621 |
+
"rewards/chosen": -4.823552131652832,
|
| 622 |
+
"rewards/margins": 1.4256280660629272,
|
| 623 |
+
"rewards/rejected": -6.249179363250732,
|
| 624 |
+
"step": 205
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 0.45901639344262296,
|
| 628 |
+
"grad_norm": 131.31523382011193,
|
| 629 |
+
"learning_rate": 6.559617056479827e-07,
|
| 630 |
+
"logits/chosen": -1.0998046398162842,
|
| 631 |
+
"logits/rejected": -1.0920783281326294,
|
| 632 |
+
"logps/chosen": -0.5168774724006653,
|
| 633 |
+
"logps/rejected": -0.6893444657325745,
|
| 634 |
+
"loss": 2.0968,
|
| 635 |
+
"rewards/accuracies": 0.78125,
|
| 636 |
+
"rewards/chosen": -5.168774127960205,
|
| 637 |
+
"rewards/margins": 1.7246698141098022,
|
| 638 |
+
"rewards/rejected": -6.893445014953613,
|
| 639 |
+
"step": 210
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.46994535519125685,
|
| 643 |
+
"grad_norm": 92.8827942851212,
|
| 644 |
+
"learning_rate": 6.376962122569567e-07,
|
| 645 |
+
"logits/chosen": -1.1027876138687134,
|
| 646 |
+
"logits/rejected": -1.0451091527938843,
|
| 647 |
+
"logps/chosen": -0.5266443490982056,
|
| 648 |
+
"logps/rejected": -0.6980344653129578,
|
| 649 |
+
"loss": 2.2334,
|
| 650 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 651 |
+
"rewards/chosen": -5.266442775726318,
|
| 652 |
+
"rewards/margins": 1.7139012813568115,
|
| 653 |
+
"rewards/rejected": -6.980344295501709,
|
| 654 |
+
"step": 215
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.4808743169398907,
|
| 658 |
+
"grad_norm": 75.78200124794627,
|
| 659 |
+
"learning_rate": 6.192296127679192e-07,
|
| 660 |
+
"logits/chosen": -1.163720965385437,
|
| 661 |
+
"logits/rejected": -1.1323108673095703,
|
| 662 |
+
"logps/chosen": -0.5482050180435181,
|
| 663 |
+
"logps/rejected": -0.6795030832290649,
|
| 664 |
+
"loss": 2.0812,
|
| 665 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 666 |
+
"rewards/chosen": -5.482050895690918,
|
| 667 |
+
"rewards/margins": 1.3129799365997314,
|
| 668 |
+
"rewards/rejected": -6.7950310707092285,
|
| 669 |
+
"step": 220
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.4918032786885246,
|
| 673 |
+
"grad_norm": 68.50625090212893,
|
| 674 |
+
"learning_rate": 6.005888777540319e-07,
|
| 675 |
+
"logits/chosen": -1.1991808414459229,
|
| 676 |
+
"logits/rejected": -1.1517974138259888,
|
| 677 |
+
"logps/chosen": -0.5413341522216797,
|
| 678 |
+
"logps/rejected": -0.6963613033294678,
|
| 679 |
+
"loss": 2.1219,
|
| 680 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 681 |
+
"rewards/chosen": -5.413341522216797,
|
| 682 |
+
"rewards/margins": 1.5502712726593018,
|
| 683 |
+
"rewards/rejected": -6.963613033294678,
|
| 684 |
+
"step": 225
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 0.5027322404371585,
|
| 688 |
+
"grad_norm": 73.44567710043167,
|
| 689 |
+
"learning_rate": 5.818012321143773e-07,
|
| 690 |
+
"logits/chosen": -1.0955681800842285,
|
| 691 |
+
"logits/rejected": -1.0907937288284302,
|
| 692 |
+
"logps/chosen": -0.5347357988357544,
|
| 693 |
+
"logps/rejected": -0.7202361226081848,
|
| 694 |
+
"loss": 2.0511,
|
| 695 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 696 |
+
"rewards/chosen": -5.347357749938965,
|
| 697 |
+
"rewards/margins": 1.855002760887146,
|
| 698 |
+
"rewards/rejected": -7.2023606300354,
|
| 699 |
+
"step": 230
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.5136612021857924,
|
| 703 |
+
"grad_norm": 84.7486173764228,
|
| 704 |
+
"learning_rate": 5.628941153118388e-07,
|
| 705 |
+
"logits/chosen": -1.113163709640503,
|
| 706 |
+
"logits/rejected": -1.0674149990081787,
|
| 707 |
+
"logps/chosen": -0.534842312335968,
|
| 708 |
+
"logps/rejected": -0.6825965046882629,
|
| 709 |
+
"loss": 2.0329,
|
| 710 |
+
"rewards/accuracies": 0.78125,
|
| 711 |
+
"rewards/chosen": -5.348422527313232,
|
| 712 |
+
"rewards/margins": 1.4775424003601074,
|
| 713 |
+
"rewards/rejected": -6.825964450836182,
|
| 714 |
+
"step": 235
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 0.5245901639344263,
|
| 718 |
+
"grad_norm": 70.6295555021999,
|
| 719 |
+
"learning_rate": 5.438951412976098e-07,
|
| 720 |
+
"logits/chosen": -1.1475393772125244,
|
| 721 |
+
"logits/rejected": -1.1534240245819092,
|
| 722 |
+
"logps/chosen": -0.5095857381820679,
|
| 723 |
+
"logps/rejected": -0.6976131200790405,
|
| 724 |
+
"loss": 1.8823,
|
| 725 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 726 |
+
"rewards/chosen": -5.095856666564941,
|
| 727 |
+
"rewards/margins": 1.8802751302719116,
|
| 728 |
+
"rewards/rejected": -6.976132392883301,
|
| 729 |
+
"step": 240
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.5355191256830601,
|
| 733 |
+
"grad_norm": 69.47924677139383,
|
| 734 |
+
"learning_rate": 5.248320581808619e-07,
|
| 735 |
+
"logits/chosen": -1.0735324621200562,
|
| 736 |
+
"logits/rejected": -1.0195956230163574,
|
| 737 |
+
"logps/chosen": -0.5255709290504456,
|
| 738 |
+
"logps/rejected": -0.7123461961746216,
|
| 739 |
+
"loss": 1.9839,
|
| 740 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 741 |
+
"rewards/chosen": -5.255709648132324,
|
| 742 |
+
"rewards/margins": 1.8677523136138916,
|
| 743 |
+
"rewards/rejected": -7.123461723327637,
|
| 744 |
+
"step": 245
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.546448087431694,
|
| 748 |
+
"grad_norm": 66.46597890028497,
|
| 749 |
+
"learning_rate": 5.057327077024744e-07,
|
| 750 |
+
"logits/chosen": -1.1226236820220947,
|
| 751 |
+
"logits/rejected": -1.0827569961547852,
|
| 752 |
+
"logps/chosen": -0.5162156224250793,
|
| 753 |
+
"logps/rejected": -0.6480467319488525,
|
| 754 |
+
"loss": 2.0699,
|
| 755 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 756 |
+
"rewards/chosen": -5.162156105041504,
|
| 757 |
+
"rewards/margins": 1.3183108568191528,
|
| 758 |
+
"rewards/rejected": -6.480466365814209,
|
| 759 |
+
"step": 250
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.5573770491803278,
|
| 763 |
+
"grad_norm": 76.19586118186004,
|
| 764 |
+
"learning_rate": 4.866249845720132e-07,
|
| 765 |
+
"logits/chosen": -1.107912540435791,
|
| 766 |
+
"logits/rejected": -1.071297287940979,
|
| 767 |
+
"logps/chosen": -0.5693866014480591,
|
| 768 |
+
"logps/rejected": -0.735679566860199,
|
| 769 |
+
"loss": 1.9599,
|
| 770 |
+
"rewards/accuracies": 0.768750011920929,
|
| 771 |
+
"rewards/chosen": -5.693865776062012,
|
| 772 |
+
"rewards/margins": 1.6629297733306885,
|
| 773 |
+
"rewards/rejected": -7.356795310974121,
|
| 774 |
+
"step": 255
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 0.5683060109289617,
|
| 778 |
+
"grad_norm": 73.20621943329614,
|
| 779 |
+
"learning_rate": 4.675367957273505e-07,
|
| 780 |
+
"logits/chosen": -1.0760225057601929,
|
| 781 |
+
"logits/rejected": -1.0634446144104004,
|
| 782 |
+
"logps/chosen": -0.5194670557975769,
|
| 783 |
+
"logps/rejected": -0.6913676261901855,
|
| 784 |
+
"loss": 1.9788,
|
| 785 |
+
"rewards/accuracies": 0.800000011920929,
|
| 786 |
+
"rewards/chosen": -5.1946702003479,
|
| 787 |
+
"rewards/margins": 1.7190055847167969,
|
| 788 |
+
"rewards/rejected": -6.9136762619018555,
|
| 789 |
+
"step": 260
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.5792349726775956,
|
| 793 |
+
"grad_norm": 111.34081159607314,
|
| 794 |
+
"learning_rate": 4.4849601957642285e-07,
|
| 795 |
+
"logits/chosen": -1.1172699928283691,
|
| 796 |
+
"logits/rejected": -1.0767419338226318,
|
| 797 |
+
"logps/chosen": -0.5310412645339966,
|
| 798 |
+
"logps/rejected": -0.6969484686851501,
|
| 799 |
+
"loss": 2.0084,
|
| 800 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 801 |
+
"rewards/chosen": -5.310412406921387,
|
| 802 |
+
"rewards/margins": 1.6590715646743774,
|
| 803 |
+
"rewards/rejected": -6.969484806060791,
|
| 804 |
+
"step": 265
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 0.5901639344262295,
|
| 808 |
+
"grad_norm": 79.77425345162416,
|
| 809 |
+
"learning_rate": 4.295304652806592e-07,
|
| 810 |
+
"logits/chosen": -1.1290314197540283,
|
| 811 |
+
"logits/rejected": -1.098478078842163,
|
| 812 |
+
"logps/chosen": -0.5287065505981445,
|
| 813 |
+
"logps/rejected": -0.7144958972930908,
|
| 814 |
+
"loss": 1.8771,
|
| 815 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 816 |
+
"rewards/chosen": -5.2870659828186035,
|
| 817 |
+
"rewards/margins": 1.8578933477401733,
|
| 818 |
+
"rewards/rejected": -7.144959449768066,
|
| 819 |
+
"step": 270
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 0.6010928961748634,
|
| 823 |
+
"grad_norm": 63.79079426666398,
|
| 824 |
+
"learning_rate": 4.106678321395433e-07,
|
| 825 |
+
"logits/chosen": -1.100162148475647,
|
| 826 |
+
"logits/rejected": -1.0285935401916504,
|
| 827 |
+
"logps/chosen": -0.5441292524337769,
|
| 828 |
+
"logps/rejected": -0.6436029672622681,
|
| 829 |
+
"loss": 1.9928,
|
| 830 |
+
"rewards/accuracies": 0.65625,
|
| 831 |
+
"rewards/chosen": -5.4412922859191895,
|
| 832 |
+
"rewards/margins": 0.9947379231452942,
|
| 833 |
+
"rewards/rejected": -6.43602991104126,
|
| 834 |
+
"step": 275
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 0.6120218579234973,
|
| 838 |
+
"grad_norm": 84.61811881052614,
|
| 839 |
+
"learning_rate": 3.9193566913562915e-07,
|
| 840 |
+
"logits/chosen": -1.055173635482788,
|
| 841 |
+
"logits/rejected": -1.0575199127197266,
|
| 842 |
+
"logps/chosen": -0.5374099612236023,
|
| 843 |
+
"logps/rejected": -0.7436326742172241,
|
| 844 |
+
"loss": 1.9752,
|
| 845 |
+
"rewards/accuracies": 0.737500011920929,
|
| 846 |
+
"rewards/chosen": -5.374099254608154,
|
| 847 |
+
"rewards/margins": 2.062227249145508,
|
| 848 |
+
"rewards/rejected": -7.436326503753662,
|
| 849 |
+
"step": 280
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.6229508196721312,
|
| 853 |
+
"grad_norm": 69.51995916576952,
|
| 854 |
+
"learning_rate": 3.7336133469909623e-07,
|
| 855 |
+
"logits/chosen": -1.1819168329238892,
|
| 856 |
+
"logits/rejected": -1.1467748880386353,
|
| 857 |
+
"logps/chosen": -0.5230523347854614,
|
| 858 |
+
"logps/rejected": -0.7193282842636108,
|
| 859 |
+
"loss": 1.923,
|
| 860 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 861 |
+
"rewards/chosen": -5.230523109436035,
|
| 862 |
+
"rewards/margins": 1.9627596139907837,
|
| 863 |
+
"rewards/rejected": -7.1932830810546875,
|
| 864 |
+
"step": 285
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.6338797814207651,
|
| 868 |
+
"grad_norm": 98.48874799948564,
|
| 869 |
+
"learning_rate": 3.549719567506076e-07,
|
| 870 |
+
"logits/chosen": -1.1354302167892456,
|
| 871 |
+
"logits/rejected": -1.0900018215179443,
|
| 872 |
+
"logps/chosen": -0.5486642718315125,
|
| 873 |
+
"logps/rejected": -0.7162230610847473,
|
| 874 |
+
"loss": 1.9529,
|
| 875 |
+
"rewards/accuracies": 0.793749988079071,
|
| 876 |
+
"rewards/chosen": -5.486642360687256,
|
| 877 |
+
"rewards/margins": 1.6755880117416382,
|
| 878 |
+
"rewards/rejected": -7.162230491638184,
|
| 879 |
+
"step": 290
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 0.644808743169399,
|
| 883 |
+
"grad_norm": 66.70134041294685,
|
| 884 |
+
"learning_rate": 3.3679439308082774e-07,
|
| 885 |
+
"logits/chosen": -1.1284725666046143,
|
| 886 |
+
"logits/rejected": -1.126910924911499,
|
| 887 |
+
"logps/chosen": -0.5425226092338562,
|
| 888 |
+
"logps/rejected": -0.7452016472816467,
|
| 889 |
+
"loss": 1.7199,
|
| 890 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 891 |
+
"rewards/chosen": -5.425226211547852,
|
| 892 |
+
"rewards/margins": 2.026790142059326,
|
| 893 |
+
"rewards/rejected": -7.452017307281494,
|
| 894 |
+
"step": 295
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 0.6557377049180327,
|
| 898 |
+
"grad_norm": 71.88498470947849,
|
| 899 |
+
"learning_rate": 3.1885519212446716e-07,
|
| 900 |
+
"logits/chosen": -1.134820580482483,
|
| 901 |
+
"logits/rejected": -1.113407850265503,
|
| 902 |
+
"logps/chosen": -0.5564647316932678,
|
| 903 |
+
"logps/rejected": -0.7468981146812439,
|
| 904 |
+
"loss": 1.8137,
|
| 905 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 906 |
+
"rewards/chosen": -5.564647197723389,
|
| 907 |
+
"rewards/margins": 1.9043340682983398,
|
| 908 |
+
"rewards/rejected": -7.4689812660217285,
|
| 909 |
+
"step": 300
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 0.6666666666666666,
|
| 913 |
+
"grad_norm": 76.90950112576293,
|
| 914 |
+
"learning_rate": 3.0118055418614295e-07,
|
| 915 |
+
"logits/chosen": -1.1550134420394897,
|
| 916 |
+
"logits/rejected": -1.0957601070404053,
|
| 917 |
+
"logps/chosen": -0.5463107228279114,
|
| 918 |
+
"logps/rejected": -0.7333046793937683,
|
| 919 |
+
"loss": 1.942,
|
| 920 |
+
"rewards/accuracies": 0.8125,
|
| 921 |
+
"rewards/chosen": -5.463107109069824,
|
| 922 |
+
"rewards/margins": 1.8699395656585693,
|
| 923 |
+
"rewards/rejected": -7.333047389984131,
|
| 924 |
+
"step": 305
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 0.6775956284153005,
|
| 928 |
+
"grad_norm": 107.13681953131237,
|
| 929 |
+
"learning_rate": 2.83796293174686e-07,
|
| 930 |
+
"logits/chosen": -1.0895421504974365,
|
| 931 |
+
"logits/rejected": -1.0931763648986816,
|
| 932 |
+
"logps/chosen": -0.5538777112960815,
|
| 933 |
+
"logps/rejected": -0.7688311338424683,
|
| 934 |
+
"loss": 2.0538,
|
| 935 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 936 |
+
"rewards/chosen": -5.538776874542236,
|
| 937 |
+
"rewards/margins": 2.149533748626709,
|
| 938 |
+
"rewards/rejected": -7.688311576843262,
|
| 939 |
+
"step": 310
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 0.6885245901639344,
|
| 943 |
+
"grad_norm": 66.34265535656564,
|
| 944 |
+
"learning_rate": 2.6672779890178046e-07,
|
| 945 |
+
"logits/chosen": -1.1496803760528564,
|
| 946 |
+
"logits/rejected": -1.1489367485046387,
|
| 947 |
+
"logps/chosen": -0.5764937400817871,
|
| 948 |
+
"logps/rejected": -0.706534206867218,
|
| 949 |
+
"loss": 1.9103,
|
| 950 |
+
"rewards/accuracies": 0.737500011920929,
|
| 951 |
+
"rewards/chosen": -5.764937400817871,
|
| 952 |
+
"rewards/margins": 1.3004045486450195,
|
| 953 |
+
"rewards/rejected": -7.065341949462891,
|
| 954 |
+
"step": 315
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.6994535519125683,
|
| 958 |
+
"grad_norm": 56.65585571674297,
|
| 959 |
+
"learning_rate": 2.500000000000001e-07,
|
| 960 |
+
"logits/chosen": -1.204594612121582,
|
| 961 |
+
"logits/rejected": -1.1554490327835083,
|
| 962 |
+
"logps/chosen": -0.556768000125885,
|
| 963 |
+
"logps/rejected": -0.7628398537635803,
|
| 964 |
+
"loss": 1.8878,
|
| 965 |
+
"rewards/accuracies": 0.824999988079071,
|
| 966 |
+
"rewards/chosen": -5.567679405212402,
|
| 967 |
+
"rewards/margins": 2.060718536376953,
|
| 968 |
+
"rewards/rejected": -7.628398895263672,
|
| 969 |
+
"step": 320
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.7103825136612022,
|
| 973 |
+
"grad_norm": 82.81929805474688,
|
| 974 |
+
"learning_rate": 2.3363732751439923e-07,
|
| 975 |
+
"logits/chosen": -1.1415125131607056,
|
| 976 |
+
"logits/rejected": -1.125905990600586,
|
| 977 |
+
"logps/chosen": -0.542820930480957,
|
| 978 |
+
"logps/rejected": -0.7379633188247681,
|
| 979 |
+
"loss": 1.8941,
|
| 980 |
+
"rewards/accuracies": 0.793749988079071,
|
| 981 |
+
"rewards/chosen": -5.42820930480957,
|
| 982 |
+
"rewards/margins": 1.9514230489730835,
|
| 983 |
+
"rewards/rejected": -7.379631996154785,
|
| 984 |
+
"step": 325
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 0.7213114754098361,
|
| 988 |
+
"grad_norm": 71.70622872848305,
|
| 989 |
+
"learning_rate": 2.1766367922083283e-07,
|
| 990 |
+
"logits/chosen": -1.0906676054000854,
|
| 991 |
+
"logits/rejected": -1.0594781637191772,
|
| 992 |
+
"logps/chosen": -0.5101396441459656,
|
| 993 |
+
"logps/rejected": -0.7594174146652222,
|
| 994 |
+
"loss": 1.8536,
|
| 995 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 996 |
+
"rewards/chosen": -5.101396083831787,
|
| 997 |
+
"rewards/margins": 2.4927775859832764,
|
| 998 |
+
"rewards/rejected": -7.594174385070801,
|
| 999 |
+
"step": 330
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 0.73224043715847,
|
| 1003 |
+
"grad_norm": 80.29511185962221,
|
| 1004 |
+
"learning_rate": 2.021023847231202e-07,
|
| 1005 |
+
"logits/chosen": -1.0921766757965088,
|
| 1006 |
+
"logits/rejected": -1.0546457767486572,
|
| 1007 |
+
"logps/chosen": -0.5786954760551453,
|
| 1008 |
+
"logps/rejected": -0.775272011756897,
|
| 1009 |
+
"loss": 1.859,
|
| 1010 |
+
"rewards/accuracies": 0.824999988079071,
|
| 1011 |
+
"rewards/chosen": -5.7869553565979,
|
| 1012 |
+
"rewards/margins": 1.9657647609710693,
|
| 1013 |
+
"rewards/rejected": -7.752719879150391,
|
| 1014 |
+
"step": 335
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 0.7431693989071039,
|
| 1018 |
+
"grad_norm": 79.83892973093212,
|
| 1019 |
+
"learning_rate": 1.869761713800254e-07,
|
| 1020 |
+
"logits/chosen": -1.1134289503097534,
|
| 1021 |
+
"logits/rejected": -1.0689969062805176,
|
| 1022 |
+
"logps/chosen": -0.5616486668586731,
|
| 1023 |
+
"logps/rejected": -0.7319185137748718,
|
| 1024 |
+
"loss": 1.9172,
|
| 1025 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1026 |
+
"rewards/chosen": -5.6164870262146,
|
| 1027 |
+
"rewards/margins": 1.7026973962783813,
|
| 1028 |
+
"rewards/rejected": -7.319184303283691,
|
| 1029 |
+
"step": 340
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 0.7540983606557377,
|
| 1033 |
+
"grad_norm": 79.67263653958703,
|
| 1034 |
+
"learning_rate": 1.7230713111182164e-07,
|
| 1035 |
+
"logits/chosen": -1.1623106002807617,
|
| 1036 |
+
"logits/rejected": -1.1600934267044067,
|
| 1037 |
+
"logps/chosen": -0.5594975352287292,
|
| 1038 |
+
"logps/rejected": -0.7684382200241089,
|
| 1039 |
+
"loss": 1.9557,
|
| 1040 |
+
"rewards/accuracies": 0.8125,
|
| 1041 |
+
"rewards/chosen": -5.594975471496582,
|
| 1042 |
+
"rewards/margins": 2.089407444000244,
|
| 1043 |
+
"rewards/rejected": -7.684383392333984,
|
| 1044 |
+
"step": 345
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 0.7650273224043715,
|
| 1048 |
+
"grad_norm": 90.3235756484309,
|
| 1049 |
+
"learning_rate": 1.5811668813491696e-07,
|
| 1050 |
+
"logits/chosen": -1.1584365367889404,
|
| 1051 |
+
"logits/rejected": -1.1381675004959106,
|
| 1052 |
+
"logps/chosen": -0.5266227126121521,
|
| 1053 |
+
"logps/rejected": -0.6895478367805481,
|
| 1054 |
+
"loss": 1.8798,
|
| 1055 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1056 |
+
"rewards/chosen": -5.266226768493652,
|
| 1057 |
+
"rewards/margins": 1.6292508840560913,
|
| 1058 |
+
"rewards/rejected": -6.895478248596191,
|
| 1059 |
+
"step": 350
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.7759562841530054,
|
| 1063 |
+
"grad_norm": 65.15244365555583,
|
| 1064 |
+
"learning_rate": 1.4442556767166369e-07,
|
| 1065 |
+
"logits/chosen": -1.1325640678405762,
|
| 1066 |
+
"logits/rejected": -1.10225510597229,
|
| 1067 |
+
"logps/chosen": -0.539125382900238,
|
| 1068 |
+
"logps/rejected": -0.7123144865036011,
|
| 1069 |
+
"loss": 1.8691,
|
| 1070 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1071 |
+
"rewards/chosen": -5.391254425048828,
|
| 1072 |
+
"rewards/margins": 1.731890082359314,
|
| 1073 |
+
"rewards/rejected": -7.123144626617432,
|
| 1074 |
+
"step": 355
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.7868852459016393,
|
| 1078 |
+
"grad_norm": 74.4651421428626,
|
| 1079 |
+
"learning_rate": 1.312537656810549e-07,
|
| 1080 |
+
"logits/chosen": -1.0851835012435913,
|
| 1081 |
+
"logits/rejected": -1.0858454704284668,
|
| 1082 |
+
"logps/chosen": -0.5476486682891846,
|
| 1083 |
+
"logps/rejected": -0.730248749256134,
|
| 1084 |
+
"loss": 1.9346,
|
| 1085 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1086 |
+
"rewards/chosen": -5.476487159729004,
|
| 1087 |
+
"rewards/margins": 1.8260005712509155,
|
| 1088 |
+
"rewards/rejected": -7.302487373352051,
|
| 1089 |
+
"step": 360
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 0.7978142076502732,
|
| 1093 |
+
"grad_norm": 153.86057279690542,
|
| 1094 |
+
"learning_rate": 1.1862051965451214e-07,
|
| 1095 |
+
"logits/chosen": -1.166272521018982,
|
| 1096 |
+
"logits/rejected": -1.165235161781311,
|
| 1097 |
+
"logps/chosen": -0.5564864873886108,
|
| 1098 |
+
"logps/rejected": -0.749941349029541,
|
| 1099 |
+
"loss": 1.8628,
|
| 1100 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 1101 |
+
"rewards/chosen": -5.5648651123046875,
|
| 1102 |
+
"rewards/margins": 1.934548020362854,
|
| 1103 |
+
"rewards/rejected": -7.49941349029541,
|
| 1104 |
+
"step": 365
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 0.8087431693989071,
|
| 1108 |
+
"grad_norm": 77.62021036106599,
|
| 1109 |
+
"learning_rate": 1.0654428051942138e-07,
|
| 1110 |
+
"logits/chosen": -1.1702772378921509,
|
| 1111 |
+
"logits/rejected": -1.1325637102127075,
|
| 1112 |
+
"logps/chosen": -0.5678818225860596,
|
| 1113 |
+
"logps/rejected": -0.7951611876487732,
|
| 1114 |
+
"loss": 1.9704,
|
| 1115 |
+
"rewards/accuracies": 0.78125,
|
| 1116 |
+
"rewards/chosen": -5.678818225860596,
|
| 1117 |
+
"rewards/margins": 2.272794246673584,
|
| 1118 |
+
"rewards/rejected": -7.951612949371338,
|
| 1119 |
+
"step": 370
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 0.819672131147541,
|
| 1123 |
+
"grad_norm": 63.25811531390013,
|
| 1124 |
+
"learning_rate": 9.504268569144763e-08,
|
| 1125 |
+
"logits/chosen": -1.2080334424972534,
|
| 1126 |
+
"logits/rejected": -1.1406437158584595,
|
| 1127 |
+
"logps/chosen": -0.5438567399978638,
|
| 1128 |
+
"logps/rejected": -0.7369149923324585,
|
| 1129 |
+
"loss": 1.892,
|
| 1130 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1131 |
+
"rewards/chosen": -5.4385666847229,
|
| 1132 |
+
"rewards/margins": 1.9305833578109741,
|
| 1133 |
+
"rewards/rejected": -7.369150638580322,
|
| 1134 |
+
"step": 375
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 0.8306010928961749,
|
| 1138 |
+
"grad_norm": 62.759245925109596,
|
| 1139 |
+
"learning_rate": 8.413253331499049e-08,
|
| 1140 |
+
"logits/chosen": -1.088577389717102,
|
| 1141 |
+
"logits/rejected": -1.1100099086761475,
|
| 1142 |
+
"logps/chosen": -0.56348717212677,
|
| 1143 |
+
"logps/rejected": -0.7516815066337585,
|
| 1144 |
+
"loss": 1.8378,
|
| 1145 |
+
"rewards/accuracies": 0.862500011920929,
|
| 1146 |
+
"rewards/chosen": -5.634871959686279,
|
| 1147 |
+
"rewards/margins": 1.8819434642791748,
|
| 1148 |
+
"rewards/rejected": -7.516815185546875,
|
| 1149 |
+
"step": 380
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 0.8415300546448088,
|
| 1153 |
+
"grad_norm": 66.42055041642078,
|
| 1154 |
+
"learning_rate": 7.382975772939865e-08,
|
| 1155 |
+
"logits/chosen": -1.182722806930542,
|
| 1156 |
+
"logits/rejected": -1.1655415296554565,
|
| 1157 |
+
"logps/chosen": -0.599307656288147,
|
| 1158 |
+
"logps/rejected": -0.7887434959411621,
|
| 1159 |
+
"loss": 1.9674,
|
| 1160 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1161 |
+
"rewards/chosen": -5.993075847625732,
|
| 1162 |
+
"rewards/margins": 1.8943592309951782,
|
| 1163 |
+
"rewards/rejected": -7.887434959411621,
|
| 1164 |
+
"step": 385
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.8524590163934426,
|
| 1168 |
+
"grad_norm": 104.21611140368309,
|
| 1169 |
+
"learning_rate": 6.414940619677734e-08,
|
| 1170 |
+
"logits/chosen": -1.171216607093811,
|
| 1171 |
+
"logits/rejected": -1.1473820209503174,
|
| 1172 |
+
"logps/chosen": -0.5480982065200806,
|
| 1173 |
+
"logps/rejected": -0.76609867811203,
|
| 1174 |
+
"loss": 1.9188,
|
| 1175 |
+
"rewards/accuracies": 0.831250011920929,
|
| 1176 |
+
"rewards/chosen": -5.480981349945068,
|
| 1177 |
+
"rewards/margins": 2.1800050735473633,
|
| 1178 |
+
"rewards/rejected": -7.660986423492432,
|
| 1179 |
+
"step": 390
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.8633879781420765,
|
| 1183 |
+
"grad_norm": 78.27102698318478,
|
| 1184 |
+
"learning_rate": 5.5105616925376296e-08,
|
| 1185 |
+
"logits/chosen": -1.1506216526031494,
|
| 1186 |
+
"logits/rejected": -1.1313217878341675,
|
| 1187 |
+
"logps/chosen": -0.5556925535202026,
|
| 1188 |
+
"logps/rejected": -0.706631064414978,
|
| 1189 |
+
"loss": 1.8449,
|
| 1190 |
+
"rewards/accuracies": 0.768750011920929,
|
| 1191 |
+
"rewards/chosen": -5.556925296783447,
|
| 1192 |
+
"rewards/margins": 1.509385347366333,
|
| 1193 |
+
"rewards/rejected": -7.066309928894043,
|
| 1194 |
+
"step": 395
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 0.8743169398907104,
|
| 1198 |
+
"grad_norm": 76.30561413947063,
|
| 1199 |
+
"learning_rate": 4.6711598420656976e-08,
|
| 1200 |
+
"logits/chosen": -1.0807750225067139,
|
| 1201 |
+
"logits/rejected": -1.0536285638809204,
|
| 1202 |
+
"logps/chosen": -0.5846288800239563,
|
| 1203 |
+
"logps/rejected": -0.7902191281318665,
|
| 1204 |
+
"loss": 1.8354,
|
| 1205 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 1206 |
+
"rewards/chosen": -5.846289157867432,
|
| 1207 |
+
"rewards/margins": 2.0559017658233643,
|
| 1208 |
+
"rewards/rejected": -7.902191162109375,
|
| 1209 |
+
"step": 400
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 0.8743169398907104,
|
| 1213 |
+
"eval_logits/chosen": -1.3447647094726562,
|
| 1214 |
+
"eval_logits/rejected": -1.3069871664047241,
|
| 1215 |
+
"eval_logps/chosen": -0.5437440872192383,
|
| 1216 |
+
"eval_logps/rejected": -0.7271364331245422,
|
| 1217 |
+
"eval_loss": 1.881577491760254,
|
| 1218 |
+
"eval_rewards/accuracies": 0.7710843086242676,
|
| 1219 |
+
"eval_rewards/chosen": -5.437441349029541,
|
| 1220 |
+
"eval_rewards/margins": 1.83392333984375,
|
| 1221 |
+
"eval_rewards/rejected": -7.271363735198975,
|
| 1222 |
+
"eval_runtime": 37.6564,
|
| 1223 |
+
"eval_samples_per_second": 35.001,
|
| 1224 |
+
"eval_steps_per_second": 2.204,
|
| 1225 |
+
"step": 400
|
| 1226 |
+
},
|
| 1227 |
+
{
|
| 1228 |
+
"epoch": 0.8852459016393442,
|
| 1229 |
+
"grad_norm": 66.16858739679557,
|
| 1230 |
+
"learning_rate": 3.897961019419516e-08,
|
| 1231 |
+
"logits/chosen": -1.1171514987945557,
|
| 1232 |
+
"logits/rejected": -1.051194190979004,
|
| 1233 |
+
"logps/chosen": -0.5250142812728882,
|
| 1234 |
+
"logps/rejected": -0.6711951494216919,
|
| 1235 |
+
"loss": 1.793,
|
| 1236 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1237 |
+
"rewards/chosen": -5.250143527984619,
|
| 1238 |
+
"rewards/margins": 1.461808204650879,
|
| 1239 |
+
"rewards/rejected": -6.71195125579834,
|
| 1240 |
+
"step": 405
|
| 1241 |
+
},
|
| 1242 |
+
{
|
| 1243 |
+
"epoch": 0.8961748633879781,
|
| 1244 |
+
"grad_norm": 77.0310414767933,
|
| 1245 |
+
"learning_rate": 3.192094485859526e-08,
|
| 1246 |
+
"logits/chosen": -1.1234816312789917,
|
| 1247 |
+
"logits/rejected": -1.144008994102478,
|
| 1248 |
+
"logps/chosen": -0.5635503530502319,
|
| 1249 |
+
"logps/rejected": -0.7934846878051758,
|
| 1250 |
+
"loss": 1.8728,
|
| 1251 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1252 |
+
"rewards/chosen": -5.63550329208374,
|
| 1253 |
+
"rewards/margins": 2.2993433475494385,
|
| 1254 |
+
"rewards/rejected": -7.934847354888916,
|
| 1255 |
+
"step": 410
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 0.907103825136612,
|
| 1259 |
+
"grad_norm": 62.5404158274799,
|
| 1260 |
+
"learning_rate": 2.5545911634565265e-08,
|
| 1261 |
+
"logits/chosen": -1.1487548351287842,
|
| 1262 |
+
"logits/rejected": -1.1474642753601074,
|
| 1263 |
+
"logps/chosen": -0.570233941078186,
|
| 1264 |
+
"logps/rejected": -0.7942467927932739,
|
| 1265 |
+
"loss": 1.8971,
|
| 1266 |
+
"rewards/accuracies": 0.824999988079071,
|
| 1267 |
+
"rewards/chosen": -5.702339172363281,
|
| 1268 |
+
"rewards/margins": 2.2401280403137207,
|
| 1269 |
+
"rewards/rejected": -7.94246768951416,
|
| 1270 |
+
"step": 415
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 0.9180327868852459,
|
| 1274 |
+
"grad_norm": 81.2075081960254,
|
| 1275 |
+
"learning_rate": 1.9863821294241522e-08,
|
| 1276 |
+
"logits/chosen": -1.150827407836914,
|
| 1277 |
+
"logits/rejected": -1.1209652423858643,
|
| 1278 |
+
"logps/chosen": -0.5287235379219055,
|
| 1279 |
+
"logps/rejected": -0.7498638033866882,
|
| 1280 |
+
"loss": 1.7791,
|
| 1281 |
+
"rewards/accuracies": 0.8125,
|
| 1282 |
+
"rewards/chosen": -5.287235260009766,
|
| 1283 |
+
"rewards/margins": 2.211402177810669,
|
| 1284 |
+
"rewards/rejected": -7.498638153076172,
|
| 1285 |
+
"step": 420
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 0.9289617486338798,
|
| 1289 |
+
"grad_norm": 74.25861489169881,
|
| 1290 |
+
"learning_rate": 1.4882972562753615e-08,
|
| 1291 |
+
"logits/chosen": -1.1429412364959717,
|
| 1292 |
+
"logits/rejected": -1.1335225105285645,
|
| 1293 |
+
"logps/chosen": -0.5883482098579407,
|
| 1294 |
+
"logps/rejected": -0.8031044006347656,
|
| 1295 |
+
"loss": 1.918,
|
| 1296 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1297 |
+
"rewards/chosen": -5.883482933044434,
|
| 1298 |
+
"rewards/margins": 2.1475605964660645,
|
| 1299 |
+
"rewards/rejected": -8.03104305267334,
|
| 1300 |
+
"step": 425
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"epoch": 0.9398907103825137,
|
| 1304 |
+
"grad_norm": 90.08434920268098,
|
| 1305 |
+
"learning_rate": 1.0610639997888915e-08,
|
| 1306 |
+
"logits/chosen": -1.0862069129943848,
|
| 1307 |
+
"logits/rejected": -1.086482286453247,
|
| 1308 |
+
"logps/chosen": -0.5331538915634155,
|
| 1309 |
+
"logps/rejected": -0.73698890209198,
|
| 1310 |
+
"loss": 1.7152,
|
| 1311 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 1312 |
+
"rewards/chosen": -5.331539154052734,
|
| 1313 |
+
"rewards/margins": 2.0383496284484863,
|
| 1314 |
+
"rewards/rejected": -7.3698883056640625,
|
| 1315 |
+
"step": 430
|
| 1316 |
+
},
|
| 1317 |
+
{
|
| 1318 |
+
"epoch": 0.9508196721311475,
|
| 1319 |
+
"grad_norm": 88.78145235758741,
|
| 1320 |
+
"learning_rate": 7.053063365559997e-09,
|
| 1321 |
+
"logits/chosen": -1.1540377140045166,
|
| 1322 |
+
"logits/rejected": -1.1776275634765625,
|
| 1323 |
+
"logps/chosen": -0.5493609309196472,
|
| 1324 |
+
"logps/rejected": -0.7795323729515076,
|
| 1325 |
+
"loss": 1.8391,
|
| 1326 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1327 |
+
"rewards/chosen": -5.493609428405762,
|
| 1328 |
+
"rewards/margins": 2.3017146587371826,
|
| 1329 |
+
"rewards/rejected": -7.795323848724365,
|
| 1330 |
+
"step": 435
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"epoch": 0.9617486338797814,
|
| 1334 |
+
"grad_norm": 76.77833268976481,
|
| 1335 |
+
"learning_rate": 4.215438526591064e-09,
|
| 1336 |
+
"logits/chosen": -1.091371774673462,
|
| 1337 |
+
"logits/rejected": -1.0515260696411133,
|
| 1338 |
+
"logps/chosen": -0.5720237493515015,
|
| 1339 |
+
"logps/rejected": -0.7283024787902832,
|
| 1340 |
+
"loss": 1.8312,
|
| 1341 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1342 |
+
"rewards/chosen": -5.720236778259277,
|
| 1343 |
+
"rewards/margins": 1.5627872943878174,
|
| 1344 |
+
"rewards/rejected": -7.283023834228516,
|
| 1345 |
+
"step": 440
|
| 1346 |
+
},
|
| 1347 |
+
{
|
| 1348 |
+
"epoch": 0.9726775956284153,
|
| 1349 |
+
"grad_norm": 85.67571762449911,
|
| 1350 |
+
"learning_rate": 2.1019098481337426e-09,
|
| 1351 |
+
"logits/chosen": -1.1658837795257568,
|
| 1352 |
+
"logits/rejected": -1.1402862071990967,
|
| 1353 |
+
"logps/chosen": -0.5455365180969238,
|
| 1354 |
+
"logps/rejected": -0.7272461652755737,
|
| 1355 |
+
"loss": 1.8014,
|
| 1356 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 1357 |
+
"rewards/chosen": -5.4553656578063965,
|
| 1358 |
+
"rewards/margins": 1.8170963525772095,
|
| 1359 |
+
"rewards/rejected": -7.272461891174316,
|
| 1360 |
+
"step": 445
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 0.9836065573770492,
|
| 1364 |
+
"grad_norm": 87.49876973882085,
|
| 1365 |
+
"learning_rate": 7.155641507955445e-10,
|
| 1366 |
+
"logits/chosen": -1.0767357349395752,
|
| 1367 |
+
"logits/rejected": -1.0696969032287598,
|
| 1368 |
+
"logps/chosen": -0.5918444395065308,
|
| 1369 |
+
"logps/rejected": -0.7698336839675903,
|
| 1370 |
+
"loss": 1.9461,
|
| 1371 |
+
"rewards/accuracies": 0.78125,
|
| 1372 |
+
"rewards/chosen": -5.918444633483887,
|
| 1373 |
+
"rewards/margins": 1.7798923254013062,
|
| 1374 |
+
"rewards/rejected": -7.698336601257324,
|
| 1375 |
+
"step": 450
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 0.994535519125683,
|
| 1379 |
+
"grad_norm": 83.15276885308347,
|
| 1380 |
+
"learning_rate": 5.842620032053824e-11,
|
| 1381 |
+
"logits/chosen": -1.0845460891723633,
|
| 1382 |
+
"logits/rejected": -1.0800822973251343,
|
| 1383 |
+
"logps/chosen": -0.5885196328163147,
|
| 1384 |
+
"logps/rejected": -0.7454328536987305,
|
| 1385 |
+
"loss": 1.9848,
|
| 1386 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 1387 |
+
"rewards/chosen": -5.885196208953857,
|
| 1388 |
+
"rewards/margins": 1.5691322088241577,
|
| 1389 |
+
"rewards/rejected": -7.4543280601501465,
|
| 1390 |
+
"step": 455
|
| 1391 |
+
},
|
| 1392 |
+
{
|
| 1393 |
+
"epoch": 0.9989071038251366,
|
| 1394 |
+
"step": 457,
|
| 1395 |
+
"total_flos": 0.0,
|
| 1396 |
+
"train_loss": 2.2740620700744363,
|
| 1397 |
+
"train_runtime": 5957.0658,
|
| 1398 |
+
"train_samples_per_second": 9.83,
|
| 1399 |
+
"train_steps_per_second": 0.077
|
| 1400 |
+
}
|
| 1401 |
+
],
|
| 1402 |
+
"logging_steps": 5,
|
| 1403 |
+
"max_steps": 457,
|
| 1404 |
+
"num_input_tokens_seen": 0,
|
| 1405 |
+
"num_train_epochs": 1,
|
| 1406 |
+
"save_steps": 1000000,
|
| 1407 |
+
"stateful_callbacks": {
|
| 1408 |
+
"TrainerControl": {
|
| 1409 |
+
"args": {
|
| 1410 |
+
"should_epoch_stop": false,
|
| 1411 |
+
"should_evaluate": false,
|
| 1412 |
+
"should_log": false,
|
| 1413 |
+
"should_save": true,
|
| 1414 |
+
"should_training_stop": true
|
| 1415 |
+
},
|
| 1416 |
+
"attributes": {}
|
| 1417 |
+
}
|
| 1418 |
+
},
|
| 1419 |
+
"total_flos": 0.0,
|
| 1420 |
+
"train_batch_size": 2,
|
| 1421 |
+
"trial_name": null,
|
| 1422 |
+
"trial_params": null
|
| 1423 |
+
}
|