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_-cpo-10-0_4-9e-7-1_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_-cpo-10-0_4-9e-7-1_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: 1.6787
|
| 22 |
+
- Rewards/chosen: -21.0507
|
| 23 |
+
- Rewards/rejected: -28.1366
|
| 24 |
+
- Rewards/accuracies: 0.8373
|
| 25 |
+
- Rewards/margins: 7.0859
|
| 26 |
+
- Logps/rejected: -2.8137
|
| 27 |
+
- Logps/chosen: -2.1051
|
| 28 |
+
- Logits/rejected: -1.2389
|
| 29 |
+
- Logits/chosen: -1.2835
|
| 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: 9e-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: 1.0
|
| 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 |
+
| 1.6588 | 0.8743 | 400 | 1.6787 | -21.0507 | -28.1366 | 0.8373 | 7.0859 | -2.8137 | -2.1051 | -1.2389 | -1.2835 |
|
| 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": 2.442784007842483,
|
| 5 |
+
"train_runtime": 5988.2154,
|
| 6 |
+
"train_samples": 58558,
|
| 7 |
+
"train_samples_per_second": 9.779,
|
| 8 |
+
"train_steps_per_second": 0.076
|
| 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": 2.442784007842483,
|
| 5 |
+
"train_runtime": 5988.2154,
|
| 6 |
+
"train_samples": 58558,
|
| 7 |
+
"train_samples_per_second": 9.779,
|
| 8 |
+
"train_steps_per_second": 0.076
|
| 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": 55.38540565612871,
|
| 14 |
+
"learning_rate": 9.782608695652174e-08,
|
| 15 |
+
"logits/chosen": -1.0178581476211548,
|
| 16 |
+
"logits/rejected": -1.0123393535614014,
|
| 17 |
+
"logps/chosen": -0.28073588013648987,
|
| 18 |
+
"logps/rejected": -0.2859407067298889,
|
| 19 |
+
"loss": 3.9566,
|
| 20 |
+
"rewards/accuracies": 0.53125,
|
| 21 |
+
"rewards/chosen": -2.807358503341675,
|
| 22 |
+
"rewards/margins": 0.05204860121011734,
|
| 23 |
+
"rewards/rejected": -2.8594069480895996,
|
| 24 |
+
"step": 5
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.02185792349726776,
|
| 28 |
+
"grad_norm": 72.60910128873313,
|
| 29 |
+
"learning_rate": 1.9565217391304347e-07,
|
| 30 |
+
"logits/chosen": -1.0586049556732178,
|
| 31 |
+
"logits/rejected": -1.0074703693389893,
|
| 32 |
+
"logps/chosen": -0.25731509923934937,
|
| 33 |
+
"logps/rejected": -0.27178820967674255,
|
| 34 |
+
"loss": 3.9165,
|
| 35 |
+
"rewards/accuracies": 0.543749988079071,
|
| 36 |
+
"rewards/chosen": -2.573150873184204,
|
| 37 |
+
"rewards/margins": 0.1447313129901886,
|
| 38 |
+
"rewards/rejected": -2.7178821563720703,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.03278688524590164,
|
| 43 |
+
"grad_norm": 61.335144114494604,
|
| 44 |
+
"learning_rate": 2.9347826086956523e-07,
|
| 45 |
+
"logits/chosen": -1.007138967514038,
|
| 46 |
+
"logits/rejected": -0.9607247114181519,
|
| 47 |
+
"logps/chosen": -0.2671090066432953,
|
| 48 |
+
"logps/rejected": -0.2734394967556,
|
| 49 |
+
"loss": 3.9076,
|
| 50 |
+
"rewards/accuracies": 0.512499988079071,
|
| 51 |
+
"rewards/chosen": -2.6710901260375977,
|
| 52 |
+
"rewards/margins": 0.06330505013465881,
|
| 53 |
+
"rewards/rejected": -2.7343950271606445,
|
| 54 |
+
"step": 15
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.04371584699453552,
|
| 58 |
+
"grad_norm": 72.83718237358336,
|
| 59 |
+
"learning_rate": 3.9130434782608694e-07,
|
| 60 |
+
"logits/chosen": -0.9411917924880981,
|
| 61 |
+
"logits/rejected": -0.8917375802993774,
|
| 62 |
+
"logps/chosen": -0.27226755023002625,
|
| 63 |
+
"logps/rejected": -0.28448182344436646,
|
| 64 |
+
"loss": 3.9517,
|
| 65 |
+
"rewards/accuracies": 0.5625,
|
| 66 |
+
"rewards/chosen": -2.7226755619049072,
|
| 67 |
+
"rewards/margins": 0.12214270979166031,
|
| 68 |
+
"rewards/rejected": -2.844818115234375,
|
| 69 |
+
"step": 20
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.0546448087431694,
|
| 73 |
+
"grad_norm": 36.310526987064065,
|
| 74 |
+
"learning_rate": 4.891304347826087e-07,
|
| 75 |
+
"logits/chosen": -0.9477519989013672,
|
| 76 |
+
"logits/rejected": -0.8758963346481323,
|
| 77 |
+
"logps/chosen": -0.2740391790866852,
|
| 78 |
+
"logps/rejected": -0.29474973678588867,
|
| 79 |
+
"loss": 3.916,
|
| 80 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 81 |
+
"rewards/chosen": -2.740391969680786,
|
| 82 |
+
"rewards/margins": 0.20710572600364685,
|
| 83 |
+
"rewards/rejected": -2.9474973678588867,
|
| 84 |
+
"step": 25
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.06557377049180328,
|
| 88 |
+
"grad_norm": 57.63053904902577,
|
| 89 |
+
"learning_rate": 5.869565217391305e-07,
|
| 90 |
+
"logits/chosen": -1.055710792541504,
|
| 91 |
+
"logits/rejected": -0.9906851649284363,
|
| 92 |
+
"logps/chosen": -0.26672154664993286,
|
| 93 |
+
"logps/rejected": -0.2856570780277252,
|
| 94 |
+
"loss": 3.8988,
|
| 95 |
+
"rewards/accuracies": 0.53125,
|
| 96 |
+
"rewards/chosen": -2.667215347290039,
|
| 97 |
+
"rewards/margins": 0.18935534358024597,
|
| 98 |
+
"rewards/rejected": -2.8565704822540283,
|
| 99 |
+
"step": 30
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.07650273224043716,
|
| 103 |
+
"grad_norm": 66.01668630917158,
|
| 104 |
+
"learning_rate": 6.847826086956522e-07,
|
| 105 |
+
"logits/chosen": -1.0005064010620117,
|
| 106 |
+
"logits/rejected": -0.9350866079330444,
|
| 107 |
+
"logps/chosen": -0.2575780749320984,
|
| 108 |
+
"logps/rejected": -0.2911999523639679,
|
| 109 |
+
"loss": 3.844,
|
| 110 |
+
"rewards/accuracies": 0.512499988079071,
|
| 111 |
+
"rewards/chosen": -2.5757808685302734,
|
| 112 |
+
"rewards/margins": 0.3362187445163727,
|
| 113 |
+
"rewards/rejected": -2.911999464035034,
|
| 114 |
+
"step": 35
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.08743169398907104,
|
| 118 |
+
"grad_norm": 76.01107140178125,
|
| 119 |
+
"learning_rate": 7.826086956521739e-07,
|
| 120 |
+
"logits/chosen": -0.959019660949707,
|
| 121 |
+
"logits/rejected": -0.8994712829589844,
|
| 122 |
+
"logps/chosen": -0.2856209874153137,
|
| 123 |
+
"logps/rejected": -0.3126066327095032,
|
| 124 |
+
"loss": 3.9495,
|
| 125 |
+
"rewards/accuracies": 0.518750011920929,
|
| 126 |
+
"rewards/chosen": -2.8562099933624268,
|
| 127 |
+
"rewards/margins": 0.2698565423488617,
|
| 128 |
+
"rewards/rejected": -3.1260666847229004,
|
| 129 |
+
"step": 40
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.09836065573770492,
|
| 133 |
+
"grad_norm": 34.55317412546812,
|
| 134 |
+
"learning_rate": 8.804347826086956e-07,
|
| 135 |
+
"logits/chosen": -1.0179210901260376,
|
| 136 |
+
"logits/rejected": -0.9362085461616516,
|
| 137 |
+
"logps/chosen": -0.2937139570713043,
|
| 138 |
+
"logps/rejected": -0.33561640977859497,
|
| 139 |
+
"loss": 3.8487,
|
| 140 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 141 |
+
"rewards/chosen": -2.9371399879455566,
|
| 142 |
+
"rewards/margins": 0.4190244674682617,
|
| 143 |
+
"rewards/rejected": -3.3561644554138184,
|
| 144 |
+
"step": 45
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.1092896174863388,
|
| 148 |
+
"grad_norm": 79.7939449629563,
|
| 149 |
+
"learning_rate": 8.997896779678812e-07,
|
| 150 |
+
"logits/chosen": -0.9532683491706848,
|
| 151 |
+
"logits/rejected": -0.8744559288024902,
|
| 152 |
+
"logps/chosen": -0.3062035143375397,
|
| 153 |
+
"logps/rejected": -0.3324257433414459,
|
| 154 |
+
"loss": 3.8515,
|
| 155 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 156 |
+
"rewards/chosen": -3.062035083770752,
|
| 157 |
+
"rewards/margins": 0.2622221112251282,
|
| 158 |
+
"rewards/rejected": -3.3242568969726562,
|
| 159 |
+
"step": 50
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.12021857923497267,
|
| 163 |
+
"grad_norm": 44.924372418868195,
|
| 164 |
+
"learning_rate": 8.989355816502526e-07,
|
| 165 |
+
"logits/chosen": -0.9827022552490234,
|
| 166 |
+
"logits/rejected": -0.872309684753418,
|
| 167 |
+
"logps/chosen": -0.2938118278980255,
|
| 168 |
+
"logps/rejected": -0.3459795117378235,
|
| 169 |
+
"loss": 3.7417,
|
| 170 |
+
"rewards/accuracies": 0.581250011920929,
|
| 171 |
+
"rewards/chosen": -2.9381184577941895,
|
| 172 |
+
"rewards/margins": 0.521676242351532,
|
| 173 |
+
"rewards/rejected": -3.4597949981689453,
|
| 174 |
+
"step": 55
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.13114754098360656,
|
| 178 |
+
"grad_norm": 39.977262723369954,
|
| 179 |
+
"learning_rate": 8.974258123725301e-07,
|
| 180 |
+
"logits/chosen": -1.0276820659637451,
|
| 181 |
+
"logits/rejected": -0.9835308790206909,
|
| 182 |
+
"logps/chosen": -0.2806386947631836,
|
| 183 |
+
"logps/rejected": -0.35135510563850403,
|
| 184 |
+
"loss": 3.6866,
|
| 185 |
+
"rewards/accuracies": 0.606249988079071,
|
| 186 |
+
"rewards/chosen": -2.806387186050415,
|
| 187 |
+
"rewards/margins": 0.7071643471717834,
|
| 188 |
+
"rewards/rejected": -3.5135509967803955,
|
| 189 |
+
"step": 60
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.14207650273224043,
|
| 193 |
+
"grad_norm": 59.29745253893998,
|
| 194 |
+
"learning_rate": 8.952625751612967e-07,
|
| 195 |
+
"logits/chosen": -1.0310702323913574,
|
| 196 |
+
"logits/rejected": -0.9638811349868774,
|
| 197 |
+
"logps/chosen": -0.3306628465652466,
|
| 198 |
+
"logps/rejected": -0.364567369222641,
|
| 199 |
+
"loss": 3.8342,
|
| 200 |
+
"rewards/accuracies": 0.518750011920929,
|
| 201 |
+
"rewards/chosen": -3.306628465652466,
|
| 202 |
+
"rewards/margins": 0.3390454351902008,
|
| 203 |
+
"rewards/rejected": -3.645674228668213,
|
| 204 |
+
"step": 65
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.15300546448087432,
|
| 208 |
+
"grad_norm": 156.8250507356202,
|
| 209 |
+
"learning_rate": 8.924490294367533e-07,
|
| 210 |
+
"logits/chosen": -0.9752784967422485,
|
| 211 |
+
"logits/rejected": -0.9566763639450073,
|
| 212 |
+
"logps/chosen": -0.31856533885002136,
|
| 213 |
+
"logps/rejected": -0.3732706904411316,
|
| 214 |
+
"loss": 3.7076,
|
| 215 |
+
"rewards/accuracies": 0.5625,
|
| 216 |
+
"rewards/chosen": -3.1856534481048584,
|
| 217 |
+
"rewards/margins": 0.5470534563064575,
|
| 218 |
+
"rewards/rejected": -3.7327072620391846,
|
| 219 |
+
"step": 70
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.16393442622950818,
|
| 223 |
+
"grad_norm": 43.26313827052752,
|
| 224 |
+
"learning_rate": 8.889892843983686e-07,
|
| 225 |
+
"logits/chosen": -0.9661557078361511,
|
| 226 |
+
"logits/rejected": -0.9509698152542114,
|
| 227 |
+
"logps/chosen": -0.3259205222129822,
|
| 228 |
+
"logps/rejected": -0.39959517121315,
|
| 229 |
+
"loss": 3.7485,
|
| 230 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 231 |
+
"rewards/chosen": -3.2592053413391113,
|
| 232 |
+
"rewards/margins": 0.7367462515830994,
|
| 233 |
+
"rewards/rejected": -3.9959514141082764,
|
| 234 |
+
"step": 75
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.17486338797814208,
|
| 238 |
+
"grad_norm": 91.2494199085889,
|
| 239 |
+
"learning_rate": 8.848883930233695e-07,
|
| 240 |
+
"logits/chosen": -0.9704955220222473,
|
| 241 |
+
"logits/rejected": -0.9103384017944336,
|
| 242 |
+
"logps/chosen": -0.3389447033405304,
|
| 243 |
+
"logps/rejected": -0.3731645941734314,
|
| 244 |
+
"loss": 3.7051,
|
| 245 |
+
"rewards/accuracies": 0.59375,
|
| 246 |
+
"rewards/chosen": -3.3894474506378174,
|
| 247 |
+
"rewards/margins": 0.34219902753829956,
|
| 248 |
+
"rewards/rejected": -3.7316460609436035,
|
| 249 |
+
"step": 80
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.18579234972677597,
|
| 253 |
+
"grad_norm": 77.88395363509508,
|
| 254 |
+
"learning_rate": 8.801523446868397e-07,
|
| 255 |
+
"logits/chosen": -0.9626806378364563,
|
| 256 |
+
"logits/rejected": -0.9148591756820679,
|
| 257 |
+
"logps/chosen": -0.33388131856918335,
|
| 258 |
+
"logps/rejected": -0.4132775664329529,
|
| 259 |
+
"loss": 3.6667,
|
| 260 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 261 |
+
"rewards/chosen": -3.338813304901123,
|
| 262 |
+
"rewards/margins": 0.7939624190330505,
|
| 263 |
+
"rewards/rejected": -4.132775783538818,
|
| 264 |
+
"step": 85
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.19672131147540983,
|
| 268 |
+
"grad_norm": 52.15269440463944,
|
| 269 |
+
"learning_rate": 8.74788056414203e-07,
|
| 270 |
+
"logits/chosen": -1.0256249904632568,
|
| 271 |
+
"logits/rejected": -0.9451953172683716,
|
| 272 |
+
"logps/chosen": -0.37562912702560425,
|
| 273 |
+
"logps/rejected": -0.4213550090789795,
|
| 274 |
+
"loss": 3.7121,
|
| 275 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 276 |
+
"rewards/chosen": -3.756291627883911,
|
| 277 |
+
"rewards/margins": 0.4572587013244629,
|
| 278 |
+
"rewards/rejected": -4.213550090789795,
|
| 279 |
+
"step": 90
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.20765027322404372,
|
| 283 |
+
"grad_norm": 56.09180280161284,
|
| 284 |
+
"learning_rate": 8.688033627788689e-07,
|
| 285 |
+
"logits/chosen": -0.9407247304916382,
|
| 286 |
+
"logits/rejected": -0.9443449974060059,
|
| 287 |
+
"logps/chosen": -0.3641100525856018,
|
| 288 |
+
"logps/rejected": -0.409776508808136,
|
| 289 |
+
"loss": 3.5942,
|
| 290 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 291 |
+
"rewards/chosen": -3.6411004066467285,
|
| 292 |
+
"rewards/margins": 0.45666465163230896,
|
| 293 |
+
"rewards/rejected": -4.09776496887207,
|
| 294 |
+
"step": 95
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.2185792349726776,
|
| 298 |
+
"grad_norm": 109.82437730440958,
|
| 299 |
+
"learning_rate": 8.622070044597933e-07,
|
| 300 |
+
"logits/chosen": -0.9516464471817017,
|
| 301 |
+
"logits/rejected": -0.9063247442245483,
|
| 302 |
+
"logps/chosen": -0.4270172119140625,
|
| 303 |
+
"logps/rejected": -0.5119742155075073,
|
| 304 |
+
"loss": 3.5993,
|
| 305 |
+
"rewards/accuracies": 0.612500011920929,
|
| 306 |
+
"rewards/chosen": -4.270172119140625,
|
| 307 |
+
"rewards/margins": 0.8495694398880005,
|
| 308 |
+
"rewards/rejected": -5.119741916656494,
|
| 309 |
+
"step": 100
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.22950819672131148,
|
| 313 |
+
"grad_norm": 65.01582060489105,
|
| 314 |
+
"learning_rate": 8.550086154756677e-07,
|
| 315 |
+
"logits/chosen": -1.0168472528457642,
|
| 316 |
+
"logits/rejected": -0.9868780374526978,
|
| 317 |
+
"logps/chosen": -0.38424795866012573,
|
| 318 |
+
"logps/rejected": -0.4684266149997711,
|
| 319 |
+
"loss": 3.5226,
|
| 320 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 321 |
+
"rewards/chosen": -3.842479705810547,
|
| 322 |
+
"rewards/margins": 0.8417860865592957,
|
| 323 |
+
"rewards/rejected": -4.684265613555908,
|
| 324 |
+
"step": 105
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.24043715846994534,
|
| 328 |
+
"grad_norm": 64.20949181772441,
|
| 329 |
+
"learning_rate": 8.472187091143803e-07,
|
| 330 |
+
"logits/chosen": -1.0123231410980225,
|
| 331 |
+
"logits/rejected": -0.9926570057868958,
|
| 332 |
+
"logps/chosen": -0.42821940779685974,
|
| 333 |
+
"logps/rejected": -0.5515985488891602,
|
| 334 |
+
"loss": 3.5137,
|
| 335 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 336 |
+
"rewards/chosen": -4.2821946144104,
|
| 337 |
+
"rewards/margins": 1.2337915897369385,
|
| 338 |
+
"rewards/rejected": -5.51598596572876,
|
| 339 |
+
"step": 110
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.25136612021857924,
|
| 343 |
+
"grad_norm": 69.07389330772023,
|
| 344 |
+
"learning_rate": 8.388486625782994e-07,
|
| 345 |
+
"logits/chosen": -1.0123811960220337,
|
| 346 |
+
"logits/rejected": -0.9744676351547241,
|
| 347 |
+
"logps/chosen": -0.43518343567848206,
|
| 348 |
+
"logps/rejected": -0.5923277139663696,
|
| 349 |
+
"loss": 3.4477,
|
| 350 |
+
"rewards/accuracies": 0.71875,
|
| 351 |
+
"rewards/chosen": -4.351834297180176,
|
| 352 |
+
"rewards/margins": 1.5714432001113892,
|
| 353 |
+
"rewards/rejected": -5.923277378082275,
|
| 354 |
+
"step": 115
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.26229508196721313,
|
| 358 |
+
"grad_norm": 52.47296311016427,
|
| 359 |
+
"learning_rate": 8.299107003678049e-07,
|
| 360 |
+
"logits/chosen": -1.0302703380584717,
|
| 361 |
+
"logits/rejected": -0.9615422487258911,
|
| 362 |
+
"logps/chosen": -0.45768284797668457,
|
| 363 |
+
"logps/rejected": -0.5608014464378357,
|
| 364 |
+
"loss": 3.3469,
|
| 365 |
+
"rewards/accuracies": 0.706250011920929,
|
| 366 |
+
"rewards/chosen": -4.576828956604004,
|
| 367 |
+
"rewards/margins": 1.0311863422393799,
|
| 368 |
+
"rewards/rejected": -5.608015537261963,
|
| 369 |
+
"step": 120
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.273224043715847,
|
| 373 |
+
"grad_norm": 113.28126785496374,
|
| 374 |
+
"learning_rate": 8.204178764273363e-07,
|
| 375 |
+
"logits/chosen": -0.932907223701477,
|
| 376 |
+
"logits/rejected": -0.9002698659896851,
|
| 377 |
+
"logps/chosen": -0.5244953632354736,
|
| 378 |
+
"logps/rejected": -0.6600042581558228,
|
| 379 |
+
"loss": 3.329,
|
| 380 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 381 |
+
"rewards/chosen": -5.2449541091918945,
|
| 382 |
+
"rewards/margins": 1.355088710784912,
|
| 383 |
+
"rewards/rejected": -6.60004186630249,
|
| 384 |
+
"step": 125
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.28415300546448086,
|
| 388 |
+
"grad_norm": 62.72252412180173,
|
| 389 |
+
"learning_rate": 8.103840550800328e-07,
|
| 390 |
+
"logits/chosen": -0.9439579844474792,
|
| 391 |
+
"logits/rejected": -0.9413612484931946,
|
| 392 |
+
"logps/chosen": -0.5499246716499329,
|
| 393 |
+
"logps/rejected": -0.7896748185157776,
|
| 394 |
+
"loss": 3.2822,
|
| 395 |
+
"rewards/accuracies": 0.731249988079071,
|
| 396 |
+
"rewards/chosen": -5.499247074127197,
|
| 397 |
+
"rewards/margins": 2.3975021839141846,
|
| 398 |
+
"rewards/rejected": -7.8967485427856445,
|
| 399 |
+
"step": 130
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.29508196721311475,
|
| 403 |
+
"grad_norm": 57.931397196362106,
|
| 404 |
+
"learning_rate": 7.998238907788109e-07,
|
| 405 |
+
"logits/chosen": -0.986513614654541,
|
| 406 |
+
"logits/rejected": -0.9044733047485352,
|
| 407 |
+
"logps/chosen": -0.5918017029762268,
|
| 408 |
+
"logps/rejected": -0.7940021753311157,
|
| 409 |
+
"loss": 3.1203,
|
| 410 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 411 |
+
"rewards/chosen": -5.918017387390137,
|
| 412 |
+
"rewards/margins": 2.022005081176758,
|
| 413 |
+
"rewards/rejected": -7.9400224685668945,
|
| 414 |
+
"step": 135
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.30601092896174864,
|
| 418 |
+
"grad_norm": 76.37622361250128,
|
| 419 |
+
"learning_rate": 7.887528067034528e-07,
|
| 420 |
+
"logits/chosen": -0.9690658450126648,
|
| 421 |
+
"logits/rejected": -0.9173405766487122,
|
| 422 |
+
"logps/chosen": -0.6729015707969666,
|
| 423 |
+
"logps/rejected": -0.9487813711166382,
|
| 424 |
+
"loss": 2.764,
|
| 425 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 426 |
+
"rewards/chosen": -6.729015350341797,
|
| 427 |
+
"rewards/margins": 2.7587978839874268,
|
| 428 |
+
"rewards/rejected": -9.487812995910645,
|
| 429 |
+
"step": 140
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.31693989071038253,
|
| 433 |
+
"grad_norm": 101.65463918673284,
|
| 434 |
+
"learning_rate": 7.77186972234965e-07,
|
| 435 |
+
"logits/chosen": -0.9030313491821289,
|
| 436 |
+
"logits/rejected": -0.8889007568359375,
|
| 437 |
+
"logps/chosen": -0.7519813776016235,
|
| 438 |
+
"logps/rejected": -0.9187248349189758,
|
| 439 |
+
"loss": 2.9287,
|
| 440 |
+
"rewards/accuracies": 0.699999988079071,
|
| 441 |
+
"rewards/chosen": -7.519814491271973,
|
| 442 |
+
"rewards/margins": 1.667433500289917,
|
| 443 |
+
"rewards/rejected": -9.187247276306152,
|
| 444 |
+
"step": 145
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.32786885245901637,
|
| 448 |
+
"grad_norm": 95.8508047628295,
|
| 449 |
+
"learning_rate": 7.651432793401065e-07,
|
| 450 |
+
"logits/chosen": -0.9583929777145386,
|
| 451 |
+
"logits/rejected": -0.9260581135749817,
|
| 452 |
+
"logps/chosen": -0.924455463886261,
|
| 453 |
+
"logps/rejected": -1.2613859176635742,
|
| 454 |
+
"loss": 2.8309,
|
| 455 |
+
"rewards/accuracies": 0.75,
|
| 456 |
+
"rewards/chosen": -9.24455451965332,
|
| 457 |
+
"rewards/margins": 3.3693041801452637,
|
| 458 |
+
"rewards/rejected": -12.613858222961426,
|
| 459 |
+
"step": 150
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.33879781420765026,
|
| 463 |
+
"grad_norm": 83.36287064815711,
|
| 464 |
+
"learning_rate": 7.526393179005737e-07,
|
| 465 |
+
"logits/chosen": -0.8817826509475708,
|
| 466 |
+
"logits/rejected": -0.8292266130447388,
|
| 467 |
+
"logps/chosen": -1.0013782978057861,
|
| 468 |
+
"logps/rejected": -1.2671784162521362,
|
| 469 |
+
"loss": 2.6483,
|
| 470 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 471 |
+
"rewards/chosen": -10.01378345489502,
|
| 472 |
+
"rewards/margins": 2.6580021381378174,
|
| 473 |
+
"rewards/rejected": -12.671785354614258,
|
| 474 |
+
"step": 155
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.34972677595628415,
|
| 478 |
+
"grad_norm": 81.57566570080871,
|
| 479 |
+
"learning_rate": 7.396933500228808e-07,
|
| 480 |
+
"logits/chosen": -0.9194300770759583,
|
| 481 |
+
"logits/rejected": -0.867072582244873,
|
| 482 |
+
"logps/chosen": -1.1953147649765015,
|
| 483 |
+
"logps/rejected": -1.5064880847930908,
|
| 484 |
+
"loss": 2.5891,
|
| 485 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 486 |
+
"rewards/chosen": -11.953146934509277,
|
| 487 |
+
"rewards/margins": 3.1117310523986816,
|
| 488 |
+
"rewards/rejected": -15.064878463745117,
|
| 489 |
+
"step": 160
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.36065573770491804,
|
| 493 |
+
"grad_norm": 147.68440937820483,
|
| 494 |
+
"learning_rate": 7.263242833664484e-07,
|
| 495 |
+
"logits/chosen": -0.8760455846786499,
|
| 496 |
+
"logits/rejected": -0.8160573244094849,
|
| 497 |
+
"logps/chosen": -1.2987282276153564,
|
| 498 |
+
"logps/rejected": -1.580674171447754,
|
| 499 |
+
"loss": 2.6585,
|
| 500 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 501 |
+
"rewards/chosen": -12.987281799316406,
|
| 502 |
+
"rewards/margins": 2.819460153579712,
|
| 503 |
+
"rewards/rejected": -15.806741714477539,
|
| 504 |
+
"step": 165
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 0.37158469945355194,
|
| 508 |
+
"grad_norm": 106.55434262084324,
|
| 509 |
+
"learning_rate": 7.1255164352886e-07,
|
| 510 |
+
"logits/chosen": -0.9024359583854675,
|
| 511 |
+
"logits/rejected": -0.8437854051589966,
|
| 512 |
+
"logps/chosen": -1.3003956079483032,
|
| 513 |
+
"logps/rejected": -1.6264398097991943,
|
| 514 |
+
"loss": 2.5513,
|
| 515 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 516 |
+
"rewards/chosen": -13.00395679473877,
|
| 517 |
+
"rewards/margins": 3.26043963432312,
|
| 518 |
+
"rewards/rejected": -16.264394760131836,
|
| 519 |
+
"step": 170
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.3825136612021858,
|
| 523 |
+
"grad_norm": 99.75404527237653,
|
| 524 |
+
"learning_rate": 6.98395545528617e-07,
|
| 525 |
+
"logits/chosen": -0.9769757986068726,
|
| 526 |
+
"logits/rejected": -0.9515819549560547,
|
| 527 |
+
"logps/chosen": -1.4226484298706055,
|
| 528 |
+
"logps/rejected": -1.6883213520050049,
|
| 529 |
+
"loss": 2.3385,
|
| 530 |
+
"rewards/accuracies": 0.675000011920929,
|
| 531 |
+
"rewards/chosen": -14.226484298706055,
|
| 532 |
+
"rewards/margins": 2.656728744506836,
|
| 533 |
+
"rewards/rejected": -16.883214950561523,
|
| 534 |
+
"step": 175
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.39344262295081966,
|
| 538 |
+
"grad_norm": 98.07261203721741,
|
| 539 |
+
"learning_rate": 6.838766644270385e-07,
|
| 540 |
+
"logits/chosen": -0.9713080525398254,
|
| 541 |
+
"logits/rejected": -0.9672197103500366,
|
| 542 |
+
"logps/chosen": -1.4138439893722534,
|
| 543 |
+
"logps/rejected": -1.909021019935608,
|
| 544 |
+
"loss": 2.234,
|
| 545 |
+
"rewards/accuracies": 0.793749988079071,
|
| 546 |
+
"rewards/chosen": -14.138440132141113,
|
| 547 |
+
"rewards/margins": 4.951767921447754,
|
| 548 |
+
"rewards/rejected": -19.090206146240234,
|
| 549 |
+
"step": 180
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.40437158469945356,
|
| 553 |
+
"grad_norm": 129.28840963645882,
|
| 554 |
+
"learning_rate": 6.690162051322191e-07,
|
| 555 |
+
"logits/chosen": -0.9913703799247742,
|
| 556 |
+
"logits/rejected": -1.0164626836776733,
|
| 557 |
+
"logps/chosen": -1.5589020252227783,
|
| 558 |
+
"logps/rejected": -2.046978235244751,
|
| 559 |
+
"loss": 2.0769,
|
| 560 |
+
"rewards/accuracies": 0.824999988079071,
|
| 561 |
+
"rewards/chosen": -15.589021682739258,
|
| 562 |
+
"rewards/margins": 4.88076114654541,
|
| 563 |
+
"rewards/rejected": -20.469783782958984,
|
| 564 |
+
"step": 185
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.41530054644808745,
|
| 568 |
+
"grad_norm": 140.495255270974,
|
| 569 |
+
"learning_rate": 6.538358714291389e-07,
|
| 570 |
+
"logits/chosen": -1.0160572528839111,
|
| 571 |
+
"logits/rejected": -0.9758199453353882,
|
| 572 |
+
"logps/chosen": -1.5961698293685913,
|
| 573 |
+
"logps/rejected": -2.09332013130188,
|
| 574 |
+
"loss": 2.2105,
|
| 575 |
+
"rewards/accuracies": 0.8125,
|
| 576 |
+
"rewards/chosen": -15.961697578430176,
|
| 577 |
+
"rewards/margins": 4.971505641937256,
|
| 578 |
+
"rewards/rejected": -20.933202743530273,
|
| 579 |
+
"step": 190
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.4262295081967213,
|
| 583 |
+
"grad_norm": 130.70095112145532,
|
| 584 |
+
"learning_rate": 6.383578342811638e-07,
|
| 585 |
+
"logits/chosen": -1.0288410186767578,
|
| 586 |
+
"logits/rejected": -1.006248116493225,
|
| 587 |
+
"logps/chosen": -1.8221418857574463,
|
| 588 |
+
"logps/rejected": -2.3388142585754395,
|
| 589 |
+
"loss": 2.2076,
|
| 590 |
+
"rewards/accuracies": 0.768750011920929,
|
| 591 |
+
"rewards/chosen": -18.221418380737305,
|
| 592 |
+
"rewards/margins": 5.166724681854248,
|
| 593 |
+
"rewards/rejected": -23.388141632080078,
|
| 594 |
+
"step": 195
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 0.4371584699453552,
|
| 598 |
+
"grad_norm": 129.43163775750423,
|
| 599 |
+
"learning_rate": 6.226046994492272e-07,
|
| 600 |
+
"logits/chosen": -0.9663163423538208,
|
| 601 |
+
"logits/rejected": -0.8765112161636353,
|
| 602 |
+
"logps/chosen": -1.7186319828033447,
|
| 603 |
+
"logps/rejected": -2.242340564727783,
|
| 604 |
+
"loss": 2.0413,
|
| 605 |
+
"rewards/accuracies": 0.800000011920929,
|
| 606 |
+
"rewards/chosen": -17.18631935119629,
|
| 607 |
+
"rewards/margins": 5.237086296081543,
|
| 608 |
+
"rewards/rejected": -22.42340660095215,
|
| 609 |
+
"step": 200
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.44808743169398907,
|
| 613 |
+
"grad_norm": 102.22819729561144,
|
| 614 |
+
"learning_rate": 6.065994744759878e-07,
|
| 615 |
+
"logits/chosen": -0.8948925733566284,
|
| 616 |
+
"logits/rejected": -0.9066826105117798,
|
| 617 |
+
"logps/chosen": -1.7214425802230835,
|
| 618 |
+
"logps/rejected": -2.249507427215576,
|
| 619 |
+
"loss": 2.1905,
|
| 620 |
+
"rewards/accuracies": 0.800000011920929,
|
| 621 |
+
"rewards/chosen": -17.214426040649414,
|
| 622 |
+
"rewards/margins": 5.280648231506348,
|
| 623 |
+
"rewards/rejected": -22.495075225830078,
|
| 624 |
+
"step": 205
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 0.45901639344262296,
|
| 628 |
+
"grad_norm": 164.64988265742974,
|
| 629 |
+
"learning_rate": 5.903655350831845e-07,
|
| 630 |
+
"logits/chosen": -0.9535326957702637,
|
| 631 |
+
"logits/rejected": -0.9509015083312988,
|
| 632 |
+
"logps/chosen": -1.7853710651397705,
|
| 633 |
+
"logps/rejected": -2.4150795936584473,
|
| 634 |
+
"loss": 1.9241,
|
| 635 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 636 |
+
"rewards/chosen": -17.853710174560547,
|
| 637 |
+
"rewards/margins": 6.297087669372559,
|
| 638 |
+
"rewards/rejected": -24.15079689025879,
|
| 639 |
+
"step": 210
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.46994535519125685,
|
| 643 |
+
"grad_norm": 147.64808928844568,
|
| 644 |
+
"learning_rate": 5.73926591031261e-07,
|
| 645 |
+
"logits/chosen": -0.9305051565170288,
|
| 646 |
+
"logits/rejected": -0.8654060363769531,
|
| 647 |
+
"logps/chosen": -1.8320693969726562,
|
| 648 |
+
"logps/rejected": -2.4488117694854736,
|
| 649 |
+
"loss": 2.1048,
|
| 650 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 651 |
+
"rewards/chosen": -18.32069206237793,
|
| 652 |
+
"rewards/margins": 6.16742467880249,
|
| 653 |
+
"rewards/rejected": -24.488117218017578,
|
| 654 |
+
"step": 215
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.4808743169398907,
|
| 658 |
+
"grad_norm": 105.48156854473766,
|
| 659 |
+
"learning_rate": 5.573066514911273e-07,
|
| 660 |
+
"logits/chosen": -0.9947813153266907,
|
| 661 |
+
"logits/rejected": -0.9732875823974609,
|
| 662 |
+
"logps/chosen": -1.8540947437286377,
|
| 663 |
+
"logps/rejected": -2.3756325244903564,
|
| 664 |
+
"loss": 2.0465,
|
| 665 |
+
"rewards/accuracies": 0.731249988079071,
|
| 666 |
+
"rewards/chosen": -18.540945053100586,
|
| 667 |
+
"rewards/margins": 5.215379238128662,
|
| 668 |
+
"rewards/rejected": -23.75632667541504,
|
| 669 |
+
"step": 220
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.4918032786885246,
|
| 673 |
+
"grad_norm": 86.49112885572626,
|
| 674 |
+
"learning_rate": 5.405299899786287e-07,
|
| 675 |
+
"logits/chosen": -1.0302679538726807,
|
| 676 |
+
"logits/rejected": -0.9820007085800171,
|
| 677 |
+
"logps/chosen": -1.8799511194229126,
|
| 678 |
+
"logps/rejected": -2.4549620151519775,
|
| 679 |
+
"loss": 2.0411,
|
| 680 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 681 |
+
"rewards/chosen": -18.799510955810547,
|
| 682 |
+
"rewards/margins": 5.750107288360596,
|
| 683 |
+
"rewards/rejected": -24.549617767333984,
|
| 684 |
+
"step": 225
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 0.5027322404371585,
|
| 688 |
+
"grad_norm": 106.5713172261187,
|
| 689 |
+
"learning_rate": 5.236211089029395e-07,
|
| 690 |
+
"logits/chosen": -0.9404142498970032,
|
| 691 |
+
"logits/rejected": -0.9487746953964233,
|
| 692 |
+
"logps/chosen": -1.8227663040161133,
|
| 693 |
+
"logps/rejected": -2.3787169456481934,
|
| 694 |
+
"loss": 2.1731,
|
| 695 |
+
"rewards/accuracies": 0.793749988079071,
|
| 696 |
+
"rewards/chosen": -18.227664947509766,
|
| 697 |
+
"rewards/margins": 5.55950403213501,
|
| 698 |
+
"rewards/rejected": -23.787168502807617,
|
| 699 |
+
"step": 230
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.5136612021857924,
|
| 703 |
+
"grad_norm": 146.52429329123075,
|
| 704 |
+
"learning_rate": 5.066047037806549e-07,
|
| 705 |
+
"logits/chosen": -0.9814826846122742,
|
| 706 |
+
"logits/rejected": -0.9384028315544128,
|
| 707 |
+
"logps/chosen": -1.871340036392212,
|
| 708 |
+
"logps/rejected": -2.362367630004883,
|
| 709 |
+
"loss": 1.8879,
|
| 710 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 711 |
+
"rewards/chosen": -18.713397979736328,
|
| 712 |
+
"rewards/margins": 4.910274505615234,
|
| 713 |
+
"rewards/rejected": -23.623672485351562,
|
| 714 |
+
"step": 235
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 0.5245901639344263,
|
| 718 |
+
"grad_norm": 109.73224210165276,
|
| 719 |
+
"learning_rate": 4.895056271678488e-07,
|
| 720 |
+
"logits/chosen": -1.0654172897338867,
|
| 721 |
+
"logits/rejected": -1.0743358135223389,
|
| 722 |
+
"logps/chosen": -1.8185491561889648,
|
| 723 |
+
"logps/rejected": -2.438718557357788,
|
| 724 |
+
"loss": 1.7806,
|
| 725 |
+
"rewards/accuracies": 0.831250011920929,
|
| 726 |
+
"rewards/chosen": -18.18549156188965,
|
| 727 |
+
"rewards/margins": 6.201694488525391,
|
| 728 |
+
"rewards/rejected": -24.387187957763672,
|
| 729 |
+
"step": 240
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.5355191256830601,
|
| 733 |
+
"grad_norm": 174.78754415853115,
|
| 734 |
+
"learning_rate": 4.723488523627758e-07,
|
| 735 |
+
"logits/chosen": -1.0102005004882812,
|
| 736 |
+
"logits/rejected": -0.9561554193496704,
|
| 737 |
+
"logps/chosen": -1.7096691131591797,
|
| 738 |
+
"logps/rejected": -2.3459925651550293,
|
| 739 |
+
"loss": 1.9498,
|
| 740 |
+
"rewards/accuracies": 0.793749988079071,
|
| 741 |
+
"rewards/chosen": -17.096691131591797,
|
| 742 |
+
"rewards/margins": 6.363234519958496,
|
| 743 |
+
"rewards/rejected": -23.45992660522461,
|
| 744 |
+
"step": 245
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.546448087431694,
|
| 748 |
+
"grad_norm": 98.18520968432665,
|
| 749 |
+
"learning_rate": 4.55159436932227e-07,
|
| 750 |
+
"logits/chosen": -1.0693771839141846,
|
| 751 |
+
"logits/rejected": -1.028814435005188,
|
| 752 |
+
"logps/chosen": -1.8196604251861572,
|
| 753 |
+
"logps/rejected": -2.324789524078369,
|
| 754 |
+
"loss": 1.9417,
|
| 755 |
+
"rewards/accuracies": 0.762499988079071,
|
| 756 |
+
"rewards/chosen": -18.196605682373047,
|
| 757 |
+
"rewards/margins": 5.051291465759277,
|
| 758 |
+
"rewards/rejected": -23.247896194458008,
|
| 759 |
+
"step": 250
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.5573770491803278,
|
| 763 |
+
"grad_norm": 118.93796678057375,
|
| 764 |
+
"learning_rate": 4.3796248611481193e-07,
|
| 765 |
+
"logits/chosen": -1.0778828859329224,
|
| 766 |
+
"logits/rejected": -1.0436246395111084,
|
| 767 |
+
"logps/chosen": -1.9404518604278564,
|
| 768 |
+
"logps/rejected": -2.521088123321533,
|
| 769 |
+
"loss": 1.9756,
|
| 770 |
+
"rewards/accuracies": 0.8125,
|
| 771 |
+
"rewards/chosen": -19.404516220092773,
|
| 772 |
+
"rewards/margins": 5.806364059448242,
|
| 773 |
+
"rewards/rejected": -25.210880279541016,
|
| 774 |
+
"step": 255
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 0.5683060109289617,
|
| 778 |
+
"grad_norm": 95.05327173585843,
|
| 779 |
+
"learning_rate": 4.2078311615461545e-07,
|
| 780 |
+
"logits/chosen": -1.053112268447876,
|
| 781 |
+
"logits/rejected": -1.038400411605835,
|
| 782 |
+
"logps/chosen": -1.8710072040557861,
|
| 783 |
+
"logps/rejected": -2.473370313644409,
|
| 784 |
+
"loss": 1.8132,
|
| 785 |
+
"rewards/accuracies": 0.824999988079071,
|
| 786 |
+
"rewards/chosen": -18.710071563720703,
|
| 787 |
+
"rewards/margins": 6.023632049560547,
|
| 788 |
+
"rewards/rejected": -24.733705520629883,
|
| 789 |
+
"step": 260
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.5792349726775956,
|
| 793 |
+
"grad_norm": 84.65373658909819,
|
| 794 |
+
"learning_rate": 4.0364641761878056e-07,
|
| 795 |
+
"logits/chosen": -1.0604503154754639,
|
| 796 |
+
"logits/rejected": -1.018703818321228,
|
| 797 |
+
"logps/chosen": -1.9676679372787476,
|
| 798 |
+
"logps/rejected": -2.594107151031494,
|
| 799 |
+
"loss": 1.8076,
|
| 800 |
+
"rewards/accuracies": 0.8125,
|
| 801 |
+
"rewards/chosen": -19.676679611206055,
|
| 802 |
+
"rewards/margins": 6.264393329620361,
|
| 803 |
+
"rewards/rejected": -25.941070556640625,
|
| 804 |
+
"step": 265
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 0.5901639344262295,
|
| 808 |
+
"grad_norm": 110.56084812406964,
|
| 809 |
+
"learning_rate": 3.8657741875259324e-07,
|
| 810 |
+
"logits/chosen": -1.0636670589447021,
|
| 811 |
+
"logits/rejected": -1.0380544662475586,
|
| 812 |
+
"logps/chosen": -2.010746955871582,
|
| 813 |
+
"logps/rejected": -2.699274778366089,
|
| 814 |
+
"loss": 1.6887,
|
| 815 |
+
"rewards/accuracies": 0.824999988079071,
|
| 816 |
+
"rewards/chosen": -20.107471466064453,
|
| 817 |
+
"rewards/margins": 6.885279178619385,
|
| 818 |
+
"rewards/rejected": -26.992746353149414,
|
| 819 |
+
"step": 270
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 0.6010928961748634,
|
| 823 |
+
"grad_norm": 126.7112018665686,
|
| 824 |
+
"learning_rate": 3.6960104892558897e-07,
|
| 825 |
+
"logits/chosen": -1.0398025512695312,
|
| 826 |
+
"logits/rejected": -0.9743970036506653,
|
| 827 |
+
"logps/chosen": -2.058713436126709,
|
| 828 |
+
"logps/rejected": -2.549133062362671,
|
| 829 |
+
"loss": 1.9337,
|
| 830 |
+
"rewards/accuracies": 0.768750011920929,
|
| 831 |
+
"rewards/chosen": -20.58713150024414,
|
| 832 |
+
"rewards/margins": 4.904201984405518,
|
| 833 |
+
"rewards/rejected": -25.491336822509766,
|
| 834 |
+
"step": 275
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 0.6120218579234973,
|
| 838 |
+
"grad_norm": 105.74498813822662,
|
| 839 |
+
"learning_rate": 3.5274210222206623e-07,
|
| 840 |
+
"logits/chosen": -1.0011359453201294,
|
| 841 |
+
"logits/rejected": -1.002959966659546,
|
| 842 |
+
"logps/chosen": -1.9459590911865234,
|
| 843 |
+
"logps/rejected": -2.6398136615753174,
|
| 844 |
+
"loss": 1.9569,
|
| 845 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 846 |
+
"rewards/chosen": -19.459592819213867,
|
| 847 |
+
"rewards/margins": 6.93854284286499,
|
| 848 |
+
"rewards/rejected": -26.39813232421875,
|
| 849 |
+
"step": 280
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.6229508196721312,
|
| 853 |
+
"grad_norm": 107.35437799710022,
|
| 854 |
+
"learning_rate": 3.360252012291866e-07,
|
| 855 |
+
"logits/chosen": -1.1627576351165771,
|
| 856 |
+
"logits/rejected": -1.128688097000122,
|
| 857 |
+
"logps/chosen": -1.9805145263671875,
|
| 858 |
+
"logps/rejected": -2.688811779022217,
|
| 859 |
+
"loss": 1.5785,
|
| 860 |
+
"rewards/accuracies": 0.856249988079071,
|
| 861 |
+
"rewards/chosen": -19.805145263671875,
|
| 862 |
+
"rewards/margins": 7.082972526550293,
|
| 863 |
+
"rewards/rejected": -26.88811683654785,
|
| 864 |
+
"step": 285
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.6338797814207651,
|
| 868 |
+
"grad_norm": 120.95521845176454,
|
| 869 |
+
"learning_rate": 3.1947476107554687e-07,
|
| 870 |
+
"logits/chosen": -1.079388976097107,
|
| 871 |
+
"logits/rejected": -1.0467464923858643,
|
| 872 |
+
"logps/chosen": -1.9965364933013916,
|
| 873 |
+
"logps/rejected": -2.6323485374450684,
|
| 874 |
+
"loss": 1.8251,
|
| 875 |
+
"rewards/accuracies": 0.831250011920929,
|
| 876 |
+
"rewards/chosen": -19.96536636352539,
|
| 877 |
+
"rewards/margins": 6.358120441436768,
|
| 878 |
+
"rewards/rejected": -26.323486328125,
|
| 879 |
+
"step": 290
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 0.644808743169399,
|
| 883 |
+
"grad_norm": 95.93515805418777,
|
| 884 |
+
"learning_rate": 3.0311495377274497e-07,
|
| 885 |
+
"logits/chosen": -1.0501229763031006,
|
| 886 |
+
"logits/rejected": -1.0500692129135132,
|
| 887 |
+
"logps/chosen": -1.992593765258789,
|
| 888 |
+
"logps/rejected": -2.7493128776550293,
|
| 889 |
+
"loss": 1.4798,
|
| 890 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 891 |
+
"rewards/chosen": -19.92593765258789,
|
| 892 |
+
"rewards/margins": 7.567191123962402,
|
| 893 |
+
"rewards/rejected": -27.493127822875977,
|
| 894 |
+
"step": 295
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 0.6557377049180327,
|
| 898 |
+
"grad_norm": 108.4794966509382,
|
| 899 |
+
"learning_rate": 2.8696967291202045e-07,
|
| 900 |
+
"logits/chosen": -1.113196849822998,
|
| 901 |
+
"logits/rejected": -1.0996946096420288,
|
| 902 |
+
"logps/chosen": -1.955629587173462,
|
| 903 |
+
"logps/rejected": -2.571319341659546,
|
| 904 |
+
"loss": 1.7287,
|
| 905 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 906 |
+
"rewards/chosen": -19.556293487548828,
|
| 907 |
+
"rewards/margins": 6.156899452209473,
|
| 908 |
+
"rewards/rejected": -25.713191986083984,
|
| 909 |
+
"step": 300
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 0.6666666666666666,
|
| 913 |
+
"grad_norm": 129.54118774129594,
|
| 914 |
+
"learning_rate": 2.710624987675287e-07,
|
| 915 |
+
"logits/chosen": -1.1094138622283936,
|
| 916 |
+
"logits/rejected": -1.0546791553497314,
|
| 917 |
+
"logps/chosen": -2.0171821117401123,
|
| 918 |
+
"logps/rejected": -2.6892952919006348,
|
| 919 |
+
"loss": 1.7859,
|
| 920 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 921 |
+
"rewards/chosen": -20.171823501586914,
|
| 922 |
+
"rewards/margins": 6.72113037109375,
|
| 923 |
+
"rewards/rejected": -26.892953872680664,
|
| 924 |
+
"step": 305
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 0.6775956284153005,
|
| 928 |
+
"grad_norm": 115.80124041818156,
|
| 929 |
+
"learning_rate": 2.554166638572174e-07,
|
| 930 |
+
"logits/chosen": -1.029006004333496,
|
| 931 |
+
"logits/rejected": -1.0432313680648804,
|
| 932 |
+
"logps/chosen": -2.045400381088257,
|
| 933 |
+
"logps/rejected": -2.7476630210876465,
|
| 934 |
+
"loss": 1.8226,
|
| 935 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 936 |
+
"rewards/chosen": -20.454004287719727,
|
| 937 |
+
"rewards/margins": 7.022623538970947,
|
| 938 |
+
"rewards/rejected": -27.476627349853516,
|
| 939 |
+
"step": 310
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 0.6885245901639344,
|
| 943 |
+
"grad_norm": 118.15275803303632,
|
| 944 |
+
"learning_rate": 2.4005501901160237e-07,
|
| 945 |
+
"logits/chosen": -1.0894476175308228,
|
| 946 |
+
"logits/rejected": -1.0898199081420898,
|
| 947 |
+
"logps/chosen": -2.1980321407318115,
|
| 948 |
+
"logps/rejected": -2.6982712745666504,
|
| 949 |
+
"loss": 1.7214,
|
| 950 |
+
"rewards/accuracies": 0.8125,
|
| 951 |
+
"rewards/chosen": -21.980321884155273,
|
| 952 |
+
"rewards/margins": 5.0023908615112305,
|
| 953 |
+
"rewards/rejected": -26.982711791992188,
|
| 954 |
+
"step": 315
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.6994535519125683,
|
| 958 |
+
"grad_norm": 102.60918367059409,
|
| 959 |
+
"learning_rate": 2.250000000000001e-07,
|
| 960 |
+
"logits/chosen": -1.1670842170715332,
|
| 961 |
+
"logits/rejected": -1.114605188369751,
|
| 962 |
+
"logps/chosen": -2.1052815914154053,
|
| 963 |
+
"logps/rejected": -2.743218183517456,
|
| 964 |
+
"loss": 1.702,
|
| 965 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 966 |
+
"rewards/chosen": -21.05281639099121,
|
| 967 |
+
"rewards/margins": 6.37936544418335,
|
| 968 |
+
"rewards/rejected": -27.432180404663086,
|
| 969 |
+
"step": 320
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.7103825136612022,
|
| 973 |
+
"grad_norm": 142.9270542485795,
|
| 974 |
+
"learning_rate": 2.1027359476295932e-07,
|
| 975 |
+
"logits/chosen": -1.1268647909164429,
|
| 976 |
+
"logits/rejected": -1.1123411655426025,
|
| 977 |
+
"logps/chosen": -1.9478269815444946,
|
| 978 |
+
"logps/rejected": -2.504878520965576,
|
| 979 |
+
"loss": 1.739,
|
| 980 |
+
"rewards/accuracies": 0.831250011920929,
|
| 981 |
+
"rewards/chosen": -19.478267669677734,
|
| 982 |
+
"rewards/margins": 5.57051944732666,
|
| 983 |
+
"rewards/rejected": -25.04878807067871,
|
| 984 |
+
"step": 325
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 0.7213114754098361,
|
| 988 |
+
"grad_norm": 87.02009137374039,
|
| 989 |
+
"learning_rate": 1.9589731129874957e-07,
|
| 990 |
+
"logits/chosen": -1.036443829536438,
|
| 991 |
+
"logits/rejected": -1.0129362344741821,
|
| 992 |
+
"logps/chosen": -1.9629653692245483,
|
| 993 |
+
"logps/rejected": -2.7208199501037598,
|
| 994 |
+
"loss": 1.6483,
|
| 995 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 996 |
+
"rewards/chosen": -19.629653930664062,
|
| 997 |
+
"rewards/margins": 7.57854700088501,
|
| 998 |
+
"rewards/rejected": -27.208200454711914,
|
| 999 |
+
"step": 330
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 0.73224043715847,
|
| 1003 |
+
"grad_norm": 88.46092221124316,
|
| 1004 |
+
"learning_rate": 1.8189214625080817e-07,
|
| 1005 |
+
"logits/chosen": -1.014481782913208,
|
| 1006 |
+
"logits/rejected": -0.9782212972640991,
|
| 1007 |
+
"logps/chosen": -2.0324487686157227,
|
| 1008 |
+
"logps/rejected": -2.6788439750671387,
|
| 1009 |
+
"loss": 1.6273,
|
| 1010 |
+
"rewards/accuracies": 0.84375,
|
| 1011 |
+
"rewards/chosen": -20.32448959350586,
|
| 1012 |
+
"rewards/margins": 6.463950157165527,
|
| 1013 |
+
"rewards/rejected": -26.788440704345703,
|
| 1014 |
+
"step": 335
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 0.7431693989071039,
|
| 1018 |
+
"grad_norm": 135.37943482075397,
|
| 1019 |
+
"learning_rate": 1.6827855424202287e-07,
|
| 1020 |
+
"logits/chosen": -1.0468891859054565,
|
| 1021 |
+
"logits/rejected": -1.002845048904419,
|
| 1022 |
+
"logps/chosen": -2.073441982269287,
|
| 1023 |
+
"logps/rejected": -2.709076166152954,
|
| 1024 |
+
"loss": 1.8334,
|
| 1025 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 1026 |
+
"rewards/chosen": -20.734418869018555,
|
| 1027 |
+
"rewards/margins": 6.3563408851623535,
|
| 1028 |
+
"rewards/rejected": -27.09075927734375,
|
| 1029 |
+
"step": 340
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 0.7540983606557377,
|
| 1033 |
+
"grad_norm": 105.42435410857779,
|
| 1034 |
+
"learning_rate": 1.550764180006395e-07,
|
| 1035 |
+
"logits/chosen": -1.1121171712875366,
|
| 1036 |
+
"logits/rejected": -1.1094890832901,
|
| 1037 |
+
"logps/chosen": -2.104811191558838,
|
| 1038 |
+
"logps/rejected": -2.760361909866333,
|
| 1039 |
+
"loss": 1.8091,
|
| 1040 |
+
"rewards/accuracies": 0.8125,
|
| 1041 |
+
"rewards/chosen": -21.048110961914062,
|
| 1042 |
+
"rewards/margins": 6.555506229400635,
|
| 1043 |
+
"rewards/rejected": -27.60361671447754,
|
| 1044 |
+
"step": 345
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 0.7650273224043715,
|
| 1048 |
+
"grad_norm": 99.44232512456259,
|
| 1049 |
+
"learning_rate": 1.4230501932142528e-07,
|
| 1050 |
+
"logits/chosen": -1.0965338945388794,
|
| 1051 |
+
"logits/rejected": -1.0752547979354858,
|
| 1052 |
+
"logps/chosen": -2.0224196910858154,
|
| 1053 |
+
"logps/rejected": -2.5650455951690674,
|
| 1054 |
+
"loss": 1.7276,
|
| 1055 |
+
"rewards/accuracies": 0.856249988079071,
|
| 1056 |
+
"rewards/chosen": -20.224197387695312,
|
| 1057 |
+
"rewards/margins": 5.4262590408325195,
|
| 1058 |
+
"rewards/rejected": -25.650455474853516,
|
| 1059 |
+
"step": 350
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.7759562841530054,
|
| 1063 |
+
"grad_norm": 95.86354961266959,
|
| 1064 |
+
"learning_rate": 1.2998301090449732e-07,
|
| 1065 |
+
"logits/chosen": -1.0493485927581787,
|
| 1066 |
+
"logits/rejected": -1.0148344039916992,
|
| 1067 |
+
"logps/chosen": -2.0027577877044678,
|
| 1068 |
+
"logps/rejected": -2.6467907428741455,
|
| 1069 |
+
"loss": 1.643,
|
| 1070 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1071 |
+
"rewards/chosen": -20.027578353881836,
|
| 1072 |
+
"rewards/margins": 6.440329074859619,
|
| 1073 |
+
"rewards/rejected": -26.467906951904297,
|
| 1074 |
+
"step": 355
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.7868852459016393,
|
| 1078 |
+
"grad_norm": 113.44769593918105,
|
| 1079 |
+
"learning_rate": 1.181283891129494e-07,
|
| 1080 |
+
"logits/chosen": -1.0109162330627441,
|
| 1081 |
+
"logits/rejected": -1.0162619352340698,
|
| 1082 |
+
"logps/chosen": -2.096867799758911,
|
| 1083 |
+
"logps/rejected": -2.732240676879883,
|
| 1084 |
+
"loss": 1.7079,
|
| 1085 |
+
"rewards/accuracies": 0.78125,
|
| 1086 |
+
"rewards/chosen": -20.968677520751953,
|
| 1087 |
+
"rewards/margins": 6.353725433349609,
|
| 1088 |
+
"rewards/rejected": -27.322406768798828,
|
| 1089 |
+
"step": 360
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 0.7978142076502732,
|
| 1093 |
+
"grad_norm": 114.10477131435252,
|
| 1094 |
+
"learning_rate": 1.0675846768906093e-07,
|
| 1095 |
+
"logits/chosen": -1.1015875339508057,
|
| 1096 |
+
"logits/rejected": -1.1041442155838013,
|
| 1097 |
+
"logps/chosen": -2.0886998176574707,
|
| 1098 |
+
"logps/rejected": -2.7303638458251953,
|
| 1099 |
+
"loss": 1.7767,
|
| 1100 |
+
"rewards/accuracies": 0.8125,
|
| 1101 |
+
"rewards/chosen": -20.88699722290039,
|
| 1102 |
+
"rewards/margins": 6.416640281677246,
|
| 1103 |
+
"rewards/rejected": -27.303638458251953,
|
| 1104 |
+
"step": 365
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 0.8087431693989071,
|
| 1108 |
+
"grad_norm": 144.30812317733663,
|
| 1109 |
+
"learning_rate": 9.588985246747924e-08,
|
| 1110 |
+
"logits/chosen": -1.121105432510376,
|
| 1111 |
+
"logits/rejected": -1.0874660015106201,
|
| 1112 |
+
"logps/chosen": -2.131044864654541,
|
| 1113 |
+
"logps/rejected": -2.8191640377044678,
|
| 1114 |
+
"loss": 1.7266,
|
| 1115 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1116 |
+
"rewards/chosen": -21.31045150756836,
|
| 1117 |
+
"rewards/margins": 6.881192207336426,
|
| 1118 |
+
"rewards/rejected": -28.191640853881836,
|
| 1119 |
+
"step": 370
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 0.819672131147541,
|
| 1123 |
+
"grad_norm": 88.75216133476276,
|
| 1124 |
+
"learning_rate": 8.553841712230287e-08,
|
| 1125 |
+
"logits/chosen": -1.1224949359893799,
|
| 1126 |
+
"logits/rejected": -1.0620272159576416,
|
| 1127 |
+
"logps/chosen": -2.079634666442871,
|
| 1128 |
+
"logps/rejected": -2.7826874256134033,
|
| 1129 |
+
"loss": 1.6434,
|
| 1130 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1131 |
+
"rewards/chosen": -20.796348571777344,
|
| 1132 |
+
"rewards/margins": 7.030525207519531,
|
| 1133 |
+
"rewards/rejected": -27.826873779296875,
|
| 1134 |
+
"step": 375
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 0.8306010928961749,
|
| 1138 |
+
"grad_norm": 103.25429828047173,
|
| 1139 |
+
"learning_rate": 7.571927998349143e-08,
|
| 1140 |
+
"logits/chosen": -1.0042684078216553,
|
| 1141 |
+
"logits/rejected": -1.0183018445968628,
|
| 1142 |
+
"logps/chosen": -2.2109534740448,
|
| 1143 |
+
"logps/rejected": -2.83004093170166,
|
| 1144 |
+
"loss": 1.6596,
|
| 1145 |
+
"rewards/accuracies": 0.84375,
|
| 1146 |
+
"rewards/chosen": -22.109533309936523,
|
| 1147 |
+
"rewards/margins": 6.190876007080078,
|
| 1148 |
+
"rewards/rejected": -28.3004093170166,
|
| 1149 |
+
"step": 380
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 0.8415300546448088,
|
| 1153 |
+
"grad_norm": 126.47652062512991,
|
| 1154 |
+
"learning_rate": 6.644678195645878e-08,
|
| 1155 |
+
"logits/chosen": -1.1039857864379883,
|
| 1156 |
+
"logits/rejected": -1.087054967880249,
|
| 1157 |
+
"logps/chosen": -2.26637601852417,
|
| 1158 |
+
"logps/rejected": -3.011500597000122,
|
| 1159 |
+
"loss": 2.0123,
|
| 1160 |
+
"rewards/accuracies": 0.8125,
|
| 1161 |
+
"rewards/chosen": -22.663759231567383,
|
| 1162 |
+
"rewards/margins": 7.451247215270996,
|
| 1163 |
+
"rewards/rejected": -30.115009307861328,
|
| 1164 |
+
"step": 385
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.8524590163934426,
|
| 1168 |
+
"grad_norm": 143.34550375994243,
|
| 1169 |
+
"learning_rate": 5.77344655770996e-08,
|
| 1170 |
+
"logits/chosen": -1.1028085947036743,
|
| 1171 |
+
"logits/rejected": -1.0805258750915527,
|
| 1172 |
+
"logps/chosen": -2.170258045196533,
|
| 1173 |
+
"logps/rejected": -2.950200319290161,
|
| 1174 |
+
"loss": 1.582,
|
| 1175 |
+
"rewards/accuracies": 0.824999988079071,
|
| 1176 |
+
"rewards/chosen": -21.702577590942383,
|
| 1177 |
+
"rewards/margins": 7.799424648284912,
|
| 1178 |
+
"rewards/rejected": -29.502002716064453,
|
| 1179 |
+
"step": 390
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.8633879781420765,
|
| 1183 |
+
"grad_norm": 137.04591581071145,
|
| 1184 |
+
"learning_rate": 4.9595055232838666e-08,
|
| 1185 |
+
"logits/chosen": -1.0706136226654053,
|
| 1186 |
+
"logits/rejected": -1.0520108938217163,
|
| 1187 |
+
"logps/chosen": -2.3184380531311035,
|
| 1188 |
+
"logps/rejected": -2.9153521060943604,
|
| 1189 |
+
"loss": 1.6624,
|
| 1190 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1191 |
+
"rewards/chosen": -23.18438148498535,
|
| 1192 |
+
"rewards/margins": 5.96914005279541,
|
| 1193 |
+
"rewards/rejected": -29.153518676757812,
|
| 1194 |
+
"step": 395
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 0.8743169398907104,
|
| 1198 |
+
"grad_norm": 115.5484757892347,
|
| 1199 |
+
"learning_rate": 4.204043857859128e-08,
|
| 1200 |
+
"logits/chosen": -1.030194640159607,
|
| 1201 |
+
"logits/rejected": -0.9990195035934448,
|
| 1202 |
+
"logps/chosen": -2.136531352996826,
|
| 1203 |
+
"logps/rejected": -2.8680989742279053,
|
| 1204 |
+
"loss": 1.6588,
|
| 1205 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 1206 |
+
"rewards/chosen": -21.365314483642578,
|
| 1207 |
+
"rewards/margins": 7.315672874450684,
|
| 1208 |
+
"rewards/rejected": -28.680988311767578,
|
| 1209 |
+
"step": 400
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 0.8743169398907104,
|
| 1213 |
+
"eval_logits/chosen": -1.2835370302200317,
|
| 1214 |
+
"eval_logits/rejected": -1.2388678789138794,
|
| 1215 |
+
"eval_logps/chosen": -2.1050703525543213,
|
| 1216 |
+
"eval_logps/rejected": -2.813657522201538,
|
| 1217 |
+
"eval_loss": 1.6786526441574097,
|
| 1218 |
+
"eval_rewards/accuracies": 0.8373494148254395,
|
| 1219 |
+
"eval_rewards/chosen": -21.050703048706055,
|
| 1220 |
+
"eval_rewards/margins": 7.0858683586120605,
|
| 1221 |
+
"eval_rewards/rejected": -28.136571884155273,
|
| 1222 |
+
"eval_runtime": 37.0957,
|
| 1223 |
+
"eval_samples_per_second": 35.53,
|
| 1224 |
+
"eval_steps_per_second": 2.237,
|
| 1225 |
+
"step": 400
|
| 1226 |
+
},
|
| 1227 |
+
{
|
| 1228 |
+
"epoch": 0.8852459016393442,
|
| 1229 |
+
"grad_norm": 111.67107585308769,
|
| 1230 |
+
"learning_rate": 3.508164917477564e-08,
|
| 1231 |
+
"logits/chosen": -1.0428274869918823,
|
| 1232 |
+
"logits/rejected": -0.9762898683547974,
|
| 1233 |
+
"logps/chosen": -2.0831797122955322,
|
| 1234 |
+
"logps/rejected": -2.7335562705993652,
|
| 1235 |
+
"loss": 1.7049,
|
| 1236 |
+
"rewards/accuracies": 0.831250011920929,
|
| 1237 |
+
"rewards/chosen": -20.831798553466797,
|
| 1238 |
+
"rewards/margins": 6.503766059875488,
|
| 1239 |
+
"rewards/rejected": -27.3355655670166,
|
| 1240 |
+
"step": 405
|
| 1241 |
+
},
|
| 1242 |
+
{
|
| 1243 |
+
"epoch": 0.8961748633879781,
|
| 1244 |
+
"grad_norm": 97.67105586980914,
|
| 1245 |
+
"learning_rate": 2.872885037273573e-08,
|
| 1246 |
+
"logits/chosen": -1.0626450777053833,
|
| 1247 |
+
"logits/rejected": -1.0900951623916626,
|
| 1248 |
+
"logps/chosen": -2.180873155593872,
|
| 1249 |
+
"logps/rejected": -2.9227170944213867,
|
| 1250 |
+
"loss": 1.6375,
|
| 1251 |
+
"rewards/accuracies": 0.8125,
|
| 1252 |
+
"rewards/chosen": -21.808731079101562,
|
| 1253 |
+
"rewards/margins": 7.418442726135254,
|
| 1254 |
+
"rewards/rejected": -29.2271728515625,
|
| 1255 |
+
"step": 410
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 0.907103825136612,
|
| 1259 |
+
"grad_norm": 97.42319064887056,
|
| 1260 |
+
"learning_rate": 2.2991320471108737e-08,
|
| 1261 |
+
"logits/chosen": -1.0922316312789917,
|
| 1262 |
+
"logits/rejected": -1.0891927480697632,
|
| 1263 |
+
"logps/chosen": -2.1282808780670166,
|
| 1264 |
+
"logps/rejected": -2.886338710784912,
|
| 1265 |
+
"loss": 1.8837,
|
| 1266 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1267 |
+
"rewards/chosen": -21.28281021118164,
|
| 1268 |
+
"rewards/margins": 7.580580711364746,
|
| 1269 |
+
"rewards/rejected": -28.863391876220703,
|
| 1270 |
+
"step": 415
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 0.9180327868852459,
|
| 1274 |
+
"grad_norm": 102.77428463977986,
|
| 1275 |
+
"learning_rate": 1.787743916481737e-08,
|
| 1276 |
+
"logits/chosen": -1.0903656482696533,
|
| 1277 |
+
"logits/rejected": -1.0697782039642334,
|
| 1278 |
+
"logps/chosen": -2.128669261932373,
|
| 1279 |
+
"logps/rejected": -2.8950395584106445,
|
| 1280 |
+
"loss": 1.5088,
|
| 1281 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 1282 |
+
"rewards/chosen": -21.286693572998047,
|
| 1283 |
+
"rewards/margins": 7.663704872131348,
|
| 1284 |
+
"rewards/rejected": -28.950397491455078,
|
| 1285 |
+
"step": 420
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 0.9289617486338798,
|
| 1289 |
+
"grad_norm": 100.13693396976505,
|
| 1290 |
+
"learning_rate": 1.3394675306478253e-08,
|
| 1291 |
+
"logits/chosen": -1.1003714799880981,
|
| 1292 |
+
"logits/rejected": -1.086613416671753,
|
| 1293 |
+
"logps/chosen": -2.245288372039795,
|
| 1294 |
+
"logps/rejected": -3.010887861251831,
|
| 1295 |
+
"loss": 1.7902,
|
| 1296 |
+
"rewards/accuracies": 0.8125,
|
| 1297 |
+
"rewards/chosen": -22.452878952026367,
|
| 1298 |
+
"rewards/margins": 7.655998229980469,
|
| 1299 |
+
"rewards/rejected": -30.108877182006836,
|
| 1300 |
+
"step": 425
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"epoch": 0.9398907103825137,
|
| 1304 |
+
"grad_norm": 105.78919470199224,
|
| 1305 |
+
"learning_rate": 9.549575998100024e-09,
|
| 1306 |
+
"logits/chosen": -1.0245463848114014,
|
| 1307 |
+
"logits/rejected": -1.0208485126495361,
|
| 1308 |
+
"logps/chosen": -1.9884769916534424,
|
| 1309 |
+
"logps/rejected": -2.7820286750793457,
|
| 1310 |
+
"loss": 1.4019,
|
| 1311 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1312 |
+
"rewards/chosen": -19.884769439697266,
|
| 1313 |
+
"rewards/margins": 7.93551778793335,
|
| 1314 |
+
"rewards/rejected": -27.820287704467773,
|
| 1315 |
+
"step": 430
|
| 1316 |
+
},
|
| 1317 |
+
{
|
| 1318 |
+
"epoch": 0.9508196721311475,
|
| 1319 |
+
"grad_norm": 106.12768935225566,
|
| 1320 |
+
"learning_rate": 6.3477570290039974e-09,
|
| 1321 |
+
"logits/chosen": -1.0841343402862549,
|
| 1322 |
+
"logits/rejected": -1.108457326889038,
|
| 1323 |
+
"logps/chosen": -2.172492027282715,
|
| 1324 |
+
"logps/rejected": -2.936332941055298,
|
| 1325 |
+
"loss": 1.6336,
|
| 1326 |
+
"rewards/accuracies": 0.875,
|
| 1327 |
+
"rewards/chosen": -21.724918365478516,
|
| 1328 |
+
"rewards/margins": 7.6384100914001465,
|
| 1329 |
+
"rewards/rejected": -29.363330841064453,
|
| 1330 |
+
"step": 435
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"epoch": 0.9617486338797814,
|
| 1334 |
+
"grad_norm": 120.9095518273233,
|
| 1335 |
+
"learning_rate": 3.7938946739319575e-09,
|
| 1336 |
+
"logits/chosen": -1.0631051063537598,
|
| 1337 |
+
"logits/rejected": -1.0275993347167969,
|
| 1338 |
+
"logps/chosen": -2.0968384742736816,
|
| 1339 |
+
"logps/rejected": -2.6874489784240723,
|
| 1340 |
+
"loss": 1.6067,
|
| 1341 |
+
"rewards/accuracies": 0.78125,
|
| 1342 |
+
"rewards/chosen": -20.968387603759766,
|
| 1343 |
+
"rewards/margins": 5.90610408782959,
|
| 1344 |
+
"rewards/rejected": -26.874492645263672,
|
| 1345 |
+
"step": 440
|
| 1346 |
+
},
|
| 1347 |
+
{
|
| 1348 |
+
"epoch": 0.9726775956284153,
|
| 1349 |
+
"grad_norm": 123.02986222538128,
|
| 1350 |
+
"learning_rate": 1.8917188633203684e-09,
|
| 1351 |
+
"logits/chosen": -1.0975539684295654,
|
| 1352 |
+
"logits/rejected": -1.0790989398956299,
|
| 1353 |
+
"logps/chosen": -2.1089742183685303,
|
| 1354 |
+
"logps/rejected": -2.8517839908599854,
|
| 1355 |
+
"loss": 1.6175,
|
| 1356 |
+
"rewards/accuracies": 0.8125,
|
| 1357 |
+
"rewards/chosen": -21.08974266052246,
|
| 1358 |
+
"rewards/margins": 7.428097724914551,
|
| 1359 |
+
"rewards/rejected": -28.517841339111328,
|
| 1360 |
+
"step": 445
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 0.9836065573770492,
|
| 1364 |
+
"grad_norm": 138.13735363884834,
|
| 1365 |
+
"learning_rate": 6.440077357159901e-10,
|
| 1366 |
+
"logits/chosen": -1.0183210372924805,
|
| 1367 |
+
"logits/rejected": -1.0120489597320557,
|
| 1368 |
+
"logps/chosen": -2.1649553775787354,
|
| 1369 |
+
"logps/rejected": -2.8207600116729736,
|
| 1370 |
+
"loss": 1.6741,
|
| 1371 |
+
"rewards/accuracies": 0.856249988079071,
|
| 1372 |
+
"rewards/chosen": -21.649555206298828,
|
| 1373 |
+
"rewards/margins": 6.558046817779541,
|
| 1374 |
+
"rewards/rejected": -28.207599639892578,
|
| 1375 |
+
"step": 450
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 0.994535519125683,
|
| 1379 |
+
"grad_norm": 133.29330256302666,
|
| 1380 |
+
"learning_rate": 5.258358028848442e-11,
|
| 1381 |
+
"logits/chosen": -1.0393953323364258,
|
| 1382 |
+
"logits/rejected": -1.0296013355255127,
|
| 1383 |
+
"logps/chosen": -2.2563390731811523,
|
| 1384 |
+
"logps/rejected": -2.771131992340088,
|
| 1385 |
+
"loss": 2.0445,
|
| 1386 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 1387 |
+
"rewards/chosen": -22.56338882446289,
|
| 1388 |
+
"rewards/margins": 5.1479291915893555,
|
| 1389 |
+
"rewards/rejected": -27.711318969726562,
|
| 1390 |
+
"step": 455
|
| 1391 |
+
},
|
| 1392 |
+
{
|
| 1393 |
+
"epoch": 0.9989071038251366,
|
| 1394 |
+
"step": 457,
|
| 1395 |
+
"total_flos": 0.0,
|
| 1396 |
+
"train_loss": 2.442784007842483,
|
| 1397 |
+
"train_runtime": 5988.2154,
|
| 1398 |
+
"train_samples_per_second": 9.779,
|
| 1399 |
+
"train_steps_per_second": 0.076
|
| 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 |
+
}
|