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_-orpo-10-0_5-1e-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_-orpo-10-0_5-1e-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: 2.4129
|
| 22 |
+
- Rewards/chosen: -11.3049
|
| 23 |
+
- Rewards/rejected: -15.3895
|
| 24 |
+
- Rewards/accuracies: 0.8133
|
| 25 |
+
- Rewards/margins: 4.0846
|
| 26 |
+
- Logps/rejected: -1.5389
|
| 27 |
+
- Logps/chosen: -1.1305
|
| 28 |
+
- Logits/rejected: -1.3507
|
| 29 |
+
- Logits/chosen: -1.3829
|
| 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 |
+
| 2.1949 | 0.8743 | 400 | 2.4129 | -11.3049 | -15.3895 | 0.8133 | 4.0846 | -1.5389 | -1.1305 | -1.3507 | -1.3829 |
|
| 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": 3.252214796955267,
|
| 5 |
+
"train_runtime": 5969.5229,
|
| 6 |
+
"train_samples": 58558,
|
| 7 |
+
"train_samples_per_second": 9.809,
|
| 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": 3.252214796955267,
|
| 5 |
+
"train_runtime": 5969.5229,
|
| 6 |
+
"train_samples": 58558,
|
| 7 |
+
"train_samples_per_second": 9.809,
|
| 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": 353.8300178137516,
|
| 14 |
+
"learning_rate": 1.0869565217391303e-07,
|
| 15 |
+
"logits/chosen": -1.0153186321258545,
|
| 16 |
+
"logits/rejected": -1.0100535154342651,
|
| 17 |
+
"logps/chosen": -0.28067928552627563,
|
| 18 |
+
"logps/rejected": -0.2859733998775482,
|
| 19 |
+
"loss": 5.0876,
|
| 20 |
+
"rewards/accuracies": 0.53125,
|
| 21 |
+
"rewards/chosen": -2.806792974472046,
|
| 22 |
+
"rewards/margins": 0.05294132977724075,
|
| 23 |
+
"rewards/rejected": -2.859734296798706,
|
| 24 |
+
"step": 5
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.02185792349726776,
|
| 28 |
+
"grad_norm": 225.98953916688063,
|
| 29 |
+
"learning_rate": 2.1739130434782607e-07,
|
| 30 |
+
"logits/chosen": -1.058803677558899,
|
| 31 |
+
"logits/rejected": -1.0075746774673462,
|
| 32 |
+
"logps/chosen": -0.25702008605003357,
|
| 33 |
+
"logps/rejected": -0.27079910039901733,
|
| 34 |
+
"loss": 5.0481,
|
| 35 |
+
"rewards/accuracies": 0.543749988079071,
|
| 36 |
+
"rewards/chosen": -2.5702006816864014,
|
| 37 |
+
"rewards/margins": 0.13779030740261078,
|
| 38 |
+
"rewards/rejected": -2.707991123199463,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.03278688524590164,
|
| 43 |
+
"grad_norm": 256.50567074528465,
|
| 44 |
+
"learning_rate": 3.260869565217391e-07,
|
| 45 |
+
"logits/chosen": -0.9965429306030273,
|
| 46 |
+
"logits/rejected": -0.9519298672676086,
|
| 47 |
+
"logps/chosen": -0.2673397660255432,
|
| 48 |
+
"logps/rejected": -0.27320751547813416,
|
| 49 |
+
"loss": 4.9705,
|
| 50 |
+
"rewards/accuracies": 0.512499988079071,
|
| 51 |
+
"rewards/chosen": -2.6733975410461426,
|
| 52 |
+
"rewards/margins": 0.05867765098810196,
|
| 53 |
+
"rewards/rejected": -2.7320752143859863,
|
| 54 |
+
"step": 15
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.04371584699453552,
|
| 58 |
+
"grad_norm": 346.54496281760004,
|
| 59 |
+
"learning_rate": 4.3478260869565214e-07,
|
| 60 |
+
"logits/chosen": -0.9495851397514343,
|
| 61 |
+
"logits/rejected": -0.8978370428085327,
|
| 62 |
+
"logps/chosen": -0.2726512551307678,
|
| 63 |
+
"logps/rejected": -0.28480401635169983,
|
| 64 |
+
"loss": 5.0659,
|
| 65 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 66 |
+
"rewards/chosen": -2.7265126705169678,
|
| 67 |
+
"rewards/margins": 0.12152715772390366,
|
| 68 |
+
"rewards/rejected": -2.8480401039123535,
|
| 69 |
+
"step": 20
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.0546448087431694,
|
| 73 |
+
"grad_norm": 308.4996786317866,
|
| 74 |
+
"learning_rate": 5.434782608695652e-07,
|
| 75 |
+
"logits/chosen": -0.9501255750656128,
|
| 76 |
+
"logits/rejected": -0.8753129839897156,
|
| 77 |
+
"logps/chosen": -0.27574610710144043,
|
| 78 |
+
"logps/rejected": -0.29435569047927856,
|
| 79 |
+
"loss": 4.9704,
|
| 80 |
+
"rewards/accuracies": 0.581250011920929,
|
| 81 |
+
"rewards/chosen": -2.7574610710144043,
|
| 82 |
+
"rewards/margins": 0.18609575927257538,
|
| 83 |
+
"rewards/rejected": -2.943556547164917,
|
| 84 |
+
"step": 25
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.06557377049180328,
|
| 88 |
+
"grad_norm": 310.0691471680414,
|
| 89 |
+
"learning_rate": 6.521739130434782e-07,
|
| 90 |
+
"logits/chosen": -1.0519192218780518,
|
| 91 |
+
"logits/rejected": -0.986653208732605,
|
| 92 |
+
"logps/chosen": -0.26742228865623474,
|
| 93 |
+
"logps/rejected": -0.28410372138023376,
|
| 94 |
+
"loss": 5.0602,
|
| 95 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 96 |
+
"rewards/chosen": -2.6742231845855713,
|
| 97 |
+
"rewards/margins": 0.1668141633272171,
|
| 98 |
+
"rewards/rejected": -2.8410372734069824,
|
| 99 |
+
"step": 30
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.07650273224043716,
|
| 103 |
+
"grad_norm": 189.7575115104871,
|
| 104 |
+
"learning_rate": 7.608695652173913e-07,
|
| 105 |
+
"logits/chosen": -0.9969251751899719,
|
| 106 |
+
"logits/rejected": -0.9315601587295532,
|
| 107 |
+
"logps/chosen": -0.25781017541885376,
|
| 108 |
+
"logps/rejected": -0.2797554135322571,
|
| 109 |
+
"loss": 4.9378,
|
| 110 |
+
"rewards/accuracies": 0.48124998807907104,
|
| 111 |
+
"rewards/chosen": -2.578101634979248,
|
| 112 |
+
"rewards/margins": 0.21945223212242126,
|
| 113 |
+
"rewards/rejected": -2.797553777694702,
|
| 114 |
+
"step": 35
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.08743169398907104,
|
| 118 |
+
"grad_norm": 225.69610863472298,
|
| 119 |
+
"learning_rate": 8.695652173913043e-07,
|
| 120 |
+
"logits/chosen": -0.9572874307632446,
|
| 121 |
+
"logits/rejected": -0.8963949084281921,
|
| 122 |
+
"logps/chosen": -0.28224068880081177,
|
| 123 |
+
"logps/rejected": -0.2995293140411377,
|
| 124 |
+
"loss": 5.1508,
|
| 125 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 126 |
+
"rewards/chosen": -2.82240629196167,
|
| 127 |
+
"rewards/margins": 0.17288652062416077,
|
| 128 |
+
"rewards/rejected": -2.995293140411377,
|
| 129 |
+
"step": 40
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.09836065573770492,
|
| 133 |
+
"grad_norm": 131.95682988894754,
|
| 134 |
+
"learning_rate": 9.782608695652173e-07,
|
| 135 |
+
"logits/chosen": -1.0119043588638306,
|
| 136 |
+
"logits/rejected": -0.9314481019973755,
|
| 137 |
+
"logps/chosen": -0.2863091230392456,
|
| 138 |
+
"logps/rejected": -0.3087444305419922,
|
| 139 |
+
"loss": 4.7676,
|
| 140 |
+
"rewards/accuracies": 0.5,
|
| 141 |
+
"rewards/chosen": -2.863091230392456,
|
| 142 |
+
"rewards/margins": 0.2243528664112091,
|
| 143 |
+
"rewards/rejected": -3.0874440670013428,
|
| 144 |
+
"step": 45
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.1092896174863388,
|
| 148 |
+
"grad_norm": 345.07474799660855,
|
| 149 |
+
"learning_rate": 9.997663088532014e-07,
|
| 150 |
+
"logits/chosen": -0.9598929286003113,
|
| 151 |
+
"logits/rejected": -0.8789777755737305,
|
| 152 |
+
"logps/chosen": -0.2817566692829132,
|
| 153 |
+
"logps/rejected": -0.2901131510734558,
|
| 154 |
+
"loss": 4.9269,
|
| 155 |
+
"rewards/accuracies": 0.48124998807907104,
|
| 156 |
+
"rewards/chosen": -2.817566394805908,
|
| 157 |
+
"rewards/margins": 0.08356456458568573,
|
| 158 |
+
"rewards/rejected": -2.9011311531066895,
|
| 159 |
+
"step": 50
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.12021857923497267,
|
| 163 |
+
"grad_norm": 162.76427513659027,
|
| 164 |
+
"learning_rate": 9.98817312944725e-07,
|
| 165 |
+
"logits/chosen": -0.9862833023071289,
|
| 166 |
+
"logits/rejected": -0.8716124296188354,
|
| 167 |
+
"logps/chosen": -0.28066322207450867,
|
| 168 |
+
"logps/rejected": -0.31615525484085083,
|
| 169 |
+
"loss": 4.707,
|
| 170 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 171 |
+
"rewards/chosen": -2.8066320419311523,
|
| 172 |
+
"rewards/margins": 0.35492032766342163,
|
| 173 |
+
"rewards/rejected": -3.161552667617798,
|
| 174 |
+
"step": 55
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.13114754098360656,
|
| 178 |
+
"grad_norm": 119.0877915995248,
|
| 179 |
+
"learning_rate": 9.971397915250336e-07,
|
| 180 |
+
"logits/chosen": -1.0102354288101196,
|
| 181 |
+
"logits/rejected": -0.9650627970695496,
|
| 182 |
+
"logps/chosen": -0.2704823911190033,
|
| 183 |
+
"logps/rejected": -0.31392908096313477,
|
| 184 |
+
"loss": 4.4957,
|
| 185 |
+
"rewards/accuracies": 0.612500011920929,
|
| 186 |
+
"rewards/chosen": -2.7048239707946777,
|
| 187 |
+
"rewards/margins": 0.4344666600227356,
|
| 188 |
+
"rewards/rejected": -3.1392903327941895,
|
| 189 |
+
"step": 60
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.14207650273224043,
|
| 193 |
+
"grad_norm": 214.6265229333486,
|
| 194 |
+
"learning_rate": 9.94736194623663e-07,
|
| 195 |
+
"logits/chosen": -1.0099595785140991,
|
| 196 |
+
"logits/rejected": -0.939832866191864,
|
| 197 |
+
"logps/chosen": -0.3105173707008362,
|
| 198 |
+
"logps/rejected": -0.34290796518325806,
|
| 199 |
+
"loss": 4.8297,
|
| 200 |
+
"rewards/accuracies": 0.53125,
|
| 201 |
+
"rewards/chosen": -3.1051735877990723,
|
| 202 |
+
"rewards/margins": 0.3239057958126068,
|
| 203 |
+
"rewards/rejected": -3.42907977104187,
|
| 204 |
+
"step": 65
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.15300546448087432,
|
| 208 |
+
"grad_norm": 473.8808962259616,
|
| 209 |
+
"learning_rate": 9.916100327075037e-07,
|
| 210 |
+
"logits/chosen": -0.9457240104675293,
|
| 211 |
+
"logits/rejected": -0.9267830848693848,
|
| 212 |
+
"logps/chosen": -0.29852330684661865,
|
| 213 |
+
"logps/rejected": -0.32594314217567444,
|
| 214 |
+
"loss": 4.6136,
|
| 215 |
+
"rewards/accuracies": 0.581250011920929,
|
| 216 |
+
"rewards/chosen": -2.9852328300476074,
|
| 217 |
+
"rewards/margins": 0.2741985023021698,
|
| 218 |
+
"rewards/rejected": -3.2594313621520996,
|
| 219 |
+
"step": 70
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.16393442622950818,
|
| 223 |
+
"grad_norm": 123.02284060076431,
|
| 224 |
+
"learning_rate": 9.877658715537428e-07,
|
| 225 |
+
"logits/chosen": -0.9411904215812683,
|
| 226 |
+
"logits/rejected": -0.9241803288459778,
|
| 227 |
+
"logps/chosen": -0.31740230321884155,
|
| 228 |
+
"logps/rejected": -0.351471483707428,
|
| 229 |
+
"loss": 4.6792,
|
| 230 |
+
"rewards/accuracies": 0.625,
|
| 231 |
+
"rewards/chosen": -3.174023389816284,
|
| 232 |
+
"rewards/margins": 0.3406919836997986,
|
| 233 |
+
"rewards/rejected": -3.5147147178649902,
|
| 234 |
+
"step": 75
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.17486338797814208,
|
| 238 |
+
"grad_norm": 136.7631938838273,
|
| 239 |
+
"learning_rate": 9.832093255815216e-07,
|
| 240 |
+
"logits/chosen": -0.9370519518852234,
|
| 241 |
+
"logits/rejected": -0.8748278617858887,
|
| 242 |
+
"logps/chosen": -0.31292015314102173,
|
| 243 |
+
"logps/rejected": -0.33769917488098145,
|
| 244 |
+
"loss": 4.6083,
|
| 245 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 246 |
+
"rewards/chosen": -3.1292014122009277,
|
| 247 |
+
"rewards/margins": 0.2477904111146927,
|
| 248 |
+
"rewards/rejected": -3.3769919872283936,
|
| 249 |
+
"step": 80
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.18579234972677597,
|
| 253 |
+
"grad_norm": 198.11968829037178,
|
| 254 |
+
"learning_rate": 9.779470496520441e-07,
|
| 255 |
+
"logits/chosen": -0.9416698217391968,
|
| 256 |
+
"logits/rejected": -0.891922116279602,
|
| 257 |
+
"logps/chosen": -0.3160974383354187,
|
| 258 |
+
"logps/rejected": -0.3771803379058838,
|
| 259 |
+
"loss": 4.4814,
|
| 260 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 261 |
+
"rewards/chosen": -3.1609745025634766,
|
| 262 |
+
"rewards/margins": 0.6108287572860718,
|
| 263 |
+
"rewards/rejected": -3.7718029022216797,
|
| 264 |
+
"step": 85
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.19672131147540983,
|
| 268 |
+
"grad_norm": 145.2843942098038,
|
| 269 |
+
"learning_rate": 9.719867293491144e-07,
|
| 270 |
+
"logits/chosen": -1.0085046291351318,
|
| 271 |
+
"logits/rejected": -0.9282514452934265,
|
| 272 |
+
"logps/chosen": -0.3432404100894928,
|
| 273 |
+
"logps/rejected": -0.381814569234848,
|
| 274 |
+
"loss": 4.5692,
|
| 275 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 276 |
+
"rewards/chosen": -3.4324042797088623,
|
| 277 |
+
"rewards/margins": 0.38574135303497314,
|
| 278 |
+
"rewards/rejected": -3.818145275115967,
|
| 279 |
+
"step": 90
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.20765027322404372,
|
| 283 |
+
"grad_norm": 224.8728680422599,
|
| 284 |
+
"learning_rate": 9.653370697542987e-07,
|
| 285 |
+
"logits/chosen": -0.9396700859069824,
|
| 286 |
+
"logits/rejected": -0.9407837986946106,
|
| 287 |
+
"logps/chosen": -0.3434152603149414,
|
| 288 |
+
"logps/rejected": -0.3718097507953644,
|
| 289 |
+
"loss": 4.3339,
|
| 290 |
+
"rewards/accuracies": 0.59375,
|
| 291 |
+
"rewards/chosen": -3.434152603149414,
|
| 292 |
+
"rewards/margins": 0.2839447855949402,
|
| 293 |
+
"rewards/rejected": -3.718097686767578,
|
| 294 |
+
"step": 95
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.2185792349726776,
|
| 298 |
+
"grad_norm": 200.93333735175037,
|
| 299 |
+
"learning_rate": 9.580077827331037e-07,
|
| 300 |
+
"logits/chosen": -0.9491629600524902,
|
| 301 |
+
"logits/rejected": -0.9065796136856079,
|
| 302 |
+
"logps/chosen": -0.38279426097869873,
|
| 303 |
+
"logps/rejected": -0.4457218050956726,
|
| 304 |
+
"loss": 4.2739,
|
| 305 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 306 |
+
"rewards/chosen": -3.8279430866241455,
|
| 307 |
+
"rewards/margins": 0.6292752027511597,
|
| 308 |
+
"rewards/rejected": -4.457218170166016,
|
| 309 |
+
"step": 100
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.22950819672131148,
|
| 313 |
+
"grad_norm": 166.91104379384572,
|
| 314 |
+
"learning_rate": 9.500095727507419e-07,
|
| 315 |
+
"logits/chosen": -1.0131243467330933,
|
| 316 |
+
"logits/rejected": -0.9821707606315613,
|
| 317 |
+
"logps/chosen": -0.36331382393836975,
|
| 318 |
+
"logps/rejected": -0.416962206363678,
|
| 319 |
+
"loss": 4.2844,
|
| 320 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 321 |
+
"rewards/chosen": -3.6331381797790527,
|
| 322 |
+
"rewards/margins": 0.5364840030670166,
|
| 323 |
+
"rewards/rejected": -4.16962194442749,
|
| 324 |
+
"step": 105
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.24043715846994534,
|
| 328 |
+
"grad_norm": 152.87786136664803,
|
| 329 |
+
"learning_rate": 9.413541212382004e-07,
|
| 330 |
+
"logits/chosen": -1.0002071857452393,
|
| 331 |
+
"logits/rejected": -0.9825676083564758,
|
| 332 |
+
"logps/chosen": -0.3780575394630432,
|
| 333 |
+
"logps/rejected": -0.46578970551490784,
|
| 334 |
+
"loss": 4.2075,
|
| 335 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 336 |
+
"rewards/chosen": -3.7805755138397217,
|
| 337 |
+
"rewards/margins": 0.8773216009140015,
|
| 338 |
+
"rewards/rejected": -4.657896995544434,
|
| 339 |
+
"step": 110
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.25136612021857924,
|
| 343 |
+
"grad_norm": 157.4597077495282,
|
| 344 |
+
"learning_rate": 9.320540695314438e-07,
|
| 345 |
+
"logits/chosen": -1.013012170791626,
|
| 346 |
+
"logits/rejected": -0.9737430810928345,
|
| 347 |
+
"logps/chosen": -0.38077500462532043,
|
| 348 |
+
"logps/rejected": -0.4924758970737457,
|
| 349 |
+
"loss": 4.0012,
|
| 350 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 351 |
+
"rewards/chosen": -3.8077499866485596,
|
| 352 |
+
"rewards/margins": 1.1170084476470947,
|
| 353 |
+
"rewards/rejected": -4.924757957458496,
|
| 354 |
+
"step": 115
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.26229508196721313,
|
| 358 |
+
"grad_norm": 125.21327412714959,
|
| 359 |
+
"learning_rate": 9.221230004086721e-07,
|
| 360 |
+
"logits/chosen": -1.029721975326538,
|
| 361 |
+
"logits/rejected": -0.9594861268997192,
|
| 362 |
+
"logps/chosen": -0.3884691596031189,
|
| 363 |
+
"logps/rejected": -0.4513280987739563,
|
| 364 |
+
"loss": 4.1127,
|
| 365 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 366 |
+
"rewards/chosen": -3.8846917152404785,
|
| 367 |
+
"rewards/margins": 0.6285887956619263,
|
| 368 |
+
"rewards/rejected": -4.513280868530273,
|
| 369 |
+
"step": 120
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.273224043715847,
|
| 373 |
+
"grad_norm": 285.25729007443874,
|
| 374 |
+
"learning_rate": 9.11575418252596e-07,
|
| 375 |
+
"logits/chosen": -0.9322704076766968,
|
| 376 |
+
"logits/rejected": -0.8982660174369812,
|
| 377 |
+
"logps/chosen": -0.4055355489253998,
|
| 378 |
+
"logps/rejected": -0.4851369261741638,
|
| 379 |
+
"loss": 3.8762,
|
| 380 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 381 |
+
"rewards/chosen": -4.055355548858643,
|
| 382 |
+
"rewards/margins": 0.7960137128829956,
|
| 383 |
+
"rewards/rejected": -4.8513689041137695,
|
| 384 |
+
"step": 125
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.28415300546448086,
|
| 388 |
+
"grad_norm": 140.757752850241,
|
| 389 |
+
"learning_rate": 9.004267278667031e-07,
|
| 390 |
+
"logits/chosen": -0.9658416509628296,
|
| 391 |
+
"logits/rejected": -0.9603641629219055,
|
| 392 |
+
"logps/chosen": -0.4266030788421631,
|
| 393 |
+
"logps/rejected": -0.5589009523391724,
|
| 394 |
+
"loss": 3.9144,
|
| 395 |
+
"rewards/accuracies": 0.737500011920929,
|
| 396 |
+
"rewards/chosen": -4.266031265258789,
|
| 397 |
+
"rewards/margins": 1.3229780197143555,
|
| 398 |
+
"rewards/rejected": -5.5890092849731445,
|
| 399 |
+
"step": 130
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.29508196721311475,
|
| 403 |
+
"grad_norm": 112.11614815018599,
|
| 404 |
+
"learning_rate": 8.886932119764565e-07,
|
| 405 |
+
"logits/chosen": -1.0118809938430786,
|
| 406 |
+
"logits/rejected": -0.9286627769470215,
|
| 407 |
+
"logps/chosen": -0.41649705171585083,
|
| 408 |
+
"logps/rejected": -0.5318835973739624,
|
| 409 |
+
"loss": 3.7328,
|
| 410 |
+
"rewards/accuracies": 0.675000011920929,
|
| 411 |
+
"rewards/chosen": -4.164970397949219,
|
| 412 |
+
"rewards/margins": 1.1538660526275635,
|
| 413 |
+
"rewards/rejected": -5.318836212158203,
|
| 414 |
+
"step": 135
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.30601092896174864,
|
| 418 |
+
"grad_norm": 151.56112609092273,
|
| 419 |
+
"learning_rate": 8.763920074482809e-07,
|
| 420 |
+
"logits/chosen": -1.00692880153656,
|
| 421 |
+
"logits/rejected": -0.9528873562812805,
|
| 422 |
+
"logps/chosen": -0.4440224766731262,
|
| 423 |
+
"logps/rejected": -0.5973733067512512,
|
| 424 |
+
"loss": 3.3329,
|
| 425 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 426 |
+
"rewards/chosen": -4.440225124359131,
|
| 427 |
+
"rewards/margins": 1.5335077047348022,
|
| 428 |
+
"rewards/rejected": -5.9737324714660645,
|
| 429 |
+
"step": 140
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.31693989071038253,
|
| 433 |
+
"grad_norm": 157.69245375817547,
|
| 434 |
+
"learning_rate": 8.635410802610723e-07,
|
| 435 |
+
"logits/chosen": -1.0103561878204346,
|
| 436 |
+
"logits/rejected": -0.9901880025863647,
|
| 437 |
+
"logps/chosen": -0.4360678195953369,
|
| 438 |
+
"logps/rejected": -0.5111598372459412,
|
| 439 |
+
"loss": 3.5759,
|
| 440 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 441 |
+
"rewards/chosen": -4.360678672790527,
|
| 442 |
+
"rewards/margins": 0.7509199380874634,
|
| 443 |
+
"rewards/rejected": -5.111598014831543,
|
| 444 |
+
"step": 145
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.32786885245901637,
|
| 448 |
+
"grad_norm": 131.42091257219136,
|
| 449 |
+
"learning_rate": 8.501591992667849e-07,
|
| 450 |
+
"logits/chosen": -1.0559594631195068,
|
| 451 |
+
"logits/rejected": -1.0228536128997803,
|
| 452 |
+
"logps/chosen": -0.47138285636901855,
|
| 453 |
+
"logps/rejected": -0.6517553329467773,
|
| 454 |
+
"loss": 3.4348,
|
| 455 |
+
"rewards/accuracies": 0.762499988079071,
|
| 456 |
+
"rewards/chosen": -4.7138285636901855,
|
| 457 |
+
"rewards/margins": 1.8037246465682983,
|
| 458 |
+
"rewards/rejected": -6.517552852630615,
|
| 459 |
+
"step": 150
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.33879781420765026,
|
| 463 |
+
"grad_norm": 121.46660804843347,
|
| 464 |
+
"learning_rate": 8.362659087784152e-07,
|
| 465 |
+
"logits/chosen": -1.007810354232788,
|
| 466 |
+
"logits/rejected": -0.952292799949646,
|
| 467 |
+
"logps/chosen": -0.4735150933265686,
|
| 468 |
+
"logps/rejected": -0.591022789478302,
|
| 469 |
+
"loss": 3.4613,
|
| 470 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 471 |
+
"rewards/chosen": -4.735150337219238,
|
| 472 |
+
"rewards/margins": 1.175077199935913,
|
| 473 |
+
"rewards/rejected": -5.9102277755737305,
|
| 474 |
+
"step": 155
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.34972677595628415,
|
| 478 |
+
"grad_norm": 126.72976057892085,
|
| 479 |
+
"learning_rate": 8.218815000254231e-07,
|
| 480 |
+
"logits/chosen": -1.0282104015350342,
|
| 481 |
+
"logits/rejected": -0.9731351733207703,
|
| 482 |
+
"logps/chosen": -0.5363224744796753,
|
| 483 |
+
"logps/rejected": -0.640962541103363,
|
| 484 |
+
"loss": 3.4898,
|
| 485 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 486 |
+
"rewards/chosen": -5.363224983215332,
|
| 487 |
+
"rewards/margins": 1.046400785446167,
|
| 488 |
+
"rewards/rejected": -6.409626007080078,
|
| 489 |
+
"step": 160
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.36065573770491804,
|
| 493 |
+
"grad_norm": 153.56145857916775,
|
| 494 |
+
"learning_rate": 8.07026981518276e-07,
|
| 495 |
+
"logits/chosen": -1.0256257057189941,
|
| 496 |
+
"logits/rejected": -0.9699680209159851,
|
| 497 |
+
"logps/chosen": -0.5307550430297852,
|
| 498 |
+
"logps/rejected": -0.6452345848083496,
|
| 499 |
+
"loss": 3.3879,
|
| 500 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 501 |
+
"rewards/chosen": -5.30755090713501,
|
| 502 |
+
"rewards/margins": 1.1447944641113281,
|
| 503 |
+
"rewards/rejected": -6.452345371246338,
|
| 504 |
+
"step": 165
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 0.37158469945355194,
|
| 508 |
+
"grad_norm": 133.9853822185735,
|
| 509 |
+
"learning_rate": 7.917240483654e-07,
|
| 510 |
+
"logits/chosen": -1.0185896158218384,
|
| 511 |
+
"logits/rejected": -0.9558917880058289,
|
| 512 |
+
"logps/chosen": -0.5270066857337952,
|
| 513 |
+
"logps/rejected": -0.6334024667739868,
|
| 514 |
+
"loss": 3.5392,
|
| 515 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 516 |
+
"rewards/chosen": -5.270066261291504,
|
| 517 |
+
"rewards/margins": 1.0639582872390747,
|
| 518 |
+
"rewards/rejected": -6.3340253829956055,
|
| 519 |
+
"step": 170
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.3825136612021858,
|
| 523 |
+
"grad_norm": 170.7287592983698,
|
| 524 |
+
"learning_rate": 7.759950505873521e-07,
|
| 525 |
+
"logits/chosen": -1.0673246383666992,
|
| 526 |
+
"logits/rejected": -1.0361274480819702,
|
| 527 |
+
"logps/chosen": -0.558329701423645,
|
| 528 |
+
"logps/rejected": -0.6510181427001953,
|
| 529 |
+
"loss": 3.3022,
|
| 530 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 531 |
+
"rewards/chosen": -5.583296775817871,
|
| 532 |
+
"rewards/margins": 0.9268839955329895,
|
| 533 |
+
"rewards/rejected": -6.510180473327637,
|
| 534 |
+
"step": 175
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.39344262295081966,
|
| 538 |
+
"grad_norm": 134.7756954954796,
|
| 539 |
+
"learning_rate": 7.598629604744872e-07,
|
| 540 |
+
"logits/chosen": -1.0714390277862549,
|
| 541 |
+
"logits/rejected": -1.0629351139068604,
|
| 542 |
+
"logps/chosen": -0.5433841347694397,
|
| 543 |
+
"logps/rejected": -0.7269644141197205,
|
| 544 |
+
"loss": 3.1167,
|
| 545 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 546 |
+
"rewards/chosen": -5.433840751647949,
|
| 547 |
+
"rewards/margins": 1.8358032703399658,
|
| 548 |
+
"rewards/rejected": -7.269643306732178,
|
| 549 |
+
"step": 180
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.40437158469945356,
|
| 553 |
+
"grad_norm": 150.12430395788675,
|
| 554 |
+
"learning_rate": 7.433513390357989e-07,
|
| 555 |
+
"logits/chosen": -1.1003663539886475,
|
| 556 |
+
"logits/rejected": -1.1102244853973389,
|
| 557 |
+
"logps/chosen": -0.5805007219314575,
|
| 558 |
+
"logps/rejected": -0.7595118880271912,
|
| 559 |
+
"loss": 3.1145,
|
| 560 |
+
"rewards/accuracies": 0.800000011920929,
|
| 561 |
+
"rewards/chosen": -5.805006980895996,
|
| 562 |
+
"rewards/margins": 1.7901118993759155,
|
| 563 |
+
"rewards/rejected": -7.595119476318359,
|
| 564 |
+
"step": 185
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.41530054644808745,
|
| 568 |
+
"grad_norm": 174.07312409793005,
|
| 569 |
+
"learning_rate": 7.264843015879321e-07,
|
| 570 |
+
"logits/chosen": -1.1195979118347168,
|
| 571 |
+
"logits/rejected": -1.0757617950439453,
|
| 572 |
+
"logps/chosen": -0.5829612612724304,
|
| 573 |
+
"logps/rejected": -0.7835941314697266,
|
| 574 |
+
"loss": 3.1898,
|
| 575 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 576 |
+
"rewards/chosen": -5.829612731933594,
|
| 577 |
+
"rewards/margins": 2.00632905960083,
|
| 578 |
+
"rewards/rejected": -7.835942268371582,
|
| 579 |
+
"step": 190
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.4262295081967213,
|
| 583 |
+
"grad_norm": 146.25540915098478,
|
| 584 |
+
"learning_rate": 7.092864825346266e-07,
|
| 585 |
+
"logits/chosen": -1.1447865962982178,
|
| 586 |
+
"logits/rejected": -1.1201746463775635,
|
| 587 |
+
"logps/chosen": -0.6956247091293335,
|
| 588 |
+
"logps/rejected": -0.9184806942939758,
|
| 589 |
+
"loss": 3.2154,
|
| 590 |
+
"rewards/accuracies": 0.762499988079071,
|
| 591 |
+
"rewards/chosen": -6.956246852874756,
|
| 592 |
+
"rewards/margins": 2.2285592555999756,
|
| 593 |
+
"rewards/rejected": -9.184806823730469,
|
| 594 |
+
"step": 195
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 0.4371584699453552,
|
| 598 |
+
"grad_norm": 143.20432181294757,
|
| 599 |
+
"learning_rate": 6.917829993880302e-07,
|
| 600 |
+
"logits/chosen": -1.1187337636947632,
|
| 601 |
+
"logits/rejected": -1.0346059799194336,
|
| 602 |
+
"logps/chosen": -0.6610385179519653,
|
| 603 |
+
"logps/rejected": -0.8449029922485352,
|
| 604 |
+
"loss": 3.0607,
|
| 605 |
+
"rewards/accuracies": 0.768750011920929,
|
| 606 |
+
"rewards/chosen": -6.610384941101074,
|
| 607 |
+
"rewards/margins": 1.83864426612854,
|
| 608 |
+
"rewards/rejected": -8.449029922485352,
|
| 609 |
+
"step": 200
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.44808743169398907,
|
| 613 |
+
"grad_norm": 114.99460631566296,
|
| 614 |
+
"learning_rate": 6.739994160844309e-07,
|
| 615 |
+
"logits/chosen": -1.0897611379623413,
|
| 616 |
+
"logits/rejected": -1.1026214361190796,
|
| 617 |
+
"logps/chosen": -0.6384384632110596,
|
| 618 |
+
"logps/rejected": -0.8304228782653809,
|
| 619 |
+
"loss": 2.9101,
|
| 620 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 621 |
+
"rewards/chosen": -6.384384632110596,
|
| 622 |
+
"rewards/margins": 1.9198448657989502,
|
| 623 |
+
"rewards/rejected": -8.304228782653809,
|
| 624 |
+
"step": 205
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 0.45901639344262296,
|
| 628 |
+
"grad_norm": 208.90885124367924,
|
| 629 |
+
"learning_rate": 6.559617056479827e-07,
|
| 630 |
+
"logits/chosen": -1.1222480535507202,
|
| 631 |
+
"logits/rejected": -1.122051477432251,
|
| 632 |
+
"logps/chosen": -0.6973533630371094,
|
| 633 |
+
"logps/rejected": -0.942583441734314,
|
| 634 |
+
"loss": 2.9151,
|
| 635 |
+
"rewards/accuracies": 0.793749988079071,
|
| 636 |
+
"rewards/chosen": -6.973533630371094,
|
| 637 |
+
"rewards/margins": 2.452300548553467,
|
| 638 |
+
"rewards/rejected": -9.425833702087402,
|
| 639 |
+
"step": 210
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.46994535519125685,
|
| 643 |
+
"grad_norm": 158.10928377564565,
|
| 644 |
+
"learning_rate": 6.376962122569567e-07,
|
| 645 |
+
"logits/chosen": -1.1337159872055054,
|
| 646 |
+
"logits/rejected": -1.0800633430480957,
|
| 647 |
+
"logps/chosen": -0.7191341519355774,
|
| 648 |
+
"logps/rejected": -0.9607221484184265,
|
| 649 |
+
"loss": 3.1913,
|
| 650 |
+
"rewards/accuracies": 0.8125,
|
| 651 |
+
"rewards/chosen": -7.191342353820801,
|
| 652 |
+
"rewards/margins": 2.4158806800842285,
|
| 653 |
+
"rewards/rejected": -9.607221603393555,
|
| 654 |
+
"step": 215
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.4808743169398907,
|
| 658 |
+
"grad_norm": 120.26950487116113,
|
| 659 |
+
"learning_rate": 6.192296127679192e-07,
|
| 660 |
+
"logits/chosen": -1.1914238929748535,
|
| 661 |
+
"logits/rejected": -1.164651870727539,
|
| 662 |
+
"logps/chosen": -0.7474638223648071,
|
| 663 |
+
"logps/rejected": -0.9536181688308716,
|
| 664 |
+
"loss": 2.9159,
|
| 665 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 666 |
+
"rewards/chosen": -7.47463846206665,
|
| 667 |
+
"rewards/margins": 2.0615434646606445,
|
| 668 |
+
"rewards/rejected": -9.536182403564453,
|
| 669 |
+
"step": 220
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.4918032786885246,
|
| 673 |
+
"grad_norm": 120.68908478001487,
|
| 674 |
+
"learning_rate": 6.005888777540319e-07,
|
| 675 |
+
"logits/chosen": -1.20607328414917,
|
| 676 |
+
"logits/rejected": -1.1677665710449219,
|
| 677 |
+
"logps/chosen": -0.7626639604568481,
|
| 678 |
+
"logps/rejected": -1.0024259090423584,
|
| 679 |
+
"loss": 2.9058,
|
| 680 |
+
"rewards/accuracies": 0.824999988079071,
|
| 681 |
+
"rewards/chosen": -7.626638889312744,
|
| 682 |
+
"rewards/margins": 2.397620439529419,
|
| 683 |
+
"rewards/rejected": -10.024259567260742,
|
| 684 |
+
"step": 225
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 0.5027322404371585,
|
| 688 |
+
"grad_norm": 122.43410251633057,
|
| 689 |
+
"learning_rate": 5.818012321143773e-07,
|
| 690 |
+
"logits/chosen": -1.116031527519226,
|
| 691 |
+
"logits/rejected": -1.117067575454712,
|
| 692 |
+
"logps/chosen": -0.767257034778595,
|
| 693 |
+
"logps/rejected": -1.0256441831588745,
|
| 694 |
+
"loss": 2.9487,
|
| 695 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 696 |
+
"rewards/chosen": -7.67257022857666,
|
| 697 |
+
"rewards/margins": 2.5838723182678223,
|
| 698 |
+
"rewards/rejected": -10.256442070007324,
|
| 699 |
+
"step": 230
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.5136612021857924,
|
| 703 |
+
"grad_norm": 146.45120973143196,
|
| 704 |
+
"learning_rate": 5.628941153118388e-07,
|
| 705 |
+
"logits/chosen": -1.1330724954605103,
|
| 706 |
+
"logits/rejected": -1.0960749387741089,
|
| 707 |
+
"logps/chosen": -0.7881630659103394,
|
| 708 |
+
"logps/rejected": -1.0228068828582764,
|
| 709 |
+
"loss": 2.7419,
|
| 710 |
+
"rewards/accuracies": 0.831250011920929,
|
| 711 |
+
"rewards/chosen": -7.881631374359131,
|
| 712 |
+
"rewards/margins": 2.346438407897949,
|
| 713 |
+
"rewards/rejected": -10.228069305419922,
|
| 714 |
+
"step": 235
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 0.5245901639344263,
|
| 718 |
+
"grad_norm": 131.87657600514555,
|
| 719 |
+
"learning_rate": 5.438951412976098e-07,
|
| 720 |
+
"logits/chosen": -1.1725775003433228,
|
| 721 |
+
"logits/rejected": -1.182626485824585,
|
| 722 |
+
"logps/chosen": -0.7777260541915894,
|
| 723 |
+
"logps/rejected": -1.0662057399749756,
|
| 724 |
+
"loss": 2.5085,
|
| 725 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 726 |
+
"rewards/chosen": -7.777259826660156,
|
| 727 |
+
"rewards/margins": 2.8847975730895996,
|
| 728 |
+
"rewards/rejected": -10.66205883026123,
|
| 729 |
+
"step": 240
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.5355191256830601,
|
| 733 |
+
"grad_norm": 135.24353045590152,
|
| 734 |
+
"learning_rate": 5.248320581808619e-07,
|
| 735 |
+
"logits/chosen": -1.1081842184066772,
|
| 736 |
+
"logits/rejected": -1.0651283264160156,
|
| 737 |
+
"logps/chosen": -0.7994285225868225,
|
| 738 |
+
"logps/rejected": -1.083718180656433,
|
| 739 |
+
"loss": 2.7299,
|
| 740 |
+
"rewards/accuracies": 0.78125,
|
| 741 |
+
"rewards/chosen": -7.994284152984619,
|
| 742 |
+
"rewards/margins": 2.842897415161133,
|
| 743 |
+
"rewards/rejected": -10.837182998657227,
|
| 744 |
+
"step": 245
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.546448087431694,
|
| 748 |
+
"grad_norm": 124.21267267369478,
|
| 749 |
+
"learning_rate": 5.057327077024744e-07,
|
| 750 |
+
"logits/chosen": -1.17474365234375,
|
| 751 |
+
"logits/rejected": -1.1420042514801025,
|
| 752 |
+
"logps/chosen": -0.8092962503433228,
|
| 753 |
+
"logps/rejected": -1.037235975265503,
|
| 754 |
+
"loss": 2.8644,
|
| 755 |
+
"rewards/accuracies": 0.737500011920929,
|
| 756 |
+
"rewards/chosen": -8.092963218688965,
|
| 757 |
+
"rewards/margins": 2.279397964477539,
|
| 758 |
+
"rewards/rejected": -10.37236213684082,
|
| 759 |
+
"step": 250
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.5573770491803278,
|
| 763 |
+
"grad_norm": 234.59899677846647,
|
| 764 |
+
"learning_rate": 4.866249845720132e-07,
|
| 765 |
+
"logits/chosen": -1.1719624996185303,
|
| 766 |
+
"logits/rejected": -1.1423064470291138,
|
| 767 |
+
"logps/chosen": -0.8820212483406067,
|
| 768 |
+
"logps/rejected": -1.1843823194503784,
|
| 769 |
+
"loss": 2.6121,
|
| 770 |
+
"rewards/accuracies": 0.793749988079071,
|
| 771 |
+
"rewards/chosen": -8.820211410522461,
|
| 772 |
+
"rewards/margins": 3.0236105918884277,
|
| 773 |
+
"rewards/rejected": -11.84382438659668,
|
| 774 |
+
"step": 255
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 0.5683060109289617,
|
| 778 |
+
"grad_norm": 136.28015337191263,
|
| 779 |
+
"learning_rate": 4.675367957273505e-07,
|
| 780 |
+
"logits/chosen": -1.1379584074020386,
|
| 781 |
+
"logits/rejected": -1.1297852993011475,
|
| 782 |
+
"logps/chosen": -0.8502503633499146,
|
| 783 |
+
"logps/rejected": -1.1393508911132812,
|
| 784 |
+
"loss": 2.6642,
|
| 785 |
+
"rewards/accuracies": 0.8125,
|
| 786 |
+
"rewards/chosen": -8.502504348754883,
|
| 787 |
+
"rewards/margins": 2.8910045623779297,
|
| 788 |
+
"rewards/rejected": -11.393507957458496,
|
| 789 |
+
"step": 260
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.5792349726775956,
|
| 793 |
+
"grad_norm": 140.269461590517,
|
| 794 |
+
"learning_rate": 4.4849601957642285e-07,
|
| 795 |
+
"logits/chosen": -1.1645795106887817,
|
| 796 |
+
"logits/rejected": -1.1305253505706787,
|
| 797 |
+
"logps/chosen": -0.8733395338058472,
|
| 798 |
+
"logps/rejected": -1.1651700735092163,
|
| 799 |
+
"loss": 2.6882,
|
| 800 |
+
"rewards/accuracies": 0.800000011920929,
|
| 801 |
+
"rewards/chosen": -8.73339557647705,
|
| 802 |
+
"rewards/margins": 2.9183056354522705,
|
| 803 |
+
"rewards/rejected": -11.651700973510742,
|
| 804 |
+
"step": 265
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 0.5901639344262295,
|
| 808 |
+
"grad_norm": 141.71312924760568,
|
| 809 |
+
"learning_rate": 4.295304652806592e-07,
|
| 810 |
+
"logits/chosen": -1.159639835357666,
|
| 811 |
+
"logits/rejected": -1.138240098953247,
|
| 812 |
+
"logps/chosen": -0.8841695785522461,
|
| 813 |
+
"logps/rejected": -1.2160850763320923,
|
| 814 |
+
"loss": 2.5159,
|
| 815 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 816 |
+
"rewards/chosen": -8.841692924499512,
|
| 817 |
+
"rewards/margins": 3.3191559314727783,
|
| 818 |
+
"rewards/rejected": -12.160849571228027,
|
| 819 |
+
"step": 270
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 0.6010928961748634,
|
| 823 |
+
"grad_norm": 110.3486106272621,
|
| 824 |
+
"learning_rate": 4.106678321395433e-07,
|
| 825 |
+
"logits/chosen": -1.133528232574463,
|
| 826 |
+
"logits/rejected": -1.0763746500015259,
|
| 827 |
+
"logps/chosen": -0.9235790967941284,
|
| 828 |
+
"logps/rejected": -1.1159861087799072,
|
| 829 |
+
"loss": 2.7155,
|
| 830 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 831 |
+
"rewards/chosen": -9.23579216003418,
|
| 832 |
+
"rewards/margins": 1.9240700006484985,
|
| 833 |
+
"rewards/rejected": -11.15986156463623,
|
| 834 |
+
"step": 275
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 0.6120218579234973,
|
| 838 |
+
"grad_norm": 130.88653954832958,
|
| 839 |
+
"learning_rate": 3.9193566913562915e-07,
|
| 840 |
+
"logits/chosen": -1.0998766422271729,
|
| 841 |
+
"logits/rejected": -1.1054426431655884,
|
| 842 |
+
"logps/chosen": -0.9381259083747864,
|
| 843 |
+
"logps/rejected": -1.2925664186477661,
|
| 844 |
+
"loss": 2.6819,
|
| 845 |
+
"rewards/accuracies": 0.737500011920929,
|
| 846 |
+
"rewards/chosen": -9.38125991821289,
|
| 847 |
+
"rewards/margins": 3.5444045066833496,
|
| 848 |
+
"rewards/rejected": -12.925663948059082,
|
| 849 |
+
"step": 280
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.6229508196721312,
|
| 853 |
+
"grad_norm": 120.47229604941695,
|
| 854 |
+
"learning_rate": 3.7336133469909623e-07,
|
| 855 |
+
"logits/chosen": -1.2325929403305054,
|
| 856 |
+
"logits/rejected": -1.2060118913650513,
|
| 857 |
+
"logps/chosen": -0.9257510900497437,
|
| 858 |
+
"logps/rejected": -1.290015697479248,
|
| 859 |
+
"loss": 2.3392,
|
| 860 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 861 |
+
"rewards/chosen": -9.257511138916016,
|
| 862 |
+
"rewards/margins": 3.6426455974578857,
|
| 863 |
+
"rewards/rejected": -12.90015697479248,
|
| 864 |
+
"step": 285
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.6338797814207651,
|
| 868 |
+
"grad_norm": 160.84314275747502,
|
| 869 |
+
"learning_rate": 3.549719567506076e-07,
|
| 870 |
+
"logits/chosen": -1.1784595251083374,
|
| 871 |
+
"logits/rejected": -1.1460840702056885,
|
| 872 |
+
"logps/chosen": -0.995724081993103,
|
| 873 |
+
"logps/rejected": -1.3252066373825073,
|
| 874 |
+
"loss": 2.6635,
|
| 875 |
+
"rewards/accuracies": 0.793749988079071,
|
| 876 |
+
"rewards/chosen": -9.95724105834961,
|
| 877 |
+
"rewards/margins": 3.2948246002197266,
|
| 878 |
+
"rewards/rejected": -13.252065658569336,
|
| 879 |
+
"step": 290
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 0.644808743169399,
|
| 883 |
+
"grad_norm": 128.31998968470484,
|
| 884 |
+
"learning_rate": 3.3679439308082774e-07,
|
| 885 |
+
"logits/chosen": -1.162412405014038,
|
| 886 |
+
"logits/rejected": -1.165961503982544,
|
| 887 |
+
"logps/chosen": -0.995215117931366,
|
| 888 |
+
"logps/rejected": -1.3827722072601318,
|
| 889 |
+
"loss": 2.1144,
|
| 890 |
+
"rewards/accuracies": 0.856249988079071,
|
| 891 |
+
"rewards/chosen": -9.95215129852295,
|
| 892 |
+
"rewards/margins": 3.875570774078369,
|
| 893 |
+
"rewards/rejected": -13.827722549438477,
|
| 894 |
+
"step": 295
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 0.6557377049180327,
|
| 898 |
+
"grad_norm": 113.0437640023239,
|
| 899 |
+
"learning_rate": 3.1885519212446716e-07,
|
| 900 |
+
"logits/chosen": -1.198628306388855,
|
| 901 |
+
"logits/rejected": -1.1889218091964722,
|
| 902 |
+
"logps/chosen": -1.035154104232788,
|
| 903 |
+
"logps/rejected": -1.390194296836853,
|
| 904 |
+
"loss": 2.3734,
|
| 905 |
+
"rewards/accuracies": 0.800000011920929,
|
| 906 |
+
"rewards/chosen": -10.351542472839355,
|
| 907 |
+
"rewards/margins": 3.550400495529175,
|
| 908 |
+
"rewards/rejected": -13.901944160461426,
|
| 909 |
+
"step": 300
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 0.6666666666666666,
|
| 913 |
+
"grad_norm": 148.57393016872774,
|
| 914 |
+
"learning_rate": 3.0118055418614295e-07,
|
| 915 |
+
"logits/chosen": -1.1905871629714966,
|
| 916 |
+
"logits/rejected": -1.147964596748352,
|
| 917 |
+
"logps/chosen": -1.0231964588165283,
|
| 918 |
+
"logps/rejected": -1.4057810306549072,
|
| 919 |
+
"loss": 2.5431,
|
| 920 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 921 |
+
"rewards/chosen": -10.231963157653809,
|
| 922 |
+
"rewards/margins": 3.825847625732422,
|
| 923 |
+
"rewards/rejected": -14.05781078338623,
|
| 924 |
+
"step": 305
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 0.6775956284153005,
|
| 928 |
+
"grad_norm": 201.17993276494983,
|
| 929 |
+
"learning_rate": 2.83796293174686e-07,
|
| 930 |
+
"logits/chosen": -1.1385854482650757,
|
| 931 |
+
"logits/rejected": -1.1492459774017334,
|
| 932 |
+
"logps/chosen": -1.0367190837860107,
|
| 933 |
+
"logps/rejected": -1.4528653621673584,
|
| 934 |
+
"loss": 2.7689,
|
| 935 |
+
"rewards/accuracies": 0.800000011920929,
|
| 936 |
+
"rewards/chosen": -10.36719036102295,
|
| 937 |
+
"rewards/margins": 4.161464214324951,
|
| 938 |
+
"rewards/rejected": -14.528656005859375,
|
| 939 |
+
"step": 310
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 0.6885245901639344,
|
| 943 |
+
"grad_norm": 150.32062982255817,
|
| 944 |
+
"learning_rate": 2.6672779890178046e-07,
|
| 945 |
+
"logits/chosen": -1.1942903995513916,
|
| 946 |
+
"logits/rejected": -1.1974905729293823,
|
| 947 |
+
"logps/chosen": -1.101068139076233,
|
| 948 |
+
"logps/rejected": -1.3708066940307617,
|
| 949 |
+
"loss": 2.4457,
|
| 950 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 951 |
+
"rewards/chosen": -11.01068115234375,
|
| 952 |
+
"rewards/margins": 2.6973862648010254,
|
| 953 |
+
"rewards/rejected": -13.70806884765625,
|
| 954 |
+
"step": 315
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.6994535519125683,
|
| 958 |
+
"grad_norm": 115.42514466297712,
|
| 959 |
+
"learning_rate": 2.500000000000001e-07,
|
| 960 |
+
"logits/chosen": -1.2662487030029297,
|
| 961 |
+
"logits/rejected": -1.2223560810089111,
|
| 962 |
+
"logps/chosen": -1.1014665365219116,
|
| 963 |
+
"logps/rejected": -1.4792479276657104,
|
| 964 |
+
"loss": 2.4845,
|
| 965 |
+
"rewards/accuracies": 0.800000011920929,
|
| 966 |
+
"rewards/chosen": -11.014665603637695,
|
| 967 |
+
"rewards/margins": 3.77781343460083,
|
| 968 |
+
"rewards/rejected": -14.79248046875,
|
| 969 |
+
"step": 320
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.7103825136612022,
|
| 973 |
+
"grad_norm": 168.00405960921228,
|
| 974 |
+
"learning_rate": 2.3363732751439923e-07,
|
| 975 |
+
"logits/chosen": -1.225192666053772,
|
| 976 |
+
"logits/rejected": -1.2128854990005493,
|
| 977 |
+
"logps/chosen": -1.0635629892349243,
|
| 978 |
+
"logps/rejected": -1.3986676931381226,
|
| 979 |
+
"loss": 2.4994,
|
| 980 |
+
"rewards/accuracies": 0.800000011920929,
|
| 981 |
+
"rewards/chosen": -10.635629653930664,
|
| 982 |
+
"rewards/margins": 3.3510470390319824,
|
| 983 |
+
"rewards/rejected": -13.986676216125488,
|
| 984 |
+
"step": 325
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 0.7213114754098361,
|
| 988 |
+
"grad_norm": 103.05949871144796,
|
| 989 |
+
"learning_rate": 2.1766367922083283e-07,
|
| 990 |
+
"logits/chosen": -1.160074234008789,
|
| 991 |
+
"logits/rejected": -1.1407663822174072,
|
| 992 |
+
"logps/chosen": -1.0139520168304443,
|
| 993 |
+
"logps/rejected": -1.4850687980651855,
|
| 994 |
+
"loss": 2.3858,
|
| 995 |
+
"rewards/accuracies": 0.856249988079071,
|
| 996 |
+
"rewards/chosen": -10.139519691467285,
|
| 997 |
+
"rewards/margins": 4.7111687660217285,
|
| 998 |
+
"rewards/rejected": -14.850689888000488,
|
| 999 |
+
"step": 330
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 0.73224043715847,
|
| 1003 |
+
"grad_norm": 140.1249582615815,
|
| 1004 |
+
"learning_rate": 2.021023847231202e-07,
|
| 1005 |
+
"logits/chosen": -1.1456273794174194,
|
| 1006 |
+
"logits/rejected": -1.115674614906311,
|
| 1007 |
+
"logps/chosen": -1.1265430450439453,
|
| 1008 |
+
"logps/rejected": -1.4902852773666382,
|
| 1009 |
+
"loss": 2.3602,
|
| 1010 |
+
"rewards/accuracies": 0.856249988079071,
|
| 1011 |
+
"rewards/chosen": -11.265429496765137,
|
| 1012 |
+
"rewards/margins": 3.637423276901245,
|
| 1013 |
+
"rewards/rejected": -14.902852058410645,
|
| 1014 |
+
"step": 335
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 0.7431693989071039,
|
| 1018 |
+
"grad_norm": 144.16110329539526,
|
| 1019 |
+
"learning_rate": 1.869761713800254e-07,
|
| 1020 |
+
"logits/chosen": -1.1668914556503296,
|
| 1021 |
+
"logits/rejected": -1.1292145252227783,
|
| 1022 |
+
"logps/chosen": -1.0967395305633545,
|
| 1023 |
+
"logps/rejected": -1.4661519527435303,
|
| 1024 |
+
"loss": 2.5221,
|
| 1025 |
+
"rewards/accuracies": 0.768750011920929,
|
| 1026 |
+
"rewards/chosen": -10.967395782470703,
|
| 1027 |
+
"rewards/margins": 3.6941237449645996,
|
| 1028 |
+
"rewards/rejected": -14.661520004272461,
|
| 1029 |
+
"step": 340
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 0.7540983606557377,
|
| 1033 |
+
"grad_norm": 149.08880264886358,
|
| 1034 |
+
"learning_rate": 1.7230713111182164e-07,
|
| 1035 |
+
"logits/chosen": -1.2098548412322998,
|
| 1036 |
+
"logits/rejected": -1.2148889303207397,
|
| 1037 |
+
"logps/chosen": -1.1389497518539429,
|
| 1038 |
+
"logps/rejected": -1.5369229316711426,
|
| 1039 |
+
"loss": 2.5085,
|
| 1040 |
+
"rewards/accuracies": 0.8125,
|
| 1041 |
+
"rewards/chosen": -11.389498710632324,
|
| 1042 |
+
"rewards/margins": 3.979731798171997,
|
| 1043 |
+
"rewards/rejected": -15.369229316711426,
|
| 1044 |
+
"step": 345
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 0.7650273224043715,
|
| 1048 |
+
"grad_norm": 144.89905388084784,
|
| 1049 |
+
"learning_rate": 1.5811668813491696e-07,
|
| 1050 |
+
"logits/chosen": -1.2163605690002441,
|
| 1051 |
+
"logits/rejected": -1.2011533975601196,
|
| 1052 |
+
"logps/chosen": -1.0724060535430908,
|
| 1053 |
+
"logps/rejected": -1.4004809856414795,
|
| 1054 |
+
"loss": 2.4309,
|
| 1055 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1056 |
+
"rewards/chosen": -10.724058151245117,
|
| 1057 |
+
"rewards/margins": 3.280749559402466,
|
| 1058 |
+
"rewards/rejected": -14.004809379577637,
|
| 1059 |
+
"step": 350
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.7759562841530054,
|
| 1063 |
+
"grad_norm": 146.85736518453848,
|
| 1064 |
+
"learning_rate": 1.4442556767166369e-07,
|
| 1065 |
+
"logits/chosen": -1.171179175376892,
|
| 1066 |
+
"logits/rejected": -1.144737958908081,
|
| 1067 |
+
"logps/chosen": -1.0694037675857544,
|
| 1068 |
+
"logps/rejected": -1.4522759914398193,
|
| 1069 |
+
"loss": 2.4166,
|
| 1070 |
+
"rewards/accuracies": 0.84375,
|
| 1071 |
+
"rewards/chosen": -10.694037437438965,
|
| 1072 |
+
"rewards/margins": 3.828723192214966,
|
| 1073 |
+
"rewards/rejected": -14.522760391235352,
|
| 1074 |
+
"step": 355
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.7868852459016393,
|
| 1078 |
+
"grad_norm": 145.85471061365715,
|
| 1079 |
+
"learning_rate": 1.312537656810549e-07,
|
| 1080 |
+
"logits/chosen": -1.1415390968322754,
|
| 1081 |
+
"logits/rejected": -1.1467649936676025,
|
| 1082 |
+
"logps/chosen": -1.1098216772079468,
|
| 1083 |
+
"logps/rejected": -1.4691263437271118,
|
| 1084 |
+
"loss": 2.4771,
|
| 1085 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1086 |
+
"rewards/chosen": -11.09821605682373,
|
| 1087 |
+
"rewards/margins": 3.5930473804473877,
|
| 1088 |
+
"rewards/rejected": -14.691264152526855,
|
| 1089 |
+
"step": 360
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 0.7978142076502732,
|
| 1093 |
+
"grad_norm": 167.3238922002567,
|
| 1094 |
+
"learning_rate": 1.1862051965451214e-07,
|
| 1095 |
+
"logits/chosen": -1.2173941135406494,
|
| 1096 |
+
"logits/rejected": -1.2231066226959229,
|
| 1097 |
+
"logps/chosen": -1.143596887588501,
|
| 1098 |
+
"logps/rejected": -1.5389248132705688,
|
| 1099 |
+
"loss": 2.3622,
|
| 1100 |
+
"rewards/accuracies": 0.78125,
|
| 1101 |
+
"rewards/chosen": -11.435969352722168,
|
| 1102 |
+
"rewards/margins": 3.9532806873321533,
|
| 1103 |
+
"rewards/rejected": -15.389248847961426,
|
| 1104 |
+
"step": 365
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 0.8087431693989071,
|
| 1108 |
+
"grad_norm": 153.79026677729559,
|
| 1109 |
+
"learning_rate": 1.0654428051942138e-07,
|
| 1110 |
+
"logits/chosen": -1.225081443786621,
|
| 1111 |
+
"logits/rejected": -1.1957951784133911,
|
| 1112 |
+
"logps/chosen": -1.163589596748352,
|
| 1113 |
+
"logps/rejected": -1.5947258472442627,
|
| 1114 |
+
"loss": 2.5258,
|
| 1115 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1116 |
+
"rewards/chosen": -11.635896682739258,
|
| 1117 |
+
"rewards/margins": 4.311361312866211,
|
| 1118 |
+
"rewards/rejected": -15.947257995605469,
|
| 1119 |
+
"step": 370
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 0.819672131147541,
|
| 1123 |
+
"grad_norm": 112.49710778946844,
|
| 1124 |
+
"learning_rate": 9.504268569144763e-08,
|
| 1125 |
+
"logits/chosen": -1.2405080795288086,
|
| 1126 |
+
"logits/rejected": -1.1872971057891846,
|
| 1127 |
+
"logps/chosen": -1.1241363286972046,
|
| 1128 |
+
"logps/rejected": -1.525253415107727,
|
| 1129 |
+
"loss": 2.385,
|
| 1130 |
+
"rewards/accuracies": 0.8125,
|
| 1131 |
+
"rewards/chosen": -11.241364479064941,
|
| 1132 |
+
"rewards/margins": 4.011169910430908,
|
| 1133 |
+
"rewards/rejected": -15.252534866333008,
|
| 1134 |
+
"step": 375
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 0.8306010928961749,
|
| 1138 |
+
"grad_norm": 133.3144411726237,
|
| 1139 |
+
"learning_rate": 8.413253331499049e-08,
|
| 1140 |
+
"logits/chosen": -1.1226742267608643,
|
| 1141 |
+
"logits/rejected": -1.1412006616592407,
|
| 1142 |
+
"logps/chosen": -1.1478770971298218,
|
| 1143 |
+
"logps/rejected": -1.5184364318847656,
|
| 1144 |
+
"loss": 2.3692,
|
| 1145 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1146 |
+
"rewards/chosen": -11.478772163391113,
|
| 1147 |
+
"rewards/margins": 3.705592393875122,
|
| 1148 |
+
"rewards/rejected": -15.184364318847656,
|
| 1149 |
+
"step": 380
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 0.8415300546448088,
|
| 1153 |
+
"grad_norm": 131.51573012388099,
|
| 1154 |
+
"learning_rate": 7.382975772939865e-08,
|
| 1155 |
+
"logits/chosen": -1.233059287071228,
|
| 1156 |
+
"logits/rejected": -1.2242696285247803,
|
| 1157 |
+
"logps/chosen": -1.211987853050232,
|
| 1158 |
+
"logps/rejected": -1.6103414297103882,
|
| 1159 |
+
"loss": 2.6625,
|
| 1160 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1161 |
+
"rewards/chosen": -12.119878768920898,
|
| 1162 |
+
"rewards/margins": 3.983535051345825,
|
| 1163 |
+
"rewards/rejected": -16.103412628173828,
|
| 1164 |
+
"step": 385
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.8524590163934426,
|
| 1168 |
+
"grad_norm": 179.13558974996818,
|
| 1169 |
+
"learning_rate": 6.414940619677734e-08,
|
| 1170 |
+
"logits/chosen": -1.2085000276565552,
|
| 1171 |
+
"logits/rejected": -1.1961381435394287,
|
| 1172 |
+
"logps/chosen": -1.1327707767486572,
|
| 1173 |
+
"logps/rejected": -1.6155046224594116,
|
| 1174 |
+
"loss": 2.3237,
|
| 1175 |
+
"rewards/accuracies": 0.831250011920929,
|
| 1176 |
+
"rewards/chosen": -11.327706336975098,
|
| 1177 |
+
"rewards/margins": 4.827339172363281,
|
| 1178 |
+
"rewards/rejected": -16.155044555664062,
|
| 1179 |
+
"step": 390
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.8633879781420765,
|
| 1183 |
+
"grad_norm": 187.32213579680786,
|
| 1184 |
+
"learning_rate": 5.5105616925376296e-08,
|
| 1185 |
+
"logits/chosen": -1.1982018947601318,
|
| 1186 |
+
"logits/rejected": -1.1823650598526,
|
| 1187 |
+
"logps/chosen": -1.1895129680633545,
|
| 1188 |
+
"logps/rejected": -1.5091607570648193,
|
| 1189 |
+
"loss": 2.3095,
|
| 1190 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1191 |
+
"rewards/chosen": -11.895132064819336,
|
| 1192 |
+
"rewards/margins": 3.196476936340332,
|
| 1193 |
+
"rewards/rejected": -15.091608047485352,
|
| 1194 |
+
"step": 395
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 0.8743169398907104,
|
| 1198 |
+
"grad_norm": 152.08655815719752,
|
| 1199 |
+
"learning_rate": 4.6711598420656976e-08,
|
| 1200 |
+
"logits/chosen": -1.1382460594177246,
|
| 1201 |
+
"logits/rejected": -1.1149357557296753,
|
| 1202 |
+
"logps/chosen": -1.1855409145355225,
|
| 1203 |
+
"logps/rejected": -1.6318585872650146,
|
| 1204 |
+
"loss": 2.1949,
|
| 1205 |
+
"rewards/accuracies": 0.875,
|
| 1206 |
+
"rewards/chosen": -11.855409622192383,
|
| 1207 |
+
"rewards/margins": 4.463176727294922,
|
| 1208 |
+
"rewards/rejected": -16.318586349487305,
|
| 1209 |
+
"step": 400
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 0.8743169398907104,
|
| 1213 |
+
"eval_logits/chosen": -1.3829468488693237,
|
| 1214 |
+
"eval_logits/rejected": -1.3507002592086792,
|
| 1215 |
+
"eval_logps/chosen": -1.1304905414581299,
|
| 1216 |
+
"eval_logps/rejected": -1.5389478206634521,
|
| 1217 |
+
"eval_loss": 2.412938117980957,
|
| 1218 |
+
"eval_rewards/accuracies": 0.8132529854774475,
|
| 1219 |
+
"eval_rewards/chosen": -11.30490493774414,
|
| 1220 |
+
"eval_rewards/margins": 4.0845723152160645,
|
| 1221 |
+
"eval_rewards/rejected": -15.389474868774414,
|
| 1222 |
+
"eval_runtime": 37.7675,
|
| 1223 |
+
"eval_samples_per_second": 34.898,
|
| 1224 |
+
"eval_steps_per_second": 2.198,
|
| 1225 |
+
"step": 400
|
| 1226 |
+
},
|
| 1227 |
+
{
|
| 1228 |
+
"epoch": 0.8852459016393442,
|
| 1229 |
+
"grad_norm": 137.4747366766835,
|
| 1230 |
+
"learning_rate": 3.897961019419516e-08,
|
| 1231 |
+
"logits/chosen": -1.1692960262298584,
|
| 1232 |
+
"logits/rejected": -1.1131761074066162,
|
| 1233 |
+
"logps/chosen": -1.078470230102539,
|
| 1234 |
+
"logps/rejected": -1.4457483291625977,
|
| 1235 |
+
"loss": 2.3038,
|
| 1236 |
+
"rewards/accuracies": 0.793749988079071,
|
| 1237 |
+
"rewards/chosen": -10.78470230102539,
|
| 1238 |
+
"rewards/margins": 3.672783613204956,
|
| 1239 |
+
"rewards/rejected": -14.457486152648926,
|
| 1240 |
+
"step": 405
|
| 1241 |
+
},
|
| 1242 |
+
{
|
| 1243 |
+
"epoch": 0.8961748633879781,
|
| 1244 |
+
"grad_norm": 128.05923930291988,
|
| 1245 |
+
"learning_rate": 3.192094485859526e-08,
|
| 1246 |
+
"logits/chosen": -1.171673059463501,
|
| 1247 |
+
"logits/rejected": -1.2017241716384888,
|
| 1248 |
+
"logps/chosen": -1.1775602102279663,
|
| 1249 |
+
"logps/rejected": -1.6198632717132568,
|
| 1250 |
+
"loss": 2.3744,
|
| 1251 |
+
"rewards/accuracies": 0.78125,
|
| 1252 |
+
"rewards/chosen": -11.775602340698242,
|
| 1253 |
+
"rewards/margins": 4.423029899597168,
|
| 1254 |
+
"rewards/rejected": -16.198633193969727,
|
| 1255 |
+
"step": 410
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 0.907103825136612,
|
| 1259 |
+
"grad_norm": 136.25040575701973,
|
| 1260 |
+
"learning_rate": 2.5545911634565265e-08,
|
| 1261 |
+
"logits/chosen": -1.2182672023773193,
|
| 1262 |
+
"logits/rejected": -1.222606897354126,
|
| 1263 |
+
"logps/chosen": -1.1761579513549805,
|
| 1264 |
+
"logps/rejected": -1.6315196752548218,
|
| 1265 |
+
"loss": 2.556,
|
| 1266 |
+
"rewards/accuracies": 0.856249988079071,
|
| 1267 |
+
"rewards/chosen": -11.761578559875488,
|
| 1268 |
+
"rewards/margins": 4.553616523742676,
|
| 1269 |
+
"rewards/rejected": -16.315195083618164,
|
| 1270 |
+
"step": 415
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 0.9180327868852459,
|
| 1274 |
+
"grad_norm": 135.49217722627617,
|
| 1275 |
+
"learning_rate": 1.9863821294241522e-08,
|
| 1276 |
+
"logits/chosen": -1.2129249572753906,
|
| 1277 |
+
"logits/rejected": -1.1957124471664429,
|
| 1278 |
+
"logps/chosen": -1.1399694681167603,
|
| 1279 |
+
"logps/rejected": -1.5886274576187134,
|
| 1280 |
+
"loss": 2.2082,
|
| 1281 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 1282 |
+
"rewards/chosen": -11.399694442749023,
|
| 1283 |
+
"rewards/margins": 4.486580848693848,
|
| 1284 |
+
"rewards/rejected": -15.886274337768555,
|
| 1285 |
+
"step": 420
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 0.9289617486338798,
|
| 1289 |
+
"grad_norm": 135.8877044935921,
|
| 1290 |
+
"learning_rate": 1.4882972562753615e-08,
|
| 1291 |
+
"logits/chosen": -1.2110410928726196,
|
| 1292 |
+
"logits/rejected": -1.2060327529907227,
|
| 1293 |
+
"logps/chosen": -1.2429840564727783,
|
| 1294 |
+
"logps/rejected": -1.6804256439208984,
|
| 1295 |
+
"loss": 2.4895,
|
| 1296 |
+
"rewards/accuracies": 0.8125,
|
| 1297 |
+
"rewards/chosen": -12.429841041564941,
|
| 1298 |
+
"rewards/margins": 4.374415397644043,
|
| 1299 |
+
"rewards/rejected": -16.804256439208984,
|
| 1300 |
+
"step": 425
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"epoch": 0.9398907103825137,
|
| 1304 |
+
"grad_norm": 132.90623812452836,
|
| 1305 |
+
"learning_rate": 1.0610639997888915e-08,
|
| 1306 |
+
"logits/chosen": -1.1448732614517212,
|
| 1307 |
+
"logits/rejected": -1.1479747295379639,
|
| 1308 |
+
"logps/chosen": -1.1143007278442383,
|
| 1309 |
+
"logps/rejected": -1.5705087184906006,
|
| 1310 |
+
"loss": 2.0744,
|
| 1311 |
+
"rewards/accuracies": 0.893750011920929,
|
| 1312 |
+
"rewards/chosen": -11.143007278442383,
|
| 1313 |
+
"rewards/margins": 4.562080383300781,
|
| 1314 |
+
"rewards/rejected": -15.705087661743164,
|
| 1315 |
+
"step": 430
|
| 1316 |
+
},
|
| 1317 |
+
{
|
| 1318 |
+
"epoch": 0.9508196721311475,
|
| 1319 |
+
"grad_norm": 150.69294443867608,
|
| 1320 |
+
"learning_rate": 7.053063365559997e-09,
|
| 1321 |
+
"logits/chosen": -1.2052174806594849,
|
| 1322 |
+
"logits/rejected": -1.2350053787231445,
|
| 1323 |
+
"logps/chosen": -1.1626297235488892,
|
| 1324 |
+
"logps/rejected": -1.625112771987915,
|
| 1325 |
+
"loss": 2.2249,
|
| 1326 |
+
"rewards/accuracies": 0.875,
|
| 1327 |
+
"rewards/chosen": -11.626296997070312,
|
| 1328 |
+
"rewards/margins": 4.6248297691345215,
|
| 1329 |
+
"rewards/rejected": -16.251127243041992,
|
| 1330 |
+
"step": 435
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"epoch": 0.9617486338797814,
|
| 1334 |
+
"grad_norm": 126.29899935620595,
|
| 1335 |
+
"learning_rate": 4.215438526591064e-09,
|
| 1336 |
+
"logits/chosen": -1.1682944297790527,
|
| 1337 |
+
"logits/rejected": -1.1369967460632324,
|
| 1338 |
+
"logps/chosen": -1.205596685409546,
|
| 1339 |
+
"logps/rejected": -1.5388822555541992,
|
| 1340 |
+
"loss": 2.426,
|
| 1341 |
+
"rewards/accuracies": 0.768750011920929,
|
| 1342 |
+
"rewards/chosen": -12.055967330932617,
|
| 1343 |
+
"rewards/margins": 3.332855701446533,
|
| 1344 |
+
"rewards/rejected": -15.388822555541992,
|
| 1345 |
+
"step": 440
|
| 1346 |
+
},
|
| 1347 |
+
{
|
| 1348 |
+
"epoch": 0.9726775956284153,
|
| 1349 |
+
"grad_norm": 145.76337247257322,
|
| 1350 |
+
"learning_rate": 2.1019098481337426e-09,
|
| 1351 |
+
"logits/chosen": -1.2075254917144775,
|
| 1352 |
+
"logits/rejected": -1.1947346925735474,
|
| 1353 |
+
"logps/chosen": -1.124491810798645,
|
| 1354 |
+
"logps/rejected": -1.569777250289917,
|
| 1355 |
+
"loss": 2.2395,
|
| 1356 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 1357 |
+
"rewards/chosen": -11.244916915893555,
|
| 1358 |
+
"rewards/margins": 4.452854156494141,
|
| 1359 |
+
"rewards/rejected": -15.697772026062012,
|
| 1360 |
+
"step": 445
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 0.9836065573770492,
|
| 1364 |
+
"grad_norm": 163.83031864944653,
|
| 1365 |
+
"learning_rate": 7.155641507955445e-10,
|
| 1366 |
+
"logits/chosen": -1.1285139322280884,
|
| 1367 |
+
"logits/rejected": -1.1291377544403076,
|
| 1368 |
+
"logps/chosen": -1.2102129459381104,
|
| 1369 |
+
"logps/rejected": -1.590519666671753,
|
| 1370 |
+
"loss": 2.5308,
|
| 1371 |
+
"rewards/accuracies": 0.824999988079071,
|
| 1372 |
+
"rewards/chosen": -12.102127075195312,
|
| 1373 |
+
"rewards/margins": 3.803069591522217,
|
| 1374 |
+
"rewards/rejected": -15.90519905090332,
|
| 1375 |
+
"step": 450
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 0.994535519125683,
|
| 1379 |
+
"grad_norm": 155.57860273708854,
|
| 1380 |
+
"learning_rate": 5.842620032053824e-11,
|
| 1381 |
+
"logits/chosen": -1.1562345027923584,
|
| 1382 |
+
"logits/rejected": -1.1542116403579712,
|
| 1383 |
+
"logps/chosen": -1.228244662284851,
|
| 1384 |
+
"logps/rejected": -1.5371659994125366,
|
| 1385 |
+
"loss": 2.7124,
|
| 1386 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1387 |
+
"rewards/chosen": -12.282448768615723,
|
| 1388 |
+
"rewards/margins": 3.0892109870910645,
|
| 1389 |
+
"rewards/rejected": -15.371660232543945,
|
| 1390 |
+
"step": 455
|
| 1391 |
+
},
|
| 1392 |
+
{
|
| 1393 |
+
"epoch": 0.9989071038251366,
|
| 1394 |
+
"step": 457,
|
| 1395 |
+
"total_flos": 0.0,
|
| 1396 |
+
"train_loss": 3.252214796955267,
|
| 1397 |
+
"train_runtime": 5969.5229,
|
| 1398 |
+
"train_samples_per_second": 9.809,
|
| 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 |
+
}
|