Commit
•
adf1bdb
1
Parent(s):
1a63694
Model save
Browse files- README.md +74 -0
- all_results.json +9 -0
- generation_config.json +6 -0
- train_results.json +9 -0
- trainer_state.json +2233 -0
README.md
ADDED
|
@@ -0,0 +1,74 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
license: apache-2.0
|
| 4 |
+
base_model: alignment-handbook/zephyr-7b-sft-full
|
| 5 |
+
tags:
|
| 6 |
+
- trl
|
| 7 |
+
- dpo
|
| 8 |
+
- generated_from_trainer
|
| 9 |
+
model-index:
|
| 10 |
+
- name: zephyr-7b-uf-rlced-conifer-dpo-2e
|
| 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 |
+
# zephyr-7b-uf-rlced-conifer-dpo-2e
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on an unknown dataset.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 0.2471
|
| 22 |
+
- Rewards/chosen: -3.9472
|
| 23 |
+
- Rewards/rejected: -12.0338
|
| 24 |
+
- Rewards/accuracies: 0.8910
|
| 25 |
+
- Rewards/margins: 8.0866
|
| 26 |
+
- Logps/rejected: -1613.5016
|
| 27 |
+
- Logps/chosen: -779.3055
|
| 28 |
+
- Logits/rejected: 4.5606
|
| 29 |
+
- Logits/chosen: 2.4398
|
| 30 |
+
|
| 31 |
+
## Model description
|
| 32 |
+
|
| 33 |
+
More information needed
|
| 34 |
+
|
| 35 |
+
## Intended uses & limitations
|
| 36 |
+
|
| 37 |
+
More information needed
|
| 38 |
+
|
| 39 |
+
## Training and evaluation data
|
| 40 |
+
|
| 41 |
+
More information needed
|
| 42 |
+
|
| 43 |
+
## Training procedure
|
| 44 |
+
|
| 45 |
+
### Training hyperparameters
|
| 46 |
+
|
| 47 |
+
The following hyperparameters were used during training:
|
| 48 |
+
- learning_rate: 5e-07
|
| 49 |
+
- train_batch_size: 8
|
| 50 |
+
- eval_batch_size: 8
|
| 51 |
+
- seed: 42
|
| 52 |
+
- distributed_type: multi-GPU
|
| 53 |
+
- num_devices: 8
|
| 54 |
+
- gradient_accumulation_steps: 4
|
| 55 |
+
- total_train_batch_size: 256
|
| 56 |
+
- total_eval_batch_size: 64
|
| 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: 2
|
| 61 |
+
|
| 62 |
+
### Training results
|
| 63 |
+
|
| 64 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 65 |
+
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 66 |
+
| 0.1587 | 1.3879 | 1000 | 0.2471 | -3.9472 | -12.0338 | 0.8910 | 8.0866 | -1613.5016 | -779.3055 | 4.5606 | 2.4398 |
|
| 67 |
+
|
| 68 |
+
|
| 69 |
+
### Framework versions
|
| 70 |
+
|
| 71 |
+
- Transformers 4.44.1
|
| 72 |
+
- Pytorch 2.1.2+cu121
|
| 73 |
+
- Datasets 2.21.0
|
| 74 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.9986120749479528,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.25082930790053476,
|
| 5 |
+
"train_runtime": 42170.4729,
|
| 6 |
+
"train_samples": 184443,
|
| 7 |
+
"train_samples_per_second": 8.747,
|
| 8 |
+
"train_steps_per_second": 0.034
|
| 9 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"eos_token_id": 2,
|
| 5 |
+
"transformers_version": "4.44.1"
|
| 6 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.9986120749479528,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.25082930790053476,
|
| 5 |
+
"train_runtime": 42170.4729,
|
| 6 |
+
"train_samples": 184443,
|
| 7 |
+
"train_samples_per_second": 8.747,
|
| 8 |
+
"train_steps_per_second": 0.034
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,2233 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.9986120749479528,
|
| 5 |
+
"eval_steps": 1000,
|
| 6 |
+
"global_step": 1440,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0013879250520471894,
|
| 13 |
+
"grad_norm": 7.310771886732375,
|
| 14 |
+
"learning_rate": 3.4722222222222217e-09,
|
| 15 |
+
"logits/chosen": -2.658149242401123,
|
| 16 |
+
"logits/rejected": -2.6729652881622314,
|
| 17 |
+
"logps/chosen": -310.6693115234375,
|
| 18 |
+
"logps/rejected": -336.3360595703125,
|
| 19 |
+
"loss": 0.6931,
|
| 20 |
+
"rewards/accuracies": 0.0,
|
| 21 |
+
"rewards/chosen": 0.0,
|
| 22 |
+
"rewards/margins": 0.0,
|
| 23 |
+
"rewards/rejected": 0.0,
|
| 24 |
+
"step": 1
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.013879250520471894,
|
| 28 |
+
"grad_norm": 6.989456718606319,
|
| 29 |
+
"learning_rate": 3.472222222222222e-08,
|
| 30 |
+
"logits/chosen": -2.7207493782043457,
|
| 31 |
+
"logits/rejected": -2.678452968597412,
|
| 32 |
+
"logps/chosen": -329.1288146972656,
|
| 33 |
+
"logps/rejected": -334.9566650390625,
|
| 34 |
+
"loss": 0.6932,
|
| 35 |
+
"rewards/accuracies": 0.4236111044883728,
|
| 36 |
+
"rewards/chosen": -0.00011655675189103931,
|
| 37 |
+
"rewards/margins": -0.00021387077867984772,
|
| 38 |
+
"rewards/rejected": 9.731399040902033e-05,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.027758501040943788,
|
| 43 |
+
"grad_norm": 7.28858517103727,
|
| 44 |
+
"learning_rate": 6.944444444444444e-08,
|
| 45 |
+
"logits/chosen": -2.690325975418091,
|
| 46 |
+
"logits/rejected": -2.6607680320739746,
|
| 47 |
+
"logps/chosen": -317.3731384277344,
|
| 48 |
+
"logps/rejected": -331.0643005371094,
|
| 49 |
+
"loss": 0.6928,
|
| 50 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 51 |
+
"rewards/chosen": 0.00018671144789550453,
|
| 52 |
+
"rewards/margins": 0.00033597589936107397,
|
| 53 |
+
"rewards/rejected": -0.00014926446601748466,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.041637751561415685,
|
| 58 |
+
"grad_norm": 7.1293734693123785,
|
| 59 |
+
"learning_rate": 1.0416666666666667e-07,
|
| 60 |
+
"logits/chosen": -2.6790878772735596,
|
| 61 |
+
"logits/rejected": -2.645470142364502,
|
| 62 |
+
"logps/chosen": -351.1817321777344,
|
| 63 |
+
"logps/rejected": -351.5977478027344,
|
| 64 |
+
"loss": 0.6914,
|
| 65 |
+
"rewards/accuracies": 0.690625011920929,
|
| 66 |
+
"rewards/chosen": 0.0023157999385148287,
|
| 67 |
+
"rewards/margins": 0.004029616713523865,
|
| 68 |
+
"rewards/rejected": -0.0017138172406703234,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.055517002081887576,
|
| 73 |
+
"grad_norm": 7.102945109434136,
|
| 74 |
+
"learning_rate": 1.3888888888888888e-07,
|
| 75 |
+
"logits/chosen": -2.698319435119629,
|
| 76 |
+
"logits/rejected": -2.6377086639404297,
|
| 77 |
+
"logps/chosen": -352.13153076171875,
|
| 78 |
+
"logps/rejected": -337.7492370605469,
|
| 79 |
+
"loss": 0.6867,
|
| 80 |
+
"rewards/accuracies": 0.762499988079071,
|
| 81 |
+
"rewards/chosen": 0.010404938831925392,
|
| 82 |
+
"rewards/margins": 0.012554061599075794,
|
| 83 |
+
"rewards/rejected": -0.0021491218358278275,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.06939625260235947,
|
| 88 |
+
"grad_norm": 6.472817798536912,
|
| 89 |
+
"learning_rate": 1.736111111111111e-07,
|
| 90 |
+
"logits/chosen": -2.6700968742370605,
|
| 91 |
+
"logits/rejected": -2.656541347503662,
|
| 92 |
+
"logps/chosen": -325.9363708496094,
|
| 93 |
+
"logps/rejected": -359.7778625488281,
|
| 94 |
+
"loss": 0.6778,
|
| 95 |
+
"rewards/accuracies": 0.7906249761581421,
|
| 96 |
+
"rewards/chosen": 0.026792842894792557,
|
| 97 |
+
"rewards/margins": 0.031667567789554596,
|
| 98 |
+
"rewards/rejected": -0.004874727688729763,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.08327550312283137,
|
| 103 |
+
"grad_norm": 7.812961434750235,
|
| 104 |
+
"learning_rate": 2.0833333333333333e-07,
|
| 105 |
+
"logits/chosen": -2.7250404357910156,
|
| 106 |
+
"logits/rejected": -2.680410861968994,
|
| 107 |
+
"logps/chosen": -325.12628173828125,
|
| 108 |
+
"logps/rejected": -345.4599914550781,
|
| 109 |
+
"loss": 0.6575,
|
| 110 |
+
"rewards/accuracies": 0.815625011920929,
|
| 111 |
+
"rewards/chosen": 0.05238135904073715,
|
| 112 |
+
"rewards/margins": 0.06940947473049164,
|
| 113 |
+
"rewards/rejected": -0.017028113827109337,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.09715475364330327,
|
| 118 |
+
"grad_norm": 8.252964521813333,
|
| 119 |
+
"learning_rate": 2.4305555555555555e-07,
|
| 120 |
+
"logits/chosen": -2.6680197715759277,
|
| 121 |
+
"logits/rejected": -2.6488897800445557,
|
| 122 |
+
"logps/chosen": -336.755615234375,
|
| 123 |
+
"logps/rejected": -372.3600158691406,
|
| 124 |
+
"loss": 0.6231,
|
| 125 |
+
"rewards/accuracies": 0.809374988079071,
|
| 126 |
+
"rewards/chosen": 0.04352443665266037,
|
| 127 |
+
"rewards/margins": 0.14054368436336517,
|
| 128 |
+
"rewards/rejected": -0.0970192551612854,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.11103400416377515,
|
| 133 |
+
"grad_norm": 10.907322040862939,
|
| 134 |
+
"learning_rate": 2.7777777777777776e-07,
|
| 135 |
+
"logits/chosen": -2.6945643424987793,
|
| 136 |
+
"logits/rejected": -2.635817050933838,
|
| 137 |
+
"logps/chosen": -345.4059143066406,
|
| 138 |
+
"logps/rejected": -395.4034423828125,
|
| 139 |
+
"loss": 0.5649,
|
| 140 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 141 |
+
"rewards/chosen": -0.05337335914373398,
|
| 142 |
+
"rewards/margins": 0.348626971244812,
|
| 143 |
+
"rewards/rejected": -0.402000367641449,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.12491325468424705,
|
| 148 |
+
"grad_norm": 14.954457613666033,
|
| 149 |
+
"learning_rate": 3.1249999999999997e-07,
|
| 150 |
+
"logits/chosen": -2.7117972373962402,
|
| 151 |
+
"logits/rejected": -2.659916877746582,
|
| 152 |
+
"logps/chosen": -357.34344482421875,
|
| 153 |
+
"logps/rejected": -424.046875,
|
| 154 |
+
"loss": 0.4833,
|
| 155 |
+
"rewards/accuracies": 0.815625011920929,
|
| 156 |
+
"rewards/chosen": -0.24065232276916504,
|
| 157 |
+
"rewards/margins": 0.6489619016647339,
|
| 158 |
+
"rewards/rejected": -0.8896142840385437,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.13879250520471895,
|
| 163 |
+
"grad_norm": 18.8809663980661,
|
| 164 |
+
"learning_rate": 3.472222222222222e-07,
|
| 165 |
+
"logits/chosen": -2.691723108291626,
|
| 166 |
+
"logits/rejected": -2.676143169403076,
|
| 167 |
+
"logps/chosen": -429.1051330566406,
|
| 168 |
+
"logps/rejected": -508.2496032714844,
|
| 169 |
+
"loss": 0.4648,
|
| 170 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 171 |
+
"rewards/chosen": -0.836562991142273,
|
| 172 |
+
"rewards/margins": 0.7794925570487976,
|
| 173 |
+
"rewards/rejected": -1.6160557270050049,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.15267175572519084,
|
| 178 |
+
"grad_norm": 17.153622207375392,
|
| 179 |
+
"learning_rate": 3.819444444444444e-07,
|
| 180 |
+
"logits/chosen": -2.7029433250427246,
|
| 181 |
+
"logits/rejected": -2.6785738468170166,
|
| 182 |
+
"logps/chosen": -463.09588623046875,
|
| 183 |
+
"logps/rejected": -551.4292602539062,
|
| 184 |
+
"loss": 0.4393,
|
| 185 |
+
"rewards/accuracies": 0.78125,
|
| 186 |
+
"rewards/chosen": -1.1502994298934937,
|
| 187 |
+
"rewards/margins": 0.9367601275444031,
|
| 188 |
+
"rewards/rejected": -2.087059736251831,
|
| 189 |
+
"step": 110
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.16655100624566274,
|
| 193 |
+
"grad_norm": 17.322428984395128,
|
| 194 |
+
"learning_rate": 4.1666666666666667e-07,
|
| 195 |
+
"logits/chosen": -2.740248203277588,
|
| 196 |
+
"logits/rejected": -2.715928554534912,
|
| 197 |
+
"logps/chosen": -466.61614990234375,
|
| 198 |
+
"logps/rejected": -588.0739135742188,
|
| 199 |
+
"loss": 0.416,
|
| 200 |
+
"rewards/accuracies": 0.746874988079071,
|
| 201 |
+
"rewards/chosen": -1.314789891242981,
|
| 202 |
+
"rewards/margins": 1.0913175344467163,
|
| 203 |
+
"rewards/rejected": -2.4061074256896973,
|
| 204 |
+
"step": 120
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.18043025676613464,
|
| 208 |
+
"grad_norm": 17.727320951964273,
|
| 209 |
+
"learning_rate": 4.513888888888889e-07,
|
| 210 |
+
"logits/chosen": -2.732884168624878,
|
| 211 |
+
"logits/rejected": -2.693039894104004,
|
| 212 |
+
"logps/chosen": -470.36712646484375,
|
| 213 |
+
"logps/rejected": -637.1500854492188,
|
| 214 |
+
"loss": 0.3922,
|
| 215 |
+
"rewards/accuracies": 0.831250011920929,
|
| 216 |
+
"rewards/chosen": -1.2835638523101807,
|
| 217 |
+
"rewards/margins": 1.5643529891967773,
|
| 218 |
+
"rewards/rejected": -2.847916841506958,
|
| 219 |
+
"step": 130
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.19430950728660654,
|
| 223 |
+
"grad_norm": 23.308808504878304,
|
| 224 |
+
"learning_rate": 4.861111111111111e-07,
|
| 225 |
+
"logits/chosen": -2.410512924194336,
|
| 226 |
+
"logits/rejected": -2.2582736015319824,
|
| 227 |
+
"logps/chosen": -466.9541931152344,
|
| 228 |
+
"logps/rejected": -714.3003540039062,
|
| 229 |
+
"loss": 0.355,
|
| 230 |
+
"rewards/accuracies": 0.846875011920929,
|
| 231 |
+
"rewards/chosen": -1.4372889995574951,
|
| 232 |
+
"rewards/margins": 2.2647476196289062,
|
| 233 |
+
"rewards/rejected": -3.7020363807678223,
|
| 234 |
+
"step": 140
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.2081887578070784,
|
| 238 |
+
"grad_norm": 15.792878048667099,
|
| 239 |
+
"learning_rate": 4.999735579817769e-07,
|
| 240 |
+
"logits/chosen": -1.8029206991195679,
|
| 241 |
+
"logits/rejected": -1.4160174131393433,
|
| 242 |
+
"logps/chosen": -489.02960205078125,
|
| 243 |
+
"logps/rejected": -732.9075927734375,
|
| 244 |
+
"loss": 0.3505,
|
| 245 |
+
"rewards/accuracies": 0.856249988079071,
|
| 246 |
+
"rewards/chosen": -1.6447805166244507,
|
| 247 |
+
"rewards/margins": 2.2803826332092285,
|
| 248 |
+
"rewards/rejected": -3.9251632690429688,
|
| 249 |
+
"step": 150
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.2220680083275503,
|
| 253 |
+
"grad_norm": 27.767234596146235,
|
| 254 |
+
"learning_rate": 4.998119881260575e-07,
|
| 255 |
+
"logits/chosen": -1.729731559753418,
|
| 256 |
+
"logits/rejected": -1.0307753086090088,
|
| 257 |
+
"logps/chosen": -489.806396484375,
|
| 258 |
+
"logps/rejected": -750.17333984375,
|
| 259 |
+
"loss": 0.3419,
|
| 260 |
+
"rewards/accuracies": 0.809374988079071,
|
| 261 |
+
"rewards/chosen": -1.7482877969741821,
|
| 262 |
+
"rewards/margins": 2.432042121887207,
|
| 263 |
+
"rewards/rejected": -4.1803297996521,
|
| 264 |
+
"step": 160
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.2359472588480222,
|
| 268 |
+
"grad_norm": 16.170525819937215,
|
| 269 |
+
"learning_rate": 4.995036332451857e-07,
|
| 270 |
+
"logits/chosen": -1.9091663360595703,
|
| 271 |
+
"logits/rejected": -0.9273399114608765,
|
| 272 |
+
"logps/chosen": -495.46661376953125,
|
| 273 |
+
"logps/rejected": -758.0303955078125,
|
| 274 |
+
"loss": 0.3289,
|
| 275 |
+
"rewards/accuracies": 0.831250011920929,
|
| 276 |
+
"rewards/chosen": -1.4501034021377563,
|
| 277 |
+
"rewards/margins": 2.5703396797180176,
|
| 278 |
+
"rewards/rejected": -4.020443439483643,
|
| 279 |
+
"step": 170
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.2498265093684941,
|
| 283 |
+
"grad_norm": 19.933464291596422,
|
| 284 |
+
"learning_rate": 4.990486745229364e-07,
|
| 285 |
+
"logits/chosen": -1.4646329879760742,
|
| 286 |
+
"logits/rejected": -0.523992121219635,
|
| 287 |
+
"logps/chosen": -475.5484924316406,
|
| 288 |
+
"logps/rejected": -712.8101806640625,
|
| 289 |
+
"loss": 0.3238,
|
| 290 |
+
"rewards/accuracies": 0.8218749761581421,
|
| 291 |
+
"rewards/chosen": -1.4877039194107056,
|
| 292 |
+
"rewards/margins": 2.2666311264038086,
|
| 293 |
+
"rewards/rejected": -3.7543349266052246,
|
| 294 |
+
"step": 180
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.263705759888966,
|
| 298 |
+
"grad_norm": 13.788976405207086,
|
| 299 |
+
"learning_rate": 4.984473792848607e-07,
|
| 300 |
+
"logits/chosen": -1.0008184909820557,
|
| 301 |
+
"logits/rejected": 0.3863092064857483,
|
| 302 |
+
"logps/chosen": -459.58367919921875,
|
| 303 |
+
"logps/rejected": -734.6851196289062,
|
| 304 |
+
"loss": 0.3131,
|
| 305 |
+
"rewards/accuracies": 0.84375,
|
| 306 |
+
"rewards/chosen": -1.3406341075897217,
|
| 307 |
+
"rewards/margins": 2.7795493602752686,
|
| 308 |
+
"rewards/rejected": -4.120182991027832,
|
| 309 |
+
"step": 190
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.2775850104094379,
|
| 313 |
+
"grad_norm": 20.384550977715232,
|
| 314 |
+
"learning_rate": 4.977001008412112e-07,
|
| 315 |
+
"logits/chosen": -0.7814763188362122,
|
| 316 |
+
"logits/rejected": 0.5796900987625122,
|
| 317 |
+
"logps/chosen": -522.8018798828125,
|
| 318 |
+
"logps/rejected": -799.6663208007812,
|
| 319 |
+
"loss": 0.333,
|
| 320 |
+
"rewards/accuracies": 0.828125,
|
| 321 |
+
"rewards/chosen": -1.9407739639282227,
|
| 322 |
+
"rewards/margins": 2.682741165161133,
|
| 323 |
+
"rewards/rejected": -4.6235151290893555,
|
| 324 |
+
"step": 200
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.2914642609299098,
|
| 328 |
+
"grad_norm": 18.232716128762664,
|
| 329 |
+
"learning_rate": 4.968072782793435e-07,
|
| 330 |
+
"logits/chosen": -0.9184268116950989,
|
| 331 |
+
"logits/rejected": 0.4480651319026947,
|
| 332 |
+
"logps/chosen": -499.62127685546875,
|
| 333 |
+
"logps/rejected": -834.462890625,
|
| 334 |
+
"loss": 0.3089,
|
| 335 |
+
"rewards/accuracies": 0.846875011920929,
|
| 336 |
+
"rewards/chosen": -1.7784448862075806,
|
| 337 |
+
"rewards/margins": 3.2359156608581543,
|
| 338 |
+
"rewards/rejected": -5.0143609046936035,
|
| 339 |
+
"step": 210
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.3053435114503817,
|
| 343 |
+
"grad_norm": 19.076527739229473,
|
| 344 |
+
"learning_rate": 4.957694362057149e-07,
|
| 345 |
+
"logits/chosen": -0.9672171473503113,
|
| 346 |
+
"logits/rejected": 0.16848711669445038,
|
| 347 |
+
"logps/chosen": -560.64453125,
|
| 348 |
+
"logps/rejected": -855.81591796875,
|
| 349 |
+
"loss": 0.3149,
|
| 350 |
+
"rewards/accuracies": 0.846875011920929,
|
| 351 |
+
"rewards/chosen": -1.9354803562164307,
|
| 352 |
+
"rewards/margins": 3.0875911712646484,
|
| 353 |
+
"rewards/rejected": -5.0230712890625,
|
| 354 |
+
"step": 220
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.3192227619708536,
|
| 358 |
+
"grad_norm": 16.258599026908875,
|
| 359 |
+
"learning_rate": 4.945871844376368e-07,
|
| 360 |
+
"logits/chosen": -1.0649207830429077,
|
| 361 |
+
"logits/rejected": 0.31253132224082947,
|
| 362 |
+
"logps/chosen": -538.4542846679688,
|
| 363 |
+
"logps/rejected": -874.3626098632812,
|
| 364 |
+
"loss": 0.3188,
|
| 365 |
+
"rewards/accuracies": 0.8968750238418579,
|
| 366 |
+
"rewards/chosen": -1.9632246494293213,
|
| 367 |
+
"rewards/margins": 3.402304172515869,
|
| 368 |
+
"rewards/rejected": -5.365528583526611,
|
| 369 |
+
"step": 230
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.3331020124913255,
|
| 373 |
+
"grad_norm": 18.705612187457067,
|
| 374 |
+
"learning_rate": 4.932612176449559e-07,
|
| 375 |
+
"logits/chosen": -0.9302694201469421,
|
| 376 |
+
"logits/rejected": 0.36828285455703735,
|
| 377 |
+
"logps/chosen": -505.27606201171875,
|
| 378 |
+
"logps/rejected": -816.0523681640625,
|
| 379 |
+
"loss": 0.3025,
|
| 380 |
+
"rewards/accuracies": 0.859375,
|
| 381 |
+
"rewards/chosen": -1.567478895187378,
|
| 382 |
+
"rewards/margins": 2.9575300216674805,
|
| 383 |
+
"rewards/rejected": -4.525008678436279,
|
| 384 |
+
"step": 240
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.3469812630117974,
|
| 388 |
+
"grad_norm": 24.629743659903227,
|
| 389 |
+
"learning_rate": 4.917923149418791e-07,
|
| 390 |
+
"logits/chosen": 0.6533899307250977,
|
| 391 |
+
"logits/rejected": 1.9370393753051758,
|
| 392 |
+
"logps/chosen": -599.1996459960938,
|
| 393 |
+
"logps/rejected": -975.5636596679688,
|
| 394 |
+
"loss": 0.2998,
|
| 395 |
+
"rewards/accuracies": 0.84375,
|
| 396 |
+
"rewards/chosen": -2.498326301574707,
|
| 397 |
+
"rewards/margins": 3.716355562210083,
|
| 398 |
+
"rewards/rejected": -6.214681625366211,
|
| 399 |
+
"step": 250
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.3608605135322693,
|
| 403 |
+
"grad_norm": 17.05308334694298,
|
| 404 |
+
"learning_rate": 4.901813394291801e-07,
|
| 405 |
+
"logits/chosen": -0.2267475426197052,
|
| 406 |
+
"logits/rejected": 1.061628818511963,
|
| 407 |
+
"logps/chosen": -510.58990478515625,
|
| 408 |
+
"logps/rejected": -825.64892578125,
|
| 409 |
+
"loss": 0.3068,
|
| 410 |
+
"rewards/accuracies": 0.8343750238418579,
|
| 411 |
+
"rewards/chosen": -1.8009761571884155,
|
| 412 |
+
"rewards/margins": 3.0452382564544678,
|
| 413 |
+
"rewards/rejected": -4.846214771270752,
|
| 414 |
+
"step": 260
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.3747397640527412,
|
| 418 |
+
"grad_norm": 16.458651892780324,
|
| 419 |
+
"learning_rate": 4.884292376870567e-07,
|
| 420 |
+
"logits/chosen": -0.5495095252990723,
|
| 421 |
+
"logits/rejected": 0.9298044443130493,
|
| 422 |
+
"logps/chosen": -518.9110107421875,
|
| 423 |
+
"logps/rejected": -838.7022705078125,
|
| 424 |
+
"loss": 0.3019,
|
| 425 |
+
"rewards/accuracies": 0.8218749761581421,
|
| 426 |
+
"rewards/chosen": -1.7417328357696533,
|
| 427 |
+
"rewards/margins": 3.202885866165161,
|
| 428 |
+
"rewards/rejected": -4.944618225097656,
|
| 429 |
+
"step": 270
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.3886190145732131,
|
| 433 |
+
"grad_norm": 17.412372040942056,
|
| 434 |
+
"learning_rate": 4.865370392189376e-07,
|
| 435 |
+
"logits/chosen": -1.044090986251831,
|
| 436 |
+
"logits/rejected": 0.6990992426872253,
|
| 437 |
+
"logps/chosen": -521.0147094726562,
|
| 438 |
+
"logps/rejected": -908.5431518554688,
|
| 439 |
+
"loss": 0.3086,
|
| 440 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 441 |
+
"rewards/chosen": -1.7731910943984985,
|
| 442 |
+
"rewards/margins": 3.855863571166992,
|
| 443 |
+
"rewards/rejected": -5.629055023193359,
|
| 444 |
+
"step": 280
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.4024982650936849,
|
| 448 |
+
"grad_norm": 18.962541699001296,
|
| 449 |
+
"learning_rate": 4.845058558465645e-07,
|
| 450 |
+
"logits/chosen": 0.2687569260597229,
|
| 451 |
+
"logits/rejected": 1.7490098476409912,
|
| 452 |
+
"logps/chosen": -562.8821411132812,
|
| 453 |
+
"logps/rejected": -911.21142578125,
|
| 454 |
+
"loss": 0.2941,
|
| 455 |
+
"rewards/accuracies": 0.862500011920929,
|
| 456 |
+
"rewards/chosen": -2.1526246070861816,
|
| 457 |
+
"rewards/margins": 3.554647445678711,
|
| 458 |
+
"rewards/rejected": -5.707272052764893,
|
| 459 |
+
"step": 290
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.4163775156141568,
|
| 463 |
+
"grad_norm": 18.283817174579646,
|
| 464 |
+
"learning_rate": 4.823368810567056e-07,
|
| 465 |
+
"logits/chosen": 0.27320951223373413,
|
| 466 |
+
"logits/rejected": 1.6727936267852783,
|
| 467 |
+
"logps/chosen": -508.94744873046875,
|
| 468 |
+
"logps/rejected": -807.9446411132812,
|
| 469 |
+
"loss": 0.3081,
|
| 470 |
+
"rewards/accuracies": 0.862500011920929,
|
| 471 |
+
"rewards/chosen": -1.861353874206543,
|
| 472 |
+
"rewards/margins": 3.0898642539978027,
|
| 473 |
+
"rewards/rejected": -4.951218605041504,
|
| 474 |
+
"step": 300
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.4302567661346287,
|
| 478 |
+
"grad_norm": 15.422967764865048,
|
| 479 |
+
"learning_rate": 4.800313892998847e-07,
|
| 480 |
+
"logits/chosen": -0.3293236196041107,
|
| 481 |
+
"logits/rejected": 1.3189566135406494,
|
| 482 |
+
"logps/chosen": -509.6499938964844,
|
| 483 |
+
"logps/rejected": -856.3865966796875,
|
| 484 |
+
"loss": 0.2919,
|
| 485 |
+
"rewards/accuracies": 0.856249988079071,
|
| 486 |
+
"rewards/chosen": -1.683760404586792,
|
| 487 |
+
"rewards/margins": 3.3120670318603516,
|
| 488 |
+
"rewards/rejected": -4.995827674865723,
|
| 489 |
+
"step": 310
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.4441360166551006,
|
| 493 |
+
"grad_norm": 18.804961569512887,
|
| 494 |
+
"learning_rate": 4.775907352415367e-07,
|
| 495 |
+
"logits/chosen": 0.22868101298809052,
|
| 496 |
+
"logits/rejected": 1.9417445659637451,
|
| 497 |
+
"logps/chosen": -556.9364013671875,
|
| 498 |
+
"logps/rejected": -950.5734252929688,
|
| 499 |
+
"loss": 0.2808,
|
| 500 |
+
"rewards/accuracies": 0.824999988079071,
|
| 501 |
+
"rewards/chosen": -1.9100916385650635,
|
| 502 |
+
"rewards/margins": 4.022894859313965,
|
| 503 |
+
"rewards/rejected": -5.932986259460449,
|
| 504 |
+
"step": 320
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 0.4580152671755725,
|
| 508 |
+
"grad_norm": 17.099019764148846,
|
| 509 |
+
"learning_rate": 4.7501635296603025e-07,
|
| 510 |
+
"logits/chosen": 1.049659013748169,
|
| 511 |
+
"logits/rejected": 2.4073307514190674,
|
| 512 |
+
"logps/chosen": -531.3202514648438,
|
| 513 |
+
"logps/rejected": -922.6915893554688,
|
| 514 |
+
"loss": 0.292,
|
| 515 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 516 |
+
"rewards/chosen": -1.8814361095428467,
|
| 517 |
+
"rewards/margins": 3.822434902191162,
|
| 518 |
+
"rewards/rejected": -5.7038702964782715,
|
| 519 |
+
"step": 330
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.4718945176960444,
|
| 523 |
+
"grad_norm": 18.898740164063316,
|
| 524 |
+
"learning_rate": 4.723097551340265e-07,
|
| 525 |
+
"logits/chosen": 1.2799204587936401,
|
| 526 |
+
"logits/rejected": 2.3121683597564697,
|
| 527 |
+
"logps/chosen": -494.0704040527344,
|
| 528 |
+
"logps/rejected": -854.8629760742188,
|
| 529 |
+
"loss": 0.292,
|
| 530 |
+
"rewards/accuracies": 0.8656250238418579,
|
| 531 |
+
"rewards/chosen": -1.7092489004135132,
|
| 532 |
+
"rewards/margins": 3.4398162364959717,
|
| 533 |
+
"rewards/rejected": -5.149065971374512,
|
| 534 |
+
"step": 340
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.4857737682165163,
|
| 538 |
+
"grad_norm": 20.5834993963944,
|
| 539 |
+
"learning_rate": 4.6947253209366613e-07,
|
| 540 |
+
"logits/chosen": 1.0228136777877808,
|
| 541 |
+
"logits/rejected": 2.397416114807129,
|
| 542 |
+
"logps/chosen": -545.7423706054688,
|
| 543 |
+
"logps/rejected": -912.7175903320312,
|
| 544 |
+
"loss": 0.2827,
|
| 545 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 546 |
+
"rewards/chosen": -2.025628089904785,
|
| 547 |
+
"rewards/margins": 3.6154162883758545,
|
| 548 |
+
"rewards/rejected": -5.6410441398620605,
|
| 549 |
+
"step": 350
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.4996530187369882,
|
| 553 |
+
"grad_norm": 14.253483616771307,
|
| 554 |
+
"learning_rate": 4.6650635094610966e-07,
|
| 555 |
+
"logits/chosen": 1.0244512557983398,
|
| 556 |
+
"logits/rejected": 2.2805395126342773,
|
| 557 |
+
"logps/chosen": -517.4966430664062,
|
| 558 |
+
"logps/rejected": -885.3656005859375,
|
| 559 |
+
"loss": 0.2767,
|
| 560 |
+
"rewards/accuracies": 0.893750011920929,
|
| 561 |
+
"rewards/chosen": -1.9133589267730713,
|
| 562 |
+
"rewards/margins": 3.5970351696014404,
|
| 563 |
+
"rewards/rejected": -5.510394096374512,
|
| 564 |
+
"step": 360
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.5135322692574601,
|
| 568 |
+
"grad_norm": 23.570388915292295,
|
| 569 |
+
"learning_rate": 4.6341295456597906e-07,
|
| 570 |
+
"logits/chosen": 0.6538206338882446,
|
| 571 |
+
"logits/rejected": 1.800736665725708,
|
| 572 |
+
"logps/chosen": -486.4225158691406,
|
| 573 |
+
"logps/rejected": -818.6160278320312,
|
| 574 |
+
"loss": 0.2972,
|
| 575 |
+
"rewards/accuracies": 0.840624988079071,
|
| 576 |
+
"rewards/chosen": -1.574295997619629,
|
| 577 |
+
"rewards/margins": 3.247091770172119,
|
| 578 |
+
"rewards/rejected": -4.821388244628906,
|
| 579 |
+
"step": 370
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.527411519777932,
|
| 583 |
+
"grad_norm": 22.92056805222348,
|
| 584 |
+
"learning_rate": 4.6019416057727577e-07,
|
| 585 |
+
"logits/chosen": 0.5617297291755676,
|
| 586 |
+
"logits/rejected": 1.9416172504425049,
|
| 587 |
+
"logps/chosen": -554.2542724609375,
|
| 588 |
+
"logps/rejected": -1017.8567504882812,
|
| 589 |
+
"loss": 0.2817,
|
| 590 |
+
"rewards/accuracies": 0.856249988079071,
|
| 591 |
+
"rewards/chosen": -1.9359180927276611,
|
| 592 |
+
"rewards/margins": 4.510493278503418,
|
| 593 |
+
"rewards/rejected": -6.4464111328125,
|
| 594 |
+
"step": 380
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 0.5412907702984039,
|
| 598 |
+
"grad_norm": 21.96046117437538,
|
| 599 |
+
"learning_rate": 4.5685186028537756e-07,
|
| 600 |
+
"logits/chosen": 0.08615957945585251,
|
| 601 |
+
"logits/rejected": 1.9250423908233643,
|
| 602 |
+
"logps/chosen": -521.3770751953125,
|
| 603 |
+
"logps/rejected": -1026.9183349609375,
|
| 604 |
+
"loss": 0.2752,
|
| 605 |
+
"rewards/accuracies": 0.903124988079071,
|
| 606 |
+
"rewards/chosen": -1.688780426979065,
|
| 607 |
+
"rewards/margins": 4.929998397827148,
|
| 608 |
+
"rewards/rejected": -6.618779182434082,
|
| 609 |
+
"step": 390
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.5551700208188758,
|
| 613 |
+
"grad_norm": 16.35709631938306,
|
| 614 |
+
"learning_rate": 4.5338801756574185e-07,
|
| 615 |
+
"logits/chosen": 0.8631726503372192,
|
| 616 |
+
"logits/rejected": 2.426805019378662,
|
| 617 |
+
"logps/chosen": -550.8397216796875,
|
| 618 |
+
"logps/rejected": -960.1754760742188,
|
| 619 |
+
"loss": 0.2975,
|
| 620 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 621 |
+
"rewards/chosen": -1.955361008644104,
|
| 622 |
+
"rewards/margins": 4.027670860290527,
|
| 623 |
+
"rewards/rejected": -5.983031749725342,
|
| 624 |
+
"step": 400
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 0.5690492713393477,
|
| 628 |
+
"grad_norm": 14.715456208460227,
|
| 629 |
+
"learning_rate": 4.498046677099674e-07,
|
| 630 |
+
"logits/chosen": 0.137635737657547,
|
| 631 |
+
"logits/rejected": 1.9270877838134766,
|
| 632 |
+
"logps/chosen": -470.42767333984375,
|
| 633 |
+
"logps/rejected": -795.9421997070312,
|
| 634 |
+
"loss": 0.2826,
|
| 635 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 636 |
+
"rewards/chosen": -1.4737566709518433,
|
| 637 |
+
"rewards/margins": 3.3337607383728027,
|
| 638 |
+
"rewards/rejected": -4.807517051696777,
|
| 639 |
+
"step": 410
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.5829285218598196,
|
| 643 |
+
"grad_norm": 15.380893236552625,
|
| 644 |
+
"learning_rate": 4.461039162298939e-07,
|
| 645 |
+
"logits/chosen": 1.0668681859970093,
|
| 646 |
+
"logits/rejected": 2.6232194900512695,
|
| 647 |
+
"logps/chosen": -546.1395874023438,
|
| 648 |
+
"logps/rejected": -970.1839599609375,
|
| 649 |
+
"loss": 0.271,
|
| 650 |
+
"rewards/accuracies": 0.8687499761581421,
|
| 651 |
+
"rewards/chosen": -1.9840848445892334,
|
| 652 |
+
"rewards/margins": 4.236409664154053,
|
| 653 |
+
"rewards/rejected": -6.220494270324707,
|
| 654 |
+
"step": 420
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.5968077723802915,
|
| 658 |
+
"grad_norm": 20.12909286909629,
|
| 659 |
+
"learning_rate": 4.4228793762044126e-07,
|
| 660 |
+
"logits/chosen": 1.8645107746124268,
|
| 661 |
+
"logits/rejected": 3.282402515411377,
|
| 662 |
+
"logps/chosen": -556.1341552734375,
|
| 663 |
+
"logps/rejected": -999.4012451171875,
|
| 664 |
+
"loss": 0.286,
|
| 665 |
+
"rewards/accuracies": 0.859375,
|
| 666 |
+
"rewards/chosen": -2.260993719100952,
|
| 667 |
+
"rewards/margins": 4.435044288635254,
|
| 668 |
+
"rewards/rejected": -6.696038246154785,
|
| 669 |
+
"step": 430
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.6106870229007634,
|
| 673 |
+
"grad_norm": 16.373631758502654,
|
| 674 |
+
"learning_rate": 4.3835897408191513e-07,
|
| 675 |
+
"logits/chosen": 1.3662347793579102,
|
| 676 |
+
"logits/rejected": 2.8811306953430176,
|
| 677 |
+
"logps/chosen": -512.532958984375,
|
| 678 |
+
"logps/rejected": -906.1589965820312,
|
| 679 |
+
"loss": 0.2862,
|
| 680 |
+
"rewards/accuracies": 0.875,
|
| 681 |
+
"rewards/chosen": -1.7359873056411743,
|
| 682 |
+
"rewards/margins": 3.9433860778808594,
|
| 683 |
+
"rewards/rejected": -5.679372787475586,
|
| 684 |
+
"step": 440
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 0.6245662734212353,
|
| 688 |
+
"grad_norm": 20.352147452552153,
|
| 689 |
+
"learning_rate": 4.34319334202531e-07,
|
| 690 |
+
"logits/chosen": 1.8553016185760498,
|
| 691 |
+
"logits/rejected": 3.245082139968872,
|
| 692 |
+
"logps/chosen": -553.4808349609375,
|
| 693 |
+
"logps/rejected": -1043.968505859375,
|
| 694 |
+
"loss": 0.2671,
|
| 695 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 696 |
+
"rewards/chosen": -2.185727834701538,
|
| 697 |
+
"rewards/margins": 4.84250545501709,
|
| 698 |
+
"rewards/rejected": -7.028233528137207,
|
| 699 |
+
"step": 450
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.6384455239417072,
|
| 703 |
+
"grad_norm": 14.530740198443842,
|
| 704 |
+
"learning_rate": 4.301713916019286e-07,
|
| 705 |
+
"logits/chosen": 1.8998327255249023,
|
| 706 |
+
"logits/rejected": 3.121591091156006,
|
| 707 |
+
"logps/chosen": -545.904296875,
|
| 708 |
+
"logps/rejected": -980.6796875,
|
| 709 |
+
"loss": 0.277,
|
| 710 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 711 |
+
"rewards/chosen": -1.9527307748794556,
|
| 712 |
+
"rewards/margins": 4.239319324493408,
|
| 713 |
+
"rewards/rejected": -6.192049980163574,
|
| 714 |
+
"step": 460
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 0.6523247744621791,
|
| 718 |
+
"grad_norm": 14.543324773787445,
|
| 719 |
+
"learning_rate": 4.2591758353647643e-07,
|
| 720 |
+
"logits/chosen": 1.0112075805664062,
|
| 721 |
+
"logits/rejected": 2.648533821105957,
|
| 722 |
+
"logps/chosen": -551.0354614257812,
|
| 723 |
+
"logps/rejected": -958.7569580078125,
|
| 724 |
+
"loss": 0.2778,
|
| 725 |
+
"rewards/accuracies": 0.8656250238418579,
|
| 726 |
+
"rewards/chosen": -1.892492651939392,
|
| 727 |
+
"rewards/margins": 4.073439598083496,
|
| 728 |
+
"rewards/rejected": -5.965932846069336,
|
| 729 |
+
"step": 470
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.666204024982651,
|
| 733 |
+
"grad_norm": 20.665751040711644,
|
| 734 |
+
"learning_rate": 4.2156040946718343e-07,
|
| 735 |
+
"logits/chosen": 1.5330281257629395,
|
| 736 |
+
"logits/rejected": 3.1015706062316895,
|
| 737 |
+
"logps/chosen": -588.2686157226562,
|
| 738 |
+
"logps/rejected": -1030.510498046875,
|
| 739 |
+
"loss": 0.2843,
|
| 740 |
+
"rewards/accuracies": 0.878125011920929,
|
| 741 |
+
"rewards/chosen": -2.3880248069763184,
|
| 742 |
+
"rewards/margins": 4.504977226257324,
|
| 743 |
+
"rewards/rejected": -6.893001556396484,
|
| 744 |
+
"step": 480
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.6800832755031229,
|
| 748 |
+
"grad_norm": 15.05601087138497,
|
| 749 |
+
"learning_rate": 4.1710242959106056e-07,
|
| 750 |
+
"logits/chosen": 0.5176582336425781,
|
| 751 |
+
"logits/rejected": 2.4557127952575684,
|
| 752 |
+
"logps/chosen": -507.008544921875,
|
| 753 |
+
"logps/rejected": -886.51416015625,
|
| 754 |
+
"loss": 0.294,
|
| 755 |
+
"rewards/accuracies": 0.846875011920929,
|
| 756 |
+
"rewards/chosen": -1.5845837593078613,
|
| 757 |
+
"rewards/margins": 3.810807704925537,
|
| 758 |
+
"rewards/rejected": -5.395391464233398,
|
| 759 |
+
"step": 490
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.6939625260235948,
|
| 763 |
+
"grad_norm": 16.718450087742482,
|
| 764 |
+
"learning_rate": 4.125462633367959e-07,
|
| 765 |
+
"logits/chosen": 1.4314930438995361,
|
| 766 |
+
"logits/rejected": 2.9996209144592285,
|
| 767 |
+
"logps/chosen": -532.1755981445312,
|
| 768 |
+
"logps/rejected": -1021.0213623046875,
|
| 769 |
+
"loss": 0.266,
|
| 770 |
+
"rewards/accuracies": 0.878125011920929,
|
| 771 |
+
"rewards/chosen": -2.0497827529907227,
|
| 772 |
+
"rewards/margins": 4.718562126159668,
|
| 773 |
+
"rewards/rejected": -6.768344879150391,
|
| 774 |
+
"step": 500
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 0.7078417765440667,
|
| 778 |
+
"grad_norm": 19.85458205442182,
|
| 779 |
+
"learning_rate": 4.0789458782562435e-07,
|
| 780 |
+
"logits/chosen": 1.293492317199707,
|
| 781 |
+
"logits/rejected": 2.812124013900757,
|
| 782 |
+
"logps/chosen": -551.9437866210938,
|
| 783 |
+
"logps/rejected": -1100.38720703125,
|
| 784 |
+
"loss": 0.2609,
|
| 785 |
+
"rewards/accuracies": 0.887499988079071,
|
| 786 |
+
"rewards/chosen": -2.3401436805725098,
|
| 787 |
+
"rewards/margins": 5.148820400238037,
|
| 788 |
+
"rewards/rejected": -7.4889631271362305,
|
| 789 |
+
"step": 510
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.7217210270645386,
|
| 793 |
+
"grad_norm": 14.477360180818952,
|
| 794 |
+
"learning_rate": 4.031501362983007e-07,
|
| 795 |
+
"logits/chosen": 0.7156480550765991,
|
| 796 |
+
"logits/rejected": 2.6717166900634766,
|
| 797 |
+
"logps/chosen": -510.4054260253906,
|
| 798 |
+
"logps/rejected": -1019.923828125,
|
| 799 |
+
"loss": 0.2805,
|
| 800 |
+
"rewards/accuracies": 0.887499988079071,
|
| 801 |
+
"rewards/chosen": -1.7749239206314087,
|
| 802 |
+
"rewards/margins": 4.909750938415527,
|
| 803 |
+
"rewards/rejected": -6.6846747398376465,
|
| 804 |
+
"step": 520
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 0.7356002775850105,
|
| 808 |
+
"grad_norm": 17.058981872372126,
|
| 809 |
+
"learning_rate": 3.9831569650909553e-07,
|
| 810 |
+
"logits/chosen": 0.01914766989648342,
|
| 811 |
+
"logits/rejected": 2.122251510620117,
|
| 812 |
+
"logps/chosen": -551.1067504882812,
|
| 813 |
+
"logps/rejected": -933.5851440429688,
|
| 814 |
+
"loss": 0.2736,
|
| 815 |
+
"rewards/accuracies": 0.846875011920929,
|
| 816 |
+
"rewards/chosen": -2.0334556102752686,
|
| 817 |
+
"rewards/margins": 3.904972791671753,
|
| 818 |
+
"rewards/rejected": -5.9384284019470215,
|
| 819 |
+
"step": 530
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 0.7494795281054824,
|
| 823 |
+
"grad_norm": 19.755034017893955,
|
| 824 |
+
"learning_rate": 3.933941090877615e-07,
|
| 825 |
+
"logits/chosen": -0.34956273436546326,
|
| 826 |
+
"logits/rejected": 2.1510236263275146,
|
| 827 |
+
"logps/chosen": -553.6760864257812,
|
| 828 |
+
"logps/rejected": -1035.0694580078125,
|
| 829 |
+
"loss": 0.2598,
|
| 830 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 831 |
+
"rewards/chosen": -2.195675849914551,
|
| 832 |
+
"rewards/margins": 4.830143928527832,
|
| 833 |
+
"rewards/rejected": -7.025820732116699,
|
| 834 |
+
"step": 540
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 0.7633587786259542,
|
| 838 |
+
"grad_norm": 15.194278158742483,
|
| 839 |
+
"learning_rate": 3.883882658704306e-07,
|
| 840 |
+
"logits/chosen": -0.1008995771408081,
|
| 841 |
+
"logits/rejected": 2.1417980194091797,
|
| 842 |
+
"logps/chosen": -562.4782104492188,
|
| 843 |
+
"logps/rejected": -1068.9832763671875,
|
| 844 |
+
"loss": 0.2769,
|
| 845 |
+
"rewards/accuracies": 0.8656250238418579,
|
| 846 |
+
"rewards/chosen": -2.3539323806762695,
|
| 847 |
+
"rewards/margins": 5.010300636291504,
|
| 848 |
+
"rewards/rejected": -7.364233493804932,
|
| 849 |
+
"step": 550
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.7772380291464261,
|
| 853 |
+
"grad_norm": 17.553015958549693,
|
| 854 |
+
"learning_rate": 3.833011082004228e-07,
|
| 855 |
+
"logits/chosen": -0.8681015968322754,
|
| 856 |
+
"logits/rejected": 1.4696462154388428,
|
| 857 |
+
"logps/chosen": -573.9002685546875,
|
| 858 |
+
"logps/rejected": -1092.654296875,
|
| 859 |
+
"loss": 0.2772,
|
| 860 |
+
"rewards/accuracies": 0.828125,
|
| 861 |
+
"rewards/chosen": -2.370056390762329,
|
| 862 |
+
"rewards/margins": 5.011393070220947,
|
| 863 |
+
"rewards/rejected": -7.381448268890381,
|
| 864 |
+
"step": 560
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.7911172796668979,
|
| 868 |
+
"grad_norm": 15.00893179617878,
|
| 869 |
+
"learning_rate": 3.781356251999663e-07,
|
| 870 |
+
"logits/chosen": -1.1506744623184204,
|
| 871 |
+
"logits/rejected": 0.8982523679733276,
|
| 872 |
+
"logps/chosen": -535.85595703125,
|
| 873 |
+
"logps/rejected": -979.8224487304688,
|
| 874 |
+
"loss": 0.2904,
|
| 875 |
+
"rewards/accuracies": 0.862500011920929,
|
| 876 |
+
"rewards/chosen": -2.1066572666168213,
|
| 877 |
+
"rewards/margins": 4.214187145233154,
|
| 878 |
+
"rewards/rejected": -6.320844650268555,
|
| 879 |
+
"step": 570
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 0.8049965301873698,
|
| 883 |
+
"grad_norm": 17.833191557593317,
|
| 884 |
+
"learning_rate": 3.728948520138426e-07,
|
| 885 |
+
"logits/chosen": -0.45408257842063904,
|
| 886 |
+
"logits/rejected": 1.915353775024414,
|
| 887 |
+
"logps/chosen": -540.576171875,
|
| 888 |
+
"logps/rejected": -994.28564453125,
|
| 889 |
+
"loss": 0.2725,
|
| 890 |
+
"rewards/accuracies": 0.859375,
|
| 891 |
+
"rewards/chosen": -2.1601402759552,
|
| 892 |
+
"rewards/margins": 4.492788314819336,
|
| 893 |
+
"rewards/rejected": -6.652928829193115,
|
| 894 |
+
"step": 580
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 0.8188757807078417,
|
| 898 |
+
"grad_norm": 16.37664073810551,
|
| 899 |
+
"learning_rate": 3.6758186802599064e-07,
|
| 900 |
+
"logits/chosen": 0.19857454299926758,
|
| 901 |
+
"logits/rejected": 2.2812840938568115,
|
| 902 |
+
"logps/chosen": -550.7555541992188,
|
| 903 |
+
"logps/rejected": -1052.941162109375,
|
| 904 |
+
"loss": 0.2681,
|
| 905 |
+
"rewards/accuracies": 0.878125011920929,
|
| 906 |
+
"rewards/chosen": -2.186647415161133,
|
| 907 |
+
"rewards/margins": 4.857481956481934,
|
| 908 |
+
"rewards/rejected": -7.044129848480225,
|
| 909 |
+
"step": 590
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 0.8327550312283136,
|
| 913 |
+
"grad_norm": 15.489252159692187,
|
| 914 |
+
"learning_rate": 3.6219979505011555e-07,
|
| 915 |
+
"logits/chosen": 0.13169285655021667,
|
| 916 |
+
"logits/rejected": 2.4862585067749023,
|
| 917 |
+
"logps/chosen": -545.8336181640625,
|
| 918 |
+
"logps/rejected": -1015.2984619140625,
|
| 919 |
+
"loss": 0.2599,
|
| 920 |
+
"rewards/accuracies": 0.8843749761581421,
|
| 921 |
+
"rewards/chosen": -2.0449271202087402,
|
| 922 |
+
"rewards/margins": 4.806704521179199,
|
| 923 |
+
"rewards/rejected": -6.851631164550781,
|
| 924 |
+
"step": 600
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 0.8466342817487855,
|
| 928 |
+
"grad_norm": 16.034098237394367,
|
| 929 |
+
"learning_rate": 3.5675179549536786e-07,
|
| 930 |
+
"logits/chosen": 1.026650309562683,
|
| 931 |
+
"logits/rejected": 2.891331672668457,
|
| 932 |
+
"logps/chosen": -562.756103515625,
|
| 933 |
+
"logps/rejected": -1112.7884521484375,
|
| 934 |
+
"loss": 0.2655,
|
| 935 |
+
"rewards/accuracies": 0.871874988079071,
|
| 936 |
+
"rewards/chosen": -2.344050884246826,
|
| 937 |
+
"rewards/margins": 5.390646934509277,
|
| 938 |
+
"rewards/rejected": -7.7346978187561035,
|
| 939 |
+
"step": 610
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 0.8605135322692574,
|
| 943 |
+
"grad_norm": 15.964353907588587,
|
| 944 |
+
"learning_rate": 3.512410705081684e-07,
|
| 945 |
+
"logits/chosen": 0.5978251695632935,
|
| 946 |
+
"logits/rejected": 2.6756691932678223,
|
| 947 |
+
"logps/chosen": -596.0946655273438,
|
| 948 |
+
"logps/rejected": -1164.9000244140625,
|
| 949 |
+
"loss": 0.2654,
|
| 950 |
+
"rewards/accuracies": 0.890625,
|
| 951 |
+
"rewards/chosen": -2.3005499839782715,
|
| 952 |
+
"rewards/margins": 5.751313209533691,
|
| 953 |
+
"rewards/rejected": -8.051862716674805,
|
| 954 |
+
"step": 620
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.8743927827897293,
|
| 958 |
+
"grad_norm": 15.534990200131912,
|
| 959 |
+
"learning_rate": 3.4567085809127245e-07,
|
| 960 |
+
"logits/chosen": 0.7654634714126587,
|
| 961 |
+
"logits/rejected": 2.563817262649536,
|
| 962 |
+
"logps/chosen": -553.0380859375,
|
| 963 |
+
"logps/rejected": -1015.8709106445312,
|
| 964 |
+
"loss": 0.2837,
|
| 965 |
+
"rewards/accuracies": 0.8343750238418579,
|
| 966 |
+
"rewards/chosen": -2.105071544647217,
|
| 967 |
+
"rewards/margins": 4.519524097442627,
|
| 968 |
+
"rewards/rejected": -6.624594688415527,
|
| 969 |
+
"step": 630
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.8882720333102012,
|
| 973 |
+
"grad_norm": 15.473999576589652,
|
| 974 |
+
"learning_rate": 3.400444312011776e-07,
|
| 975 |
+
"logits/chosen": 1.6810909509658813,
|
| 976 |
+
"logits/rejected": 3.29026460647583,
|
| 977 |
+
"logps/chosen": -547.7821044921875,
|
| 978 |
+
"logps/rejected": -979.19580078125,
|
| 979 |
+
"loss": 0.2808,
|
| 980 |
+
"rewards/accuracies": 0.871874988079071,
|
| 981 |
+
"rewards/chosen": -2.2526745796203613,
|
| 982 |
+
"rewards/margins": 4.376776695251465,
|
| 983 |
+
"rewards/rejected": -6.629450798034668,
|
| 984 |
+
"step": 640
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 0.9021512838306731,
|
| 988 |
+
"grad_norm": 18.65677766855292,
|
| 989 |
+
"learning_rate": 3.343650958249935e-07,
|
| 990 |
+
"logits/chosen": 1.9508390426635742,
|
| 991 |
+
"logits/rejected": 3.384152889251709,
|
| 992 |
+
"logps/chosen": -611.1751708984375,
|
| 993 |
+
"logps/rejected": -1097.154541015625,
|
| 994 |
+
"loss": 0.2717,
|
| 995 |
+
"rewards/accuracies": 0.875,
|
| 996 |
+
"rewards/chosen": -2.623706340789795,
|
| 997 |
+
"rewards/margins": 4.866621494293213,
|
| 998 |
+
"rewards/rejected": -7.49032735824585,
|
| 999 |
+
"step": 650
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 0.916030534351145,
|
| 1003 |
+
"grad_norm": 15.172341090722101,
|
| 1004 |
+
"learning_rate": 3.286361890379034e-07,
|
| 1005 |
+
"logits/chosen": 1.418536901473999,
|
| 1006 |
+
"logits/rejected": 2.9168481826782227,
|
| 1007 |
+
"logps/chosen": -542.9044799804688,
|
| 1008 |
+
"logps/rejected": -998.8190307617188,
|
| 1009 |
+
"loss": 0.2649,
|
| 1010 |
+
"rewards/accuracies": 0.878125011920929,
|
| 1011 |
+
"rewards/chosen": -2.128884792327881,
|
| 1012 |
+
"rewards/margins": 4.480809688568115,
|
| 1013 |
+
"rewards/rejected": -6.6096954345703125,
|
| 1014 |
+
"step": 660
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 0.9299097848716169,
|
| 1018 |
+
"grad_norm": 21.03897492647871,
|
| 1019 |
+
"learning_rate": 3.2286107704235875e-07,
|
| 1020 |
+
"logits/chosen": 0.6511715650558472,
|
| 1021 |
+
"logits/rejected": 2.4751362800598145,
|
| 1022 |
+
"logps/chosen": -516.9407348632812,
|
| 1023 |
+
"logps/rejected": -994.5736083984375,
|
| 1024 |
+
"loss": 0.2687,
|
| 1025 |
+
"rewards/accuracies": 0.856249988079071,
|
| 1026 |
+
"rewards/chosen": -1.8717918395996094,
|
| 1027 |
+
"rewards/margins": 4.618001937866211,
|
| 1028 |
+
"rewards/rejected": -6.489793300628662,
|
| 1029 |
+
"step": 670
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 0.9437890353920888,
|
| 1033 |
+
"grad_norm": 23.147160211659624,
|
| 1034 |
+
"learning_rate": 3.1704315319015936e-07,
|
| 1035 |
+
"logits/chosen": 0.42685967683792114,
|
| 1036 |
+
"logits/rejected": 2.4200854301452637,
|
| 1037 |
+
"logps/chosen": -569.0222778320312,
|
| 1038 |
+
"logps/rejected": -1023.0499877929688,
|
| 1039 |
+
"loss": 0.2637,
|
| 1040 |
+
"rewards/accuracies": 0.856249988079071,
|
| 1041 |
+
"rewards/chosen": -2.241736650466919,
|
| 1042 |
+
"rewards/margins": 4.569226264953613,
|
| 1043 |
+
"rewards/rejected": -6.8109636306762695,
|
| 1044 |
+
"step": 680
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 0.9576682859125607,
|
| 1048 |
+
"grad_norm": 19.074650599712797,
|
| 1049 |
+
"learning_rate": 3.1118583598858094e-07,
|
| 1050 |
+
"logits/chosen": 0.3251928687095642,
|
| 1051 |
+
"logits/rejected": 2.361527442932129,
|
| 1052 |
+
"logps/chosen": -536.3818969726562,
|
| 1053 |
+
"logps/rejected": -1084.29150390625,
|
| 1054 |
+
"loss": 0.269,
|
| 1055 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 1056 |
+
"rewards/chosen": -2.159126043319702,
|
| 1057 |
+
"rewards/margins": 5.291468620300293,
|
| 1058 |
+
"rewards/rejected": -7.450594425201416,
|
| 1059 |
+
"step": 690
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.9715475364330326,
|
| 1063 |
+
"grad_norm": 16.46319504999634,
|
| 1064 |
+
"learning_rate": 3.052925670917219e-07,
|
| 1065 |
+
"logits/chosen": 0.3188991844654083,
|
| 1066 |
+
"logits/rejected": 2.1961517333984375,
|
| 1067 |
+
"logps/chosen": -590.7283935546875,
|
| 1068 |
+
"logps/rejected": -1070.7264404296875,
|
| 1069 |
+
"loss": 0.2754,
|
| 1070 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 1071 |
+
"rewards/chosen": -2.466348171234131,
|
| 1072 |
+
"rewards/margins": 4.865278720855713,
|
| 1073 |
+
"rewards/rejected": -7.33162784576416,
|
| 1074 |
+
"step": 700
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.9854267869535045,
|
| 1078 |
+
"grad_norm": 22.275614803003936,
|
| 1079 |
+
"learning_rate": 2.9936680927824935e-07,
|
| 1080 |
+
"logits/chosen": 0.7669705152511597,
|
| 1081 |
+
"logits/rejected": 2.4150214195251465,
|
| 1082 |
+
"logps/chosen": -574.384033203125,
|
| 1083 |
+
"logps/rejected": -1084.837646484375,
|
| 1084 |
+
"loss": 0.2543,
|
| 1085 |
+
"rewards/accuracies": 0.856249988079071,
|
| 1086 |
+
"rewards/chosen": -2.393491268157959,
|
| 1087 |
+
"rewards/margins": 5.025046348571777,
|
| 1088 |
+
"rewards/rejected": -7.4185380935668945,
|
| 1089 |
+
"step": 710
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 0.9993060374739764,
|
| 1093 |
+
"grad_norm": 22.856869601641986,
|
| 1094 |
+
"learning_rate": 2.934120444167326e-07,
|
| 1095 |
+
"logits/chosen": 0.018332133069634438,
|
| 1096 |
+
"logits/rejected": 1.9890985488891602,
|
| 1097 |
+
"logps/chosen": -554.123291015625,
|
| 1098 |
+
"logps/rejected": -1072.498046875,
|
| 1099 |
+
"loss": 0.2651,
|
| 1100 |
+
"rewards/accuracies": 0.8656250238418579,
|
| 1101 |
+
"rewards/chosen": -2.201577663421631,
|
| 1102 |
+
"rewards/margins": 5.162604331970215,
|
| 1103 |
+
"rewards/rejected": -7.3641815185546875,
|
| 1104 |
+
"step": 720
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 1.0131852879944483,
|
| 1108 |
+
"grad_norm": 16.323336520132766,
|
| 1109 |
+
"learning_rate": 2.8743177141975993e-07,
|
| 1110 |
+
"logits/chosen": 0.2875978350639343,
|
| 1111 |
+
"logits/rejected": 2.0843756198883057,
|
| 1112 |
+
"logps/chosen": -598.0647583007812,
|
| 1113 |
+
"logps/rejected": -1214.6568603515625,
|
| 1114 |
+
"loss": 0.1848,
|
| 1115 |
+
"rewards/accuracies": 0.921875,
|
| 1116 |
+
"rewards/chosen": -2.478050708770752,
|
| 1117 |
+
"rewards/margins": 6.1301589012146,
|
| 1118 |
+
"rewards/rejected": -8.608209609985352,
|
| 1119 |
+
"step": 730
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 1.0270645385149202,
|
| 1123 |
+
"grad_norm": 17.717617068699322,
|
| 1124 |
+
"learning_rate": 2.814295041880407e-07,
|
| 1125 |
+
"logits/chosen": 1.5749400854110718,
|
| 1126 |
+
"logits/rejected": 3.2800605297088623,
|
| 1127 |
+
"logps/chosen": -720.4405517578125,
|
| 1128 |
+
"logps/rejected": -1425.5921630859375,
|
| 1129 |
+
"loss": 0.1738,
|
| 1130 |
+
"rewards/accuracies": 0.9156249761581421,
|
| 1131 |
+
"rewards/chosen": -3.6376540660858154,
|
| 1132 |
+
"rewards/margins": 6.913857460021973,
|
| 1133 |
+
"rewards/rejected": -10.55151081085205,
|
| 1134 |
+
"step": 740
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 1.040943789035392,
|
| 1138 |
+
"grad_norm": 16.138966847361562,
|
| 1139 |
+
"learning_rate": 2.754087695457005e-07,
|
| 1140 |
+
"logits/chosen": 1.3446929454803467,
|
| 1141 |
+
"logits/rejected": 3.647613525390625,
|
| 1142 |
+
"logps/chosen": -660.2525024414062,
|
| 1143 |
+
"logps/rejected": -1274.7076416015625,
|
| 1144 |
+
"loss": 0.1695,
|
| 1145 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1146 |
+
"rewards/chosen": -3.0320277214050293,
|
| 1147 |
+
"rewards/margins": 6.136539936065674,
|
| 1148 |
+
"rewards/rejected": -9.168566703796387,
|
| 1149 |
+
"step": 750
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 1.054823039555864,
|
| 1153 |
+
"grad_norm": 15.364782345427345,
|
| 1154 |
+
"learning_rate": 2.6937310516798275e-07,
|
| 1155 |
+
"logits/chosen": 1.2250503301620483,
|
| 1156 |
+
"logits/rejected": 3.3649864196777344,
|
| 1157 |
+
"logps/chosen": -629.1475830078125,
|
| 1158 |
+
"logps/rejected": -1302.7933349609375,
|
| 1159 |
+
"loss": 0.1799,
|
| 1160 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 1161 |
+
"rewards/chosen": -2.8058159351348877,
|
| 1162 |
+
"rewards/margins": 6.627597808837891,
|
| 1163 |
+
"rewards/rejected": -9.4334135055542,
|
| 1164 |
+
"step": 760
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 1.0687022900763359,
|
| 1168 |
+
"grad_norm": 18.530915715706502,
|
| 1169 |
+
"learning_rate": 2.6332605750257456e-07,
|
| 1170 |
+
"logits/chosen": 0.9841400384902954,
|
| 1171 |
+
"logits/rejected": 3.341764450073242,
|
| 1172 |
+
"logps/chosen": -656.33251953125,
|
| 1173 |
+
"logps/rejected": -1349.439697265625,
|
| 1174 |
+
"loss": 0.1607,
|
| 1175 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1176 |
+
"rewards/chosen": -3.064764976501465,
|
| 1177 |
+
"rewards/margins": 6.942727565765381,
|
| 1178 |
+
"rewards/rejected": -10.007492065429688,
|
| 1179 |
+
"step": 770
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 1.0825815405968078,
|
| 1183 |
+
"grad_norm": 18.346652449927596,
|
| 1184 |
+
"learning_rate": 2.5727117968577785e-07,
|
| 1185 |
+
"logits/chosen": 1.088732361793518,
|
| 1186 |
+
"logits/rejected": 3.094304323196411,
|
| 1187 |
+
"logps/chosen": -656.88134765625,
|
| 1188 |
+
"logps/rejected": -1397.916259765625,
|
| 1189 |
+
"loss": 0.1514,
|
| 1190 |
+
"rewards/accuracies": 0.918749988079071,
|
| 1191 |
+
"rewards/chosen": -3.20575213432312,
|
| 1192 |
+
"rewards/margins": 7.160604000091553,
|
| 1193 |
+
"rewards/rejected": -10.366357803344727,
|
| 1194 |
+
"step": 780
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 1.0964607911172797,
|
| 1198 |
+
"grad_norm": 17.314542797248876,
|
| 1199 |
+
"learning_rate": 2.5121202945475043e-07,
|
| 1200 |
+
"logits/chosen": 1.5640583038330078,
|
| 1201 |
+
"logits/rejected": 3.803560256958008,
|
| 1202 |
+
"logps/chosen": -630.0179443359375,
|
| 1203 |
+
"logps/rejected": -1349.983642578125,
|
| 1204 |
+
"loss": 0.178,
|
| 1205 |
+
"rewards/accuracies": 0.909375011920929,
|
| 1206 |
+
"rewards/chosen": -2.848733425140381,
|
| 1207 |
+
"rewards/margins": 7.226857662200928,
|
| 1208 |
+
"rewards/rejected": -10.075590133666992,
|
| 1209 |
+
"step": 790
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 1.1103400416377516,
|
| 1213 |
+
"grad_norm": 14.955502239768675,
|
| 1214 |
+
"learning_rate": 2.4515216705704393e-07,
|
| 1215 |
+
"logits/chosen": 1.8664920330047607,
|
| 1216 |
+
"logits/rejected": 3.7396292686462402,
|
| 1217 |
+
"logps/chosen": -608.3690795898438,
|
| 1218 |
+
"logps/rejected": -1366.815673828125,
|
| 1219 |
+
"loss": 0.1637,
|
| 1220 |
+
"rewards/accuracies": 0.953125,
|
| 1221 |
+
"rewards/chosen": -2.7840514183044434,
|
| 1222 |
+
"rewards/margins": 7.429004669189453,
|
| 1223 |
+
"rewards/rejected": -10.213056564331055,
|
| 1224 |
+
"step": 800
|
| 1225 |
+
},
|
| 1226 |
+
{
|
| 1227 |
+
"epoch": 1.1242192921582235,
|
| 1228 |
+
"grad_norm": 18.71030295389215,
|
| 1229 |
+
"learning_rate": 2.39095153158666e-07,
|
| 1230 |
+
"logits/chosen": 1.694549560546875,
|
| 1231 |
+
"logits/rejected": 3.747912645339966,
|
| 1232 |
+
"logps/chosen": -657.016845703125,
|
| 1233 |
+
"logps/rejected": -1396.2716064453125,
|
| 1234 |
+
"loss": 0.1581,
|
| 1235 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 1236 |
+
"rewards/chosen": -3.1420130729675293,
|
| 1237 |
+
"rewards/margins": 7.294529914855957,
|
| 1238 |
+
"rewards/rejected": -10.436542510986328,
|
| 1239 |
+
"step": 810
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 1.1380985426786954,
|
| 1243 |
+
"grad_norm": 14.829984041140372,
|
| 1244 |
+
"learning_rate": 2.330445467518977e-07,
|
| 1245 |
+
"logits/chosen": 0.7256888747215271,
|
| 1246 |
+
"logits/rejected": 2.9225716590881348,
|
| 1247 |
+
"logps/chosen": -660.5631103515625,
|
| 1248 |
+
"logps/rejected": -1347.63134765625,
|
| 1249 |
+
"loss": 0.1619,
|
| 1250 |
+
"rewards/accuracies": 0.918749988079071,
|
| 1251 |
+
"rewards/chosen": -3.208239793777466,
|
| 1252 |
+
"rewards/margins": 6.785590171813965,
|
| 1253 |
+
"rewards/rejected": -9.993829727172852,
|
| 1254 |
+
"step": 820
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 1.1519777931991673,
|
| 1258 |
+
"grad_norm": 22.137973289587368,
|
| 1259 |
+
"learning_rate": 2.270039030640931e-07,
|
| 1260 |
+
"logits/chosen": 1.578162431716919,
|
| 1261 |
+
"logits/rejected": 3.47196888923645,
|
| 1262 |
+
"logps/chosen": -628.019287109375,
|
| 1263 |
+
"logps/rejected": -1320.9600830078125,
|
| 1264 |
+
"loss": 0.1749,
|
| 1265 |
+
"rewards/accuracies": 0.893750011920929,
|
| 1266 |
+
"rewards/chosen": -2.9641802310943604,
|
| 1267 |
+
"rewards/margins": 6.771336555480957,
|
| 1268 |
+
"rewards/rejected": -9.735516548156738,
|
| 1269 |
+
"step": 830
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 1.1658570437196392,
|
| 1273 |
+
"grad_norm": 19.216307710411925,
|
| 1274 |
+
"learning_rate": 2.209767714686924e-07,
|
| 1275 |
+
"logits/chosen": 1.5978351831436157,
|
| 1276 |
+
"logits/rejected": 3.7053027153015137,
|
| 1277 |
+
"logps/chosen": -653.354736328125,
|
| 1278 |
+
"logps/rejected": -1406.637451171875,
|
| 1279 |
+
"loss": 0.1606,
|
| 1280 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1281 |
+
"rewards/chosen": -3.2107303142547607,
|
| 1282 |
+
"rewards/margins": 7.3466925621032715,
|
| 1283 |
+
"rewards/rejected": -10.557422637939453,
|
| 1284 |
+
"step": 840
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 1.179736294240111,
|
| 1288 |
+
"grad_norm": 21.415640762520873,
|
| 1289 |
+
"learning_rate": 2.1496669339967344e-07,
|
| 1290 |
+
"logits/chosen": 1.8785479068756104,
|
| 1291 |
+
"logits/rejected": 3.8594677448272705,
|
| 1292 |
+
"logps/chosen": -668.572265625,
|
| 1293 |
+
"logps/rejected": -1460.521240234375,
|
| 1294 |
+
"loss": 0.1611,
|
| 1295 |
+
"rewards/accuracies": 0.9156249761581421,
|
| 1296 |
+
"rewards/chosen": -3.3157646656036377,
|
| 1297 |
+
"rewards/margins": 7.8289475440979,
|
| 1298 |
+
"rewards/rejected": -11.144712448120117,
|
| 1299 |
+
"step": 850
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 1.193615544760583,
|
| 1303 |
+
"grad_norm": 19.68985482074639,
|
| 1304 |
+
"learning_rate": 2.0897720027066897e-07,
|
| 1305 |
+
"logits/chosen": 1.5298289060592651,
|
| 1306 |
+
"logits/rejected": 3.88330078125,
|
| 1307 |
+
"logps/chosen": -660.7781982421875,
|
| 1308 |
+
"logps/rejected": -1415.609130859375,
|
| 1309 |
+
"loss": 0.1637,
|
| 1310 |
+
"rewards/accuracies": 0.9437500238418579,
|
| 1311 |
+
"rewards/chosen": -3.227867603302002,
|
| 1312 |
+
"rewards/margins": 7.6040802001953125,
|
| 1313 |
+
"rewards/rejected": -10.831949234008789,
|
| 1314 |
+
"step": 860
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 1.2074947952810549,
|
| 1318 |
+
"grad_norm": 19.76316476703645,
|
| 1319 |
+
"learning_rate": 2.0301181139997202e-07,
|
| 1320 |
+
"logits/chosen": 1.5017510652542114,
|
| 1321 |
+
"logits/rejected": 3.601763963699341,
|
| 1322 |
+
"logps/chosen": -664.7364501953125,
|
| 1323 |
+
"logps/rejected": -1365.153564453125,
|
| 1324 |
+
"loss": 0.1716,
|
| 1325 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 1326 |
+
"rewards/chosen": -3.1189022064208984,
|
| 1327 |
+
"rewards/margins": 6.980868339538574,
|
| 1328 |
+
"rewards/rejected": -10.099771499633789,
|
| 1329 |
+
"step": 870
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 1.2213740458015268,
|
| 1333 |
+
"grad_norm": 15.231393479477973,
|
| 1334 |
+
"learning_rate": 1.970740319426474e-07,
|
| 1335 |
+
"logits/chosen": 2.0277457237243652,
|
| 1336 |
+
"logits/rejected": 3.8836147785186768,
|
| 1337 |
+
"logps/chosen": -615.454833984375,
|
| 1338 |
+
"logps/rejected": -1219.456298828125,
|
| 1339 |
+
"loss": 0.1803,
|
| 1340 |
+
"rewards/accuracies": 0.9156249761581421,
|
| 1341 |
+
"rewards/chosen": -2.8338332176208496,
|
| 1342 |
+
"rewards/margins": 5.970818519592285,
|
| 1343 |
+
"rewards/rejected": -8.804651260375977,
|
| 1344 |
+
"step": 880
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 1.2352532963219987,
|
| 1348 |
+
"grad_norm": 23.68508206139152,
|
| 1349 |
+
"learning_rate": 1.911673508309656e-07,
|
| 1350 |
+
"logits/chosen": 1.9813998937606812,
|
| 1351 |
+
"logits/rejected": 3.9024734497070312,
|
| 1352 |
+
"logps/chosen": -628.2916259765625,
|
| 1353 |
+
"logps/rejected": -1264.0699462890625,
|
| 1354 |
+
"loss": 0.1643,
|
| 1355 |
+
"rewards/accuracies": 0.9437500238418579,
|
| 1356 |
+
"rewards/chosen": -2.8454182147979736,
|
| 1357 |
+
"rewards/margins": 6.396851539611816,
|
| 1358 |
+
"rewards/rejected": -9.242269515991211,
|
| 1359 |
+
"step": 890
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 1.2491325468424705,
|
| 1363 |
+
"grad_norm": 19.952403887659834,
|
| 1364 |
+
"learning_rate": 1.8529523872436977e-07,
|
| 1365 |
+
"logits/chosen": 1.032080054283142,
|
| 1366 |
+
"logits/rejected": 3.210120677947998,
|
| 1367 |
+
"logps/chosen": -671.1250610351562,
|
| 1368 |
+
"logps/rejected": -1549.0615234375,
|
| 1369 |
+
"loss": 0.16,
|
| 1370 |
+
"rewards/accuracies": 0.956250011920929,
|
| 1371 |
+
"rewards/chosen": -3.1327102184295654,
|
| 1372 |
+
"rewards/margins": 8.625667572021484,
|
| 1373 |
+
"rewards/rejected": -11.758378982543945,
|
| 1374 |
+
"step": 900
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 1.2630117973629424,
|
| 1378 |
+
"grad_norm": 22.651112148813382,
|
| 1379 |
+
"learning_rate": 1.7946114597017808e-07,
|
| 1380 |
+
"logits/chosen": 1.6984176635742188,
|
| 1381 |
+
"logits/rejected": 3.7526144981384277,
|
| 1382 |
+
"logps/chosen": -660.7144165039062,
|
| 1383 |
+
"logps/rejected": -1447.436767578125,
|
| 1384 |
+
"loss": 0.1619,
|
| 1385 |
+
"rewards/accuracies": 0.940625011920929,
|
| 1386 |
+
"rewards/chosen": -3.2229180335998535,
|
| 1387 |
+
"rewards/margins": 7.7218732833862305,
|
| 1388 |
+
"rewards/rejected": -10.944790840148926,
|
| 1389 |
+
"step": 910
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 1.2768910478834143,
|
| 1393 |
+
"grad_norm": 22.126302318317872,
|
| 1394 |
+
"learning_rate": 1.7366850057622172e-07,
|
| 1395 |
+
"logits/chosen": 1.6554454565048218,
|
| 1396 |
+
"logits/rejected": 3.758930206298828,
|
| 1397 |
+
"logps/chosen": -654.619140625,
|
| 1398 |
+
"logps/rejected": -1429.7637939453125,
|
| 1399 |
+
"loss": 0.1653,
|
| 1400 |
+
"rewards/accuracies": 0.9468749761581421,
|
| 1401 |
+
"rewards/chosen": -3.282761812210083,
|
| 1402 |
+
"rewards/margins": 7.560622215270996,
|
| 1403 |
+
"rewards/rejected": -10.8433837890625,
|
| 1404 |
+
"step": 920
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 1.2907702984038862,
|
| 1408 |
+
"grad_norm": 17.175640354603786,
|
| 1409 |
+
"learning_rate": 1.6792070619660974e-07,
|
| 1410 |
+
"logits/chosen": 1.6677768230438232,
|
| 1411 |
+
"logits/rejected": 3.8167777061462402,
|
| 1412 |
+
"logps/chosen": -654.5197143554688,
|
| 1413 |
+
"logps/rejected": -1449.1417236328125,
|
| 1414 |
+
"loss": 0.1654,
|
| 1415 |
+
"rewards/accuracies": 0.9375,
|
| 1416 |
+
"rewards/chosen": -3.2044639587402344,
|
| 1417 |
+
"rewards/margins": 7.7717132568359375,
|
| 1418 |
+
"rewards/rejected": -10.976176261901855,
|
| 1419 |
+
"step": 930
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 1.3046495489243581,
|
| 1423 |
+
"grad_norm": 19.342272367773113,
|
| 1424 |
+
"learning_rate": 1.622211401318028e-07,
|
| 1425 |
+
"logits/chosen": 1.567354440689087,
|
| 1426 |
+
"logits/rejected": 4.129142761230469,
|
| 1427 |
+
"logps/chosen": -657.1278686523438,
|
| 1428 |
+
"logps/rejected": -1453.904541015625,
|
| 1429 |
+
"loss": 0.1595,
|
| 1430 |
+
"rewards/accuracies": 0.9437500238418579,
|
| 1431 |
+
"rewards/chosen": -3.066843271255493,
|
| 1432 |
+
"rewards/margins": 7.886987209320068,
|
| 1433 |
+
"rewards/rejected": -10.953829765319824,
|
| 1434 |
+
"step": 940
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 1.31852879944483,
|
| 1438 |
+
"grad_norm": 23.040092913699233,
|
| 1439 |
+
"learning_rate": 1.5657315134417244e-07,
|
| 1440 |
+
"logits/chosen": 1.8149755001068115,
|
| 1441 |
+
"logits/rejected": 4.2323408126831055,
|
| 1442 |
+
"logps/chosen": -688.6448364257812,
|
| 1443 |
+
"logps/rejected": -1585.87841796875,
|
| 1444 |
+
"loss": 0.1665,
|
| 1445 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1446 |
+
"rewards/chosen": -3.4780020713806152,
|
| 1447 |
+
"rewards/margins": 8.797907829284668,
|
| 1448 |
+
"rewards/rejected": -12.275908470153809,
|
| 1449 |
+
"step": 950
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 1.332408049965302,
|
| 1453 |
+
"grad_norm": 17.336766543398397,
|
| 1454 |
+
"learning_rate": 1.5098005849021078e-07,
|
| 1455 |
+
"logits/chosen": 1.7025401592254639,
|
| 1456 |
+
"logits/rejected": 4.534233570098877,
|
| 1457 |
+
"logps/chosen": -703.167724609375,
|
| 1458 |
+
"logps/rejected": -1630.665283203125,
|
| 1459 |
+
"loss": 0.1672,
|
| 1460 |
+
"rewards/accuracies": 0.9281250238418579,
|
| 1461 |
+
"rewards/chosen": -3.5511913299560547,
|
| 1462 |
+
"rewards/margins": 9.27750301361084,
|
| 1463 |
+
"rewards/rejected": -12.828694343566895,
|
| 1464 |
+
"step": 960
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 1.3462873004857738,
|
| 1468 |
+
"grad_norm": 20.603249005514577,
|
| 1469 |
+
"learning_rate": 1.454451479705484e-07,
|
| 1470 |
+
"logits/chosen": 2.531970500946045,
|
| 1471 |
+
"logits/rejected": 4.76376485824585,
|
| 1472 |
+
"logps/chosen": -658.04931640625,
|
| 1473 |
+
"logps/rejected": -1445.8599853515625,
|
| 1474 |
+
"loss": 0.1611,
|
| 1475 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 1476 |
+
"rewards/chosen": -3.303406238555908,
|
| 1477 |
+
"rewards/margins": 7.701096534729004,
|
| 1478 |
+
"rewards/rejected": -11.00450325012207,
|
| 1479 |
+
"step": 970
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 1.3601665510062457,
|
| 1483 |
+
"grad_norm": 26.637520706621387,
|
| 1484 |
+
"learning_rate": 1.3997167199892385e-07,
|
| 1485 |
+
"logits/chosen": 2.412917375564575,
|
| 1486 |
+
"logits/rejected": 4.9495530128479,
|
| 1487 |
+
"logps/chosen": -706.7518310546875,
|
| 1488 |
+
"logps/rejected": -1550.7310791015625,
|
| 1489 |
+
"loss": 0.16,
|
| 1490 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1491 |
+
"rewards/chosen": -3.5566086769104004,
|
| 1492 |
+
"rewards/margins": 8.430598258972168,
|
| 1493 |
+
"rewards/rejected": -11.987205505371094,
|
| 1494 |
+
"step": 980
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 1.3740458015267176,
|
| 1498 |
+
"grad_norm": 22.552075685705777,
|
| 1499 |
+
"learning_rate": 1.3456284669124157e-07,
|
| 1500 |
+
"logits/chosen": 2.5810890197753906,
|
| 1501 |
+
"logits/rejected": 4.767507076263428,
|
| 1502 |
+
"logps/chosen": -692.7928466796875,
|
| 1503 |
+
"logps/rejected": -1483.0191650390625,
|
| 1504 |
+
"loss": 0.1661,
|
| 1505 |
+
"rewards/accuracies": 0.9437500238418579,
|
| 1506 |
+
"rewards/chosen": -3.5196609497070312,
|
| 1507 |
+
"rewards/margins": 7.722577095031738,
|
| 1508 |
+
"rewards/rejected": -11.24223804473877,
|
| 1509 |
+
"step": 990
|
| 1510 |
+
},
|
| 1511 |
+
{
|
| 1512 |
+
"epoch": 1.3879250520471895,
|
| 1513 |
+
"grad_norm": 18.63062778053758,
|
| 1514 |
+
"learning_rate": 1.2922185017584036e-07,
|
| 1515 |
+
"logits/chosen": 2.9480197429656982,
|
| 1516 |
+
"logits/rejected": 5.796414375305176,
|
| 1517 |
+
"logps/chosen": -692.9706420898438,
|
| 1518 |
+
"logps/rejected": -1486.6146240234375,
|
| 1519 |
+
"loss": 0.1587,
|
| 1520 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1521 |
+
"rewards/chosen": -3.4646847248077393,
|
| 1522 |
+
"rewards/margins": 7.979840278625488,
|
| 1523 |
+
"rewards/rejected": -11.444524765014648,
|
| 1524 |
+
"step": 1000
|
| 1525 |
+
},
|
| 1526 |
+
{
|
| 1527 |
+
"epoch": 1.3879250520471895,
|
| 1528 |
+
"eval_logits/chosen": 2.4397730827331543,
|
| 1529 |
+
"eval_logits/rejected": 4.560609817504883,
|
| 1530 |
+
"eval_logps/chosen": -779.3055419921875,
|
| 1531 |
+
"eval_logps/rejected": -1613.5015869140625,
|
| 1532 |
+
"eval_loss": 0.24712695181369781,
|
| 1533 |
+
"eval_rewards/accuracies": 0.8909774422645569,
|
| 1534 |
+
"eval_rewards/chosen": -3.947213649749756,
|
| 1535 |
+
"eval_rewards/margins": 8.086578369140625,
|
| 1536 |
+
"eval_rewards/rejected": -12.033791542053223,
|
| 1537 |
+
"eval_runtime": 385.5049,
|
| 1538 |
+
"eval_samples_per_second": 22.026,
|
| 1539 |
+
"eval_steps_per_second": 0.345,
|
| 1540 |
+
"step": 1000
|
| 1541 |
+
},
|
| 1542 |
+
{
|
| 1543 |
+
"epoch": 1.4018043025676614,
|
| 1544 |
+
"grad_norm": 17.720503294892175,
|
| 1545 |
+
"learning_rate": 1.2395182072608245e-07,
|
| 1546 |
+
"logits/chosen": 2.783371925354004,
|
| 1547 |
+
"logits/rejected": 5.293555736541748,
|
| 1548 |
+
"logps/chosen": -693.8554077148438,
|
| 1549 |
+
"logps/rejected": -1455.6026611328125,
|
| 1550 |
+
"loss": 0.1578,
|
| 1551 |
+
"rewards/accuracies": 0.918749988079071,
|
| 1552 |
+
"rewards/chosen": -3.711820125579834,
|
| 1553 |
+
"rewards/margins": 7.577448844909668,
|
| 1554 |
+
"rewards/rejected": -11.289270401000977,
|
| 1555 |
+
"step": 1010
|
| 1556 |
+
},
|
| 1557 |
+
{
|
| 1558 |
+
"epoch": 1.4156835530881333,
|
| 1559 |
+
"grad_norm": 39.790575003507485,
|
| 1560 |
+
"learning_rate": 1.1875585491635998e-07,
|
| 1561 |
+
"logits/chosen": 3.120913028717041,
|
| 1562 |
+
"logits/rejected": 5.793875217437744,
|
| 1563 |
+
"logps/chosen": -757.794921875,
|
| 1564 |
+
"logps/rejected": -1674.984375,
|
| 1565 |
+
"loss": 0.1529,
|
| 1566 |
+
"rewards/accuracies": 0.9375,
|
| 1567 |
+
"rewards/chosen": -4.312003135681152,
|
| 1568 |
+
"rewards/margins": 8.977596282958984,
|
| 1569 |
+
"rewards/rejected": -13.289599418640137,
|
| 1570 |
+
"step": 1020
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 1.4295628036086052,
|
| 1574 |
+
"grad_norm": 69.28439112059148,
|
| 1575 |
+
"learning_rate": 1.1363700580260438e-07,
|
| 1576 |
+
"logits/chosen": 3.0012755393981934,
|
| 1577 |
+
"logits/rejected": 5.11987829208374,
|
| 1578 |
+
"logps/chosen": -737.9568481445312,
|
| 1579 |
+
"logps/rejected": -1678.4951171875,
|
| 1580 |
+
"loss": 0.161,
|
| 1581 |
+
"rewards/accuracies": 0.8968750238418579,
|
| 1582 |
+
"rewards/chosen": -4.191954612731934,
|
| 1583 |
+
"rewards/margins": 9.063277244567871,
|
| 1584 |
+
"rewards/rejected": -13.255231857299805,
|
| 1585 |
+
"step": 1030
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 1.4434420541290771,
|
| 1589 |
+
"grad_norm": 18.312529268747,
|
| 1590 |
+
"learning_rate": 1.0859828112836539e-07,
|
| 1591 |
+
"logits/chosen": 2.285982847213745,
|
| 1592 |
+
"logits/rejected": 4.8581390380859375,
|
| 1593 |
+
"logps/chosen": -739.9261474609375,
|
| 1594 |
+
"logps/rejected": -1586.6966552734375,
|
| 1595 |
+
"loss": 0.1639,
|
| 1596 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 1597 |
+
"rewards/chosen": -3.9794273376464844,
|
| 1598 |
+
"rewards/margins": 8.331348419189453,
|
| 1599 |
+
"rewards/rejected": -12.310776710510254,
|
| 1600 |
+
"step": 1040
|
| 1601 |
+
},
|
| 1602 |
+
{
|
| 1603 |
+
"epoch": 1.457321304649549,
|
| 1604 |
+
"grad_norm": 25.436043150499486,
|
| 1605 |
+
"learning_rate": 1.0364264155751487e-07,
|
| 1606 |
+
"logits/chosen": 2.474147081375122,
|
| 1607 |
+
"logits/rejected": 4.805792331695557,
|
| 1608 |
+
"logps/chosen": -723.0252685546875,
|
| 1609 |
+
"logps/rejected": -1613.719970703125,
|
| 1610 |
+
"loss": 0.1599,
|
| 1611 |
+
"rewards/accuracies": 0.9281250238418579,
|
| 1612 |
+
"rewards/chosen": -4.043066501617432,
|
| 1613 |
+
"rewards/margins": 8.693578720092773,
|
| 1614 |
+
"rewards/rejected": -12.736645698547363,
|
| 1615 |
+
"step": 1050
|
| 1616 |
+
},
|
| 1617 |
+
{
|
| 1618 |
+
"epoch": 1.4712005551700207,
|
| 1619 |
+
"grad_norm": 18.578653218988183,
|
| 1620 |
+
"learning_rate": 9.877299893461455e-08,
|
| 1621 |
+
"logits/chosen": 2.1861484050750732,
|
| 1622 |
+
"logits/rejected": 5.042906761169434,
|
| 1623 |
+
"logps/chosen": -748.6861572265625,
|
| 1624 |
+
"logps/rejected": -1565.2005615234375,
|
| 1625 |
+
"loss": 0.1539,
|
| 1626 |
+
"rewards/accuracies": 0.9468749761581421,
|
| 1627 |
+
"rewards/chosen": -4.071117401123047,
|
| 1628 |
+
"rewards/margins": 8.12061595916748,
|
| 1629 |
+
"rewards/rejected": -12.191734313964844,
|
| 1630 |
+
"step": 1060
|
| 1631 |
+
},
|
| 1632 |
+
{
|
| 1633 |
+
"epoch": 1.4850798056904928,
|
| 1634 |
+
"grad_norm": 21.104448029863153,
|
| 1635 |
+
"learning_rate": 9.39922145739683e-08,
|
| 1636 |
+
"logits/chosen": 2.260671854019165,
|
| 1637 |
+
"logits/rejected": 4.919951438903809,
|
| 1638 |
+
"logps/chosen": -785.3671875,
|
| 1639 |
+
"logps/rejected": -1578.3946533203125,
|
| 1640 |
+
"loss": 0.1559,
|
| 1641 |
+
"rewards/accuracies": 0.9437500238418579,
|
| 1642 |
+
"rewards/chosen": -4.341281414031982,
|
| 1643 |
+
"rewards/margins": 7.95664119720459,
|
| 1644 |
+
"rewards/rejected": -12.29792308807373,
|
| 1645 |
+
"step": 1070
|
| 1646 |
+
},
|
| 1647 |
+
{
|
| 1648 |
+
"epoch": 1.4989590562109645,
|
| 1649 |
+
"grad_norm": 30.412482027711867,
|
| 1650 |
+
"learning_rate": 8.930309757836516e-08,
|
| 1651 |
+
"logits/chosen": 2.812704563140869,
|
| 1652 |
+
"logits/rejected": 5.714043140411377,
|
| 1653 |
+
"logps/chosen": -803.9837036132812,
|
| 1654 |
+
"logps/rejected": -1662.1617431640625,
|
| 1655 |
+
"loss": 0.1624,
|
| 1656 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 1657 |
+
"rewards/chosen": -4.532739162445068,
|
| 1658 |
+
"rewards/margins": 8.574421882629395,
|
| 1659 |
+
"rewards/rejected": -13.107160568237305,
|
| 1660 |
+
"step": 1080
|
| 1661 |
+
},
|
| 1662 |
+
{
|
| 1663 |
+
"epoch": 1.5128383067314366,
|
| 1664 |
+
"grad_norm": 27.560035648494992,
|
| 1665 |
+
"learning_rate": 8.470840318850168e-08,
|
| 1666 |
+
"logits/chosen": 2.2375552654266357,
|
| 1667 |
+
"logits/rejected": 5.399850368499756,
|
| 1668 |
+
"logps/chosen": -772.5816650390625,
|
| 1669 |
+
"logps/rejected": -1577.226806640625,
|
| 1670 |
+
"loss": 0.1576,
|
| 1671 |
+
"rewards/accuracies": 0.953125,
|
| 1672 |
+
"rewards/chosen": -4.188956260681152,
|
| 1673 |
+
"rewards/margins": 8.190386772155762,
|
| 1674 |
+
"rewards/rejected": -12.379343032836914,
|
| 1675 |
+
"step": 1090
|
| 1676 |
+
},
|
| 1677 |
+
{
|
| 1678 |
+
"epoch": 1.5267175572519083,
|
| 1679 |
+
"grad_norm": 21.688406462998472,
|
| 1680 |
+
"learning_rate": 8.021083116405173e-08,
|
| 1681 |
+
"logits/chosen": 2.3147189617156982,
|
| 1682 |
+
"logits/rejected": 5.235360145568848,
|
| 1683 |
+
"logps/chosen": -789.7681884765625,
|
| 1684 |
+
"logps/rejected": -1536.5394287109375,
|
| 1685 |
+
"loss": 0.1576,
|
| 1686 |
+
"rewards/accuracies": 0.9281250238418579,
|
| 1687 |
+
"rewards/chosen": -4.277130126953125,
|
| 1688 |
+
"rewards/margins": 7.6214165687561035,
|
| 1689 |
+
"rewards/rejected": -11.898547172546387,
|
| 1690 |
+
"step": 1100
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 1.5405968077723804,
|
| 1694 |
+
"grad_norm": 17.348108637258065,
|
| 1695 |
+
"learning_rate": 7.581302419733632e-08,
|
| 1696 |
+
"logits/chosen": 2.7001595497131348,
|
| 1697 |
+
"logits/rejected": 5.277278900146484,
|
| 1698 |
+
"logps/chosen": -735.4856567382812,
|
| 1699 |
+
"logps/rejected": -1611.0947265625,
|
| 1700 |
+
"loss": 0.1519,
|
| 1701 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1702 |
+
"rewards/chosen": -4.2074875831604,
|
| 1703 |
+
"rewards/margins": 8.55141830444336,
|
| 1704 |
+
"rewards/rejected": -12.758907318115234,
|
| 1705 |
+
"step": 1110
|
| 1706 |
+
},
|
| 1707 |
+
{
|
| 1708 |
+
"epoch": 1.554476058292852,
|
| 1709 |
+
"grad_norm": 26.00795087450143,
|
| 1710 |
+
"learning_rate": 7.151756636052527e-08,
|
| 1711 |
+
"logits/chosen": 2.411917209625244,
|
| 1712 |
+
"logits/rejected": 5.364395618438721,
|
| 1713 |
+
"logps/chosen": -790.9634399414062,
|
| 1714 |
+
"logps/rejected": -1729.780517578125,
|
| 1715 |
+
"loss": 0.1535,
|
| 1716 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1717 |
+
"rewards/chosen": -4.578307151794434,
|
| 1718 |
+
"rewards/margins": 9.278319358825684,
|
| 1719 |
+
"rewards/rejected": -13.856626510620117,
|
| 1720 |
+
"step": 1120
|
| 1721 |
+
},
|
| 1722 |
+
{
|
| 1723 |
+
"epoch": 1.5683553088133242,
|
| 1724 |
+
"grad_norm": 19.597840266227173,
|
| 1725 |
+
"learning_rate": 6.732698158728315e-08,
|
| 1726 |
+
"logits/chosen": 2.3103692531585693,
|
| 1727 |
+
"logits/rejected": 5.478982448577881,
|
| 1728 |
+
"logps/chosen": -766.3836059570312,
|
| 1729 |
+
"logps/rejected": -1625.908203125,
|
| 1730 |
+
"loss": 0.1485,
|
| 1731 |
+
"rewards/accuracies": 0.934374988079071,
|
| 1732 |
+
"rewards/chosen": -4.306478977203369,
|
| 1733 |
+
"rewards/margins": 8.519509315490723,
|
| 1734 |
+
"rewards/rejected": -12.82598876953125,
|
| 1735 |
+
"step": 1130
|
| 1736 |
+
},
|
| 1737 |
+
{
|
| 1738 |
+
"epoch": 1.5822345593337959,
|
| 1739 |
+
"grad_norm": 37.32265853585954,
|
| 1740 |
+
"learning_rate": 6.324373218975104e-08,
|
| 1741 |
+
"logits/chosen": 2.6391475200653076,
|
| 1742 |
+
"logits/rejected": 5.588366508483887,
|
| 1743 |
+
"logps/chosen": -738.453125,
|
| 1744 |
+
"logps/rejected": -1503.0093994140625,
|
| 1745 |
+
"loss": 0.1743,
|
| 1746 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 1747 |
+
"rewards/chosen": -4.1589436531066895,
|
| 1748 |
+
"rewards/margins": 7.595011234283447,
|
| 1749 |
+
"rewards/rejected": -11.75395393371582,
|
| 1750 |
+
"step": 1140
|
| 1751 |
+
},
|
| 1752 |
+
{
|
| 1753 |
+
"epoch": 1.596113809854268,
|
| 1754 |
+
"grad_norm": 34.02286786789855,
|
| 1755 |
+
"learning_rate": 5.927021741173624e-08,
|
| 1756 |
+
"logits/chosen": 2.2584009170532227,
|
| 1757 |
+
"logits/rejected": 5.177404403686523,
|
| 1758 |
+
"logps/chosen": -716.9951171875,
|
| 1759 |
+
"logps/rejected": -1559.7662353515625,
|
| 1760 |
+
"loss": 0.1604,
|
| 1761 |
+
"rewards/accuracies": 0.9468749761581421,
|
| 1762 |
+
"rewards/chosen": -4.111263275146484,
|
| 1763 |
+
"rewards/margins": 8.235260963439941,
|
| 1764 |
+
"rewards/rejected": -12.346525192260742,
|
| 1765 |
+
"step": 1150
|
| 1766 |
+
},
|
| 1767 |
+
{
|
| 1768 |
+
"epoch": 1.6099930603747397,
|
| 1769 |
+
"grad_norm": 20.932494952987984,
|
| 1770 |
+
"learning_rate": 5.5408772018959996e-08,
|
| 1771 |
+
"logits/chosen": 2.0977630615234375,
|
| 1772 |
+
"logits/rejected": 5.0792131423950195,
|
| 1773 |
+
"logps/chosen": -744.5228271484375,
|
| 1774 |
+
"logps/rejected": -1543.290283203125,
|
| 1775 |
+
"loss": 0.1549,
|
| 1776 |
+
"rewards/accuracies": 0.9156249761581421,
|
| 1777 |
+
"rewards/chosen": -3.9662253856658936,
|
| 1778 |
+
"rewards/margins": 7.838561058044434,
|
| 1779 |
+
"rewards/rejected": -11.80478572845459,
|
| 1780 |
+
"step": 1160
|
| 1781 |
+
},
|
| 1782 |
+
{
|
| 1783 |
+
"epoch": 1.6238723108952118,
|
| 1784 |
+
"grad_norm": 18.156628188332558,
|
| 1785 |
+
"learning_rate": 5.166166492719124e-08,
|
| 1786 |
+
"logits/chosen": 2.717636823654175,
|
| 1787 |
+
"logits/rejected": 5.455983638763428,
|
| 1788 |
+
"logps/chosen": -789.8887939453125,
|
| 1789 |
+
"logps/rejected": -1659.1243896484375,
|
| 1790 |
+
"loss": 0.1431,
|
| 1791 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 1792 |
+
"rewards/chosen": -4.5077619552612305,
|
| 1793 |
+
"rewards/margins": 8.738015174865723,
|
| 1794 |
+
"rewards/rejected": -13.245776176452637,
|
| 1795 |
+
"step": 1170
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 1.6377515614156835,
|
| 1799 |
+
"grad_norm": 31.62818641679718,
|
| 1800 |
+
"learning_rate": 4.8031097869072225e-08,
|
| 1801 |
+
"logits/chosen": 2.479032516479492,
|
| 1802 |
+
"logits/rejected": 5.483838081359863,
|
| 1803 |
+
"logps/chosen": -843.8903198242188,
|
| 1804 |
+
"logps/rejected": -1682.486328125,
|
| 1805 |
+
"loss": 0.1808,
|
| 1806 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 1807 |
+
"rewards/chosen": -4.988056182861328,
|
| 1808 |
+
"rewards/margins": 8.428102493286133,
|
| 1809 |
+
"rewards/rejected": -13.416158676147461,
|
| 1810 |
+
"step": 1180
|
| 1811 |
+
},
|
| 1812 |
+
{
|
| 1813 |
+
"epoch": 1.6516308119361556,
|
| 1814 |
+
"grad_norm": 23.367654380160257,
|
| 1815 |
+
"learning_rate": 4.451920410042048e-08,
|
| 1816 |
+
"logits/chosen": 2.4087483882904053,
|
| 1817 |
+
"logits/rejected": 5.190215110778809,
|
| 1818 |
+
"logps/chosen": -769.288818359375,
|
| 1819 |
+
"logps/rejected": -1569.325927734375,
|
| 1820 |
+
"loss": 0.1421,
|
| 1821 |
+
"rewards/accuracies": 0.956250011920929,
|
| 1822 |
+
"rewards/chosen": -4.326132297515869,
|
| 1823 |
+
"rewards/margins": 7.9340925216674805,
|
| 1824 |
+
"rewards/rejected": -12.260224342346191,
|
| 1825 |
+
"step": 1190
|
| 1826 |
+
},
|
| 1827 |
+
{
|
| 1828 |
+
"epoch": 1.6655100624566272,
|
| 1829 |
+
"grad_norm": 19.705530443812812,
|
| 1830 |
+
"learning_rate": 4.112804714676593e-08,
|
| 1831 |
+
"logits/chosen": 2.707061529159546,
|
| 1832 |
+
"logits/rejected": 5.535910606384277,
|
| 1833 |
+
"logps/chosen": -801.08935546875,
|
| 1834 |
+
"logps/rejected": -1702.814453125,
|
| 1835 |
+
"loss": 0.1442,
|
| 1836 |
+
"rewards/accuracies": 0.953125,
|
| 1837 |
+
"rewards/chosen": -4.545718193054199,
|
| 1838 |
+
"rewards/margins": 8.991861343383789,
|
| 1839 |
+
"rewards/rejected": -13.537579536437988,
|
| 1840 |
+
"step": 1200
|
| 1841 |
+
},
|
| 1842 |
+
{
|
| 1843 |
+
"epoch": 1.6793893129770994,
|
| 1844 |
+
"grad_norm": 18.481993519274262,
|
| 1845 |
+
"learning_rate": 3.785961959086026e-08,
|
| 1846 |
+
"logits/chosen": 2.372300624847412,
|
| 1847 |
+
"logits/rejected": 4.822690010070801,
|
| 1848 |
+
"logps/chosen": -803.9580688476562,
|
| 1849 |
+
"logps/rejected": -1608.513671875,
|
| 1850 |
+
"loss": 0.1649,
|
| 1851 |
+
"rewards/accuracies": 0.9281250238418579,
|
| 1852 |
+
"rewards/chosen": -4.596296787261963,
|
| 1853 |
+
"rewards/margins": 7.976349830627441,
|
| 1854 |
+
"rewards/rejected": -12.572647094726562,
|
| 1855 |
+
"step": 1210
|
| 1856 |
+
},
|
| 1857 |
+
{
|
| 1858 |
+
"epoch": 1.693268563497571,
|
| 1859 |
+
"grad_norm": 17.963706056533137,
|
| 1860 |
+
"learning_rate": 3.4715841901871545e-08,
|
| 1861 |
+
"logits/chosen": 2.230515956878662,
|
| 1862 |
+
"logits/rejected": 4.730217456817627,
|
| 1863 |
+
"logps/chosen": -780.9340209960938,
|
| 1864 |
+
"logps/rejected": -1548.3511962890625,
|
| 1865 |
+
"loss": 0.1478,
|
| 1866 |
+
"rewards/accuracies": 0.956250011920929,
|
| 1867 |
+
"rewards/chosen": -4.49829626083374,
|
| 1868 |
+
"rewards/margins": 7.652662754058838,
|
| 1869 |
+
"rewards/rejected": -12.150960922241211,
|
| 1870 |
+
"step": 1220
|
| 1871 |
+
},
|
| 1872 |
+
{
|
| 1873 |
+
"epoch": 1.7071478140180432,
|
| 1874 |
+
"grad_norm": 19.452721114097585,
|
| 1875 |
+
"learning_rate": 3.169856130695106e-08,
|
| 1876 |
+
"logits/chosen": 2.7935097217559814,
|
| 1877 |
+
"logits/rejected": 5.348397254943848,
|
| 1878 |
+
"logps/chosen": -837.3346557617188,
|
| 1879 |
+
"logps/rejected": -1694.122314453125,
|
| 1880 |
+
"loss": 0.1637,
|
| 1881 |
+
"rewards/accuracies": 0.90625,
|
| 1882 |
+
"rewards/chosen": -4.844791412353516,
|
| 1883 |
+
"rewards/margins": 8.404356002807617,
|
| 1884 |
+
"rewards/rejected": -13.249147415161133,
|
| 1885 |
+
"step": 1230
|
| 1886 |
+
},
|
| 1887 |
+
{
|
| 1888 |
+
"epoch": 1.7210270645385148,
|
| 1889 |
+
"grad_norm": 24.207735428419646,
|
| 1890 |
+
"learning_rate": 2.8809550705835546e-08,
|
| 1891 |
+
"logits/chosen": 2.0396385192871094,
|
| 1892 |
+
"logits/rejected": 5.347299098968506,
|
| 1893 |
+
"logps/chosen": -809.9459838867188,
|
| 1894 |
+
"logps/rejected": -1699.661865234375,
|
| 1895 |
+
"loss": 0.1595,
|
| 1896 |
+
"rewards/accuracies": 0.9468749761581421,
|
| 1897 |
+
"rewards/chosen": -4.592886447906494,
|
| 1898 |
+
"rewards/margins": 8.931785583496094,
|
| 1899 |
+
"rewards/rejected": -13.52467155456543,
|
| 1900 |
+
"step": 1240
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"epoch": 1.734906315058987,
|
| 1904 |
+
"grad_norm": 24.611504297665018,
|
| 1905 |
+
"learning_rate": 2.6050507629123724e-08,
|
| 1906 |
+
"logits/chosen": 2.3445913791656494,
|
| 1907 |
+
"logits/rejected": 5.292250633239746,
|
| 1908 |
+
"logps/chosen": -763.6495361328125,
|
| 1909 |
+
"logps/rejected": -1591.069091796875,
|
| 1910 |
+
"loss": 0.1746,
|
| 1911 |
+
"rewards/accuracies": 0.921875,
|
| 1912 |
+
"rewards/chosen": -4.202335834503174,
|
| 1913 |
+
"rewards/margins": 8.237665176391602,
|
| 1914 |
+
"rewards/rejected": -12.440000534057617,
|
| 1915 |
+
"step": 1250
|
| 1916 |
+
},
|
| 1917 |
+
{
|
| 1918 |
+
"epoch": 1.7487855655794586,
|
| 1919 |
+
"grad_norm": 17.351441576745916,
|
| 1920 |
+
"learning_rate": 2.3423053240837514e-08,
|
| 1921 |
+
"logits/chosen": 2.1456151008605957,
|
| 1922 |
+
"logits/rejected": 5.197157859802246,
|
| 1923 |
+
"logps/chosen": -775.988525390625,
|
| 1924 |
+
"logps/rejected": -1635.9970703125,
|
| 1925 |
+
"loss": 0.1602,
|
| 1926 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1927 |
+
"rewards/chosen": -4.428194999694824,
|
| 1928 |
+
"rewards/margins": 8.511189460754395,
|
| 1929 |
+
"rewards/rejected": -12.939386367797852,
|
| 1930 |
+
"step": 1260
|
| 1931 |
+
},
|
| 1932 |
+
{
|
| 1933 |
+
"epoch": 1.7626648160999308,
|
| 1934 |
+
"grad_norm": 23.373280732240797,
|
| 1935 |
+
"learning_rate": 2.0928731385855548e-08,
|
| 1936 |
+
"logits/chosen": 2.3203883171081543,
|
| 1937 |
+
"logits/rejected": 5.307148456573486,
|
| 1938 |
+
"logps/chosen": -750.8080444335938,
|
| 1939 |
+
"logps/rejected": -1596.9986572265625,
|
| 1940 |
+
"loss": 0.1503,
|
| 1941 |
+
"rewards/accuracies": 0.921875,
|
| 1942 |
+
"rewards/chosen": -4.260045051574707,
|
| 1943 |
+
"rewards/margins": 8.443410873413086,
|
| 1944 |
+
"rewards/rejected": -12.703454971313477,
|
| 1945 |
+
"step": 1270
|
| 1946 |
+
},
|
| 1947 |
+
{
|
| 1948 |
+
"epoch": 1.7765440666204024,
|
| 1949 |
+
"grad_norm": 21.91095019107269,
|
| 1950 |
+
"learning_rate": 1.8569007682777415e-08,
|
| 1951 |
+
"logits/chosen": 2.3937430381774902,
|
| 1952 |
+
"logits/rejected": 5.018430233001709,
|
| 1953 |
+
"logps/chosen": -775.3260498046875,
|
| 1954 |
+
"logps/rejected": -1653.3775634765625,
|
| 1955 |
+
"loss": 0.1517,
|
| 1956 |
+
"rewards/accuracies": 0.9468749761581421,
|
| 1957 |
+
"rewards/chosen": -4.4354681968688965,
|
| 1958 |
+
"rewards/margins": 8.597478866577148,
|
| 1959 |
+
"rewards/rejected": -13.032946586608887,
|
| 1960 |
+
"step": 1280
|
| 1961 |
+
},
|
| 1962 |
+
{
|
| 1963 |
+
"epoch": 1.7904233171408745,
|
| 1964 |
+
"grad_norm": 18.785456105009327,
|
| 1965 |
+
"learning_rate": 1.6345268662752904e-08,
|
| 1966 |
+
"logits/chosen": 2.1837244033813477,
|
| 1967 |
+
"logits/rejected": 4.996502876281738,
|
| 1968 |
+
"logps/chosen": -797.1134643554688,
|
| 1969 |
+
"logps/rejected": -1595.029541015625,
|
| 1970 |
+
"loss": 0.1469,
|
| 1971 |
+
"rewards/accuracies": 0.934374988079071,
|
| 1972 |
+
"rewards/chosen": -4.397826671600342,
|
| 1973 |
+
"rewards/margins": 7.9338555335998535,
|
| 1974 |
+
"rewards/rejected": -12.331681251525879,
|
| 1975 |
+
"step": 1290
|
| 1976 |
+
},
|
| 1977 |
+
{
|
| 1978 |
+
"epoch": 1.8043025676613462,
|
| 1979 |
+
"grad_norm": 22.286605985901687,
|
| 1980 |
+
"learning_rate": 1.4258820954781037e-08,
|
| 1981 |
+
"logits/chosen": 2.1744418144226074,
|
| 1982 |
+
"logits/rejected": 5.030113697052002,
|
| 1983 |
+
"logps/chosen": -767.9185180664062,
|
| 1984 |
+
"logps/rejected": -1573.216552734375,
|
| 1985 |
+
"loss": 0.1686,
|
| 1986 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1987 |
+
"rewards/chosen": -4.429121971130371,
|
| 1988 |
+
"rewards/margins": 7.945558071136475,
|
| 1989 |
+
"rewards/rejected": -12.37468147277832,
|
| 1990 |
+
"step": 1300
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 1.8181818181818183,
|
| 1994 |
+
"grad_norm": 25.01416006471772,
|
| 1995 |
+
"learning_rate": 1.2310890517958389e-08,
|
| 1996 |
+
"logits/chosen": 2.1604392528533936,
|
| 1997 |
+
"logits/rejected": 5.2982378005981445,
|
| 1998 |
+
"logps/chosen": -799.98046875,
|
| 1999 |
+
"logps/rejected": -1684.6910400390625,
|
| 2000 |
+
"loss": 0.1451,
|
| 2001 |
+
"rewards/accuracies": 0.918749988079071,
|
| 2002 |
+
"rewards/chosen": -4.48465633392334,
|
| 2003 |
+
"rewards/margins": 8.867886543273926,
|
| 2004 |
+
"rewards/rejected": -13.352543830871582,
|
| 2005 |
+
"step": 1310
|
| 2006 |
+
},
|
| 2007 |
+
{
|
| 2008 |
+
"epoch": 1.83206106870229,
|
| 2009 |
+
"grad_norm": 22.26904532535015,
|
| 2010 |
+
"learning_rate": 1.0502621921127774e-08,
|
| 2011 |
+
"logits/chosen": 2.3922975063323975,
|
| 2012 |
+
"logits/rejected": 5.1035685539245605,
|
| 2013 |
+
"logps/chosen": -778.9490966796875,
|
| 2014 |
+
"logps/rejected": -1598.516845703125,
|
| 2015 |
+
"loss": 0.1596,
|
| 2016 |
+
"rewards/accuracies": 0.909375011920929,
|
| 2017 |
+
"rewards/chosen": -4.48614501953125,
|
| 2018 |
+
"rewards/margins": 8.004111289978027,
|
| 2019 |
+
"rewards/rejected": -12.490256309509277,
|
| 2020 |
+
"step": 1320
|
| 2021 |
+
},
|
| 2022 |
+
{
|
| 2023 |
+
"epoch": 1.845940319222762,
|
| 2024 |
+
"grad_norm": 20.889996927509948,
|
| 2025 |
+
"learning_rate": 8.83507767035016e-09,
|
| 2026 |
+
"logits/chosen": 2.3073320388793945,
|
| 2027 |
+
"logits/rejected": 4.977648735046387,
|
| 2028 |
+
"logps/chosen": -797.042236328125,
|
| 2029 |
+
"logps/rejected": -1626.1595458984375,
|
| 2030 |
+
"loss": 0.1657,
|
| 2031 |
+
"rewards/accuracies": 0.918749988079071,
|
| 2032 |
+
"rewards/chosen": -4.542098045349121,
|
| 2033 |
+
"rewards/margins": 8.27629280090332,
|
| 2034 |
+
"rewards/rejected": -12.818391799926758,
|
| 2035 |
+
"step": 1330
|
| 2036 |
+
},
|
| 2037 |
+
{
|
| 2038 |
+
"epoch": 1.8598195697432338,
|
| 2039 |
+
"grad_norm": 26.353814591753384,
|
| 2040 |
+
"learning_rate": 7.309237584595007e-09,
|
| 2041 |
+
"logits/chosen": 2.40840220451355,
|
| 2042 |
+
"logits/rejected": 5.275250434875488,
|
| 2043 |
+
"logps/chosen": -784.8826904296875,
|
| 2044 |
+
"logps/rejected": -1593.8211669921875,
|
| 2045 |
+
"loss": 0.1604,
|
| 2046 |
+
"rewards/accuracies": 0.934374988079071,
|
| 2047 |
+
"rewards/chosen": -4.507115364074707,
|
| 2048 |
+
"rewards/margins": 8.047722816467285,
|
| 2049 |
+
"rewards/rejected": -12.554839134216309,
|
| 2050 |
+
"step": 1340
|
| 2051 |
+
},
|
| 2052 |
+
{
|
| 2053 |
+
"epoch": 1.8736988202637057,
|
| 2054 |
+
"grad_norm": 27.69480128946826,
|
| 2055 |
+
"learning_rate": 5.925998220016659e-09,
|
| 2056 |
+
"logits/chosen": 2.0922293663024902,
|
| 2057 |
+
"logits/rejected": 5.0452680587768555,
|
| 2058 |
+
"logps/chosen": -795.8992919921875,
|
| 2059 |
+
"logps/rejected": -1614.81298828125,
|
| 2060 |
+
"loss": 0.1657,
|
| 2061 |
+
"rewards/accuracies": 0.940625011920929,
|
| 2062 |
+
"rewards/chosen": -4.478159427642822,
|
| 2063 |
+
"rewards/margins": 8.272706985473633,
|
| 2064 |
+
"rewards/rejected": -12.75086498260498,
|
| 2065 |
+
"step": 1350
|
| 2066 |
+
},
|
| 2067 |
+
{
|
| 2068 |
+
"epoch": 1.8875780707841776,
|
| 2069 |
+
"grad_norm": 21.698890899471028,
|
| 2070 |
+
"learning_rate": 4.6861723431538265e-09,
|
| 2071 |
+
"logits/chosen": 2.527801275253296,
|
| 2072 |
+
"logits/rejected": 5.509335517883301,
|
| 2073 |
+
"logps/chosen": -762.1702270507812,
|
| 2074 |
+
"logps/rejected": -1630.4300537109375,
|
| 2075 |
+
"loss": 0.16,
|
| 2076 |
+
"rewards/accuracies": 0.9281250238418579,
|
| 2077 |
+
"rewards/chosen": -4.3918232917785645,
|
| 2078 |
+
"rewards/margins": 8.475740432739258,
|
| 2079 |
+
"rewards/rejected": -12.867563247680664,
|
| 2080 |
+
"step": 1360
|
| 2081 |
+
},
|
| 2082 |
+
{
|
| 2083 |
+
"epoch": 1.9014573213046495,
|
| 2084 |
+
"grad_norm": 23.606565024567274,
|
| 2085 |
+
"learning_rate": 3.5904884533627113e-09,
|
| 2086 |
+
"logits/chosen": 2.040306568145752,
|
| 2087 |
+
"logits/rejected": 5.0976457595825195,
|
| 2088 |
+
"logps/chosen": -809.3186645507812,
|
| 2089 |
+
"logps/rejected": -1723.1839599609375,
|
| 2090 |
+
"loss": 0.1661,
|
| 2091 |
+
"rewards/accuracies": 0.956250011920929,
|
| 2092 |
+
"rewards/chosen": -4.630982398986816,
|
| 2093 |
+
"rewards/margins": 9.00139331817627,
|
| 2094 |
+
"rewards/rejected": -13.63237476348877,
|
| 2095 |
+
"step": 1370
|
| 2096 |
+
},
|
| 2097 |
+
{
|
| 2098 |
+
"epoch": 1.9153365718251214,
|
| 2099 |
+
"grad_norm": 23.741237778017844,
|
| 2100 |
+
"learning_rate": 2.639590354763882e-09,
|
| 2101 |
+
"logits/chosen": 2.073582410812378,
|
| 2102 |
+
"logits/rejected": 4.958869934082031,
|
| 2103 |
+
"logps/chosen": -795.91162109375,
|
| 2104 |
+
"logps/rejected": -1612.056396484375,
|
| 2105 |
+
"loss": 0.1613,
|
| 2106 |
+
"rewards/accuracies": 0.940625011920929,
|
| 2107 |
+
"rewards/chosen": -4.596648693084717,
|
| 2108 |
+
"rewards/margins": 8.110963821411133,
|
| 2109 |
+
"rewards/rejected": -12.707611083984375,
|
| 2110 |
+
"step": 1380
|
| 2111 |
+
},
|
| 2112 |
+
{
|
| 2113 |
+
"epoch": 1.9292158223455933,
|
| 2114 |
+
"grad_norm": 18.578608062996363,
|
| 2115 |
+
"learning_rate": 1.8340367779545452e-09,
|
| 2116 |
+
"logits/chosen": 2.297240972518921,
|
| 2117 |
+
"logits/rejected": 5.08568811416626,
|
| 2118 |
+
"logps/chosen": -794.4322509765625,
|
| 2119 |
+
"logps/rejected": -1613.2269287109375,
|
| 2120 |
+
"loss": 0.1486,
|
| 2121 |
+
"rewards/accuracies": 0.9375,
|
| 2122 |
+
"rewards/chosen": -4.632008075714111,
|
| 2123 |
+
"rewards/margins": 8.13344955444336,
|
| 2124 |
+
"rewards/rejected": -12.765457153320312,
|
| 2125 |
+
"step": 1390
|
| 2126 |
+
},
|
| 2127 |
+
{
|
| 2128 |
+
"epoch": 1.9430950728660652,
|
| 2129 |
+
"grad_norm": 18.67065103551978,
|
| 2130 |
+
"learning_rate": 1.1743010517085427e-09,
|
| 2131 |
+
"logits/chosen": 2.2125773429870605,
|
| 2132 |
+
"logits/rejected": 5.183515548706055,
|
| 2133 |
+
"logps/chosen": -828.9417724609375,
|
| 2134 |
+
"logps/rejected": -1649.573974609375,
|
| 2135 |
+
"loss": 0.1548,
|
| 2136 |
+
"rewards/accuracies": 0.9437500238418579,
|
| 2137 |
+
"rewards/chosen": -4.853725433349609,
|
| 2138 |
+
"rewards/margins": 8.179043769836426,
|
| 2139 |
+
"rewards/rejected": -13.032768249511719,
|
| 2140 |
+
"step": 1400
|
| 2141 |
+
},
|
| 2142 |
+
{
|
| 2143 |
+
"epoch": 1.956974323386537,
|
| 2144 |
+
"grad_norm": 22.95032078584552,
|
| 2145 |
+
"learning_rate": 6.607708248569377e-10,
|
| 2146 |
+
"logits/chosen": 2.1667537689208984,
|
| 2147 |
+
"logits/rejected": 4.85316276550293,
|
| 2148 |
+
"logps/chosen": -793.8030395507812,
|
| 2149 |
+
"logps/rejected": -1551.1268310546875,
|
| 2150 |
+
"loss": 0.1672,
|
| 2151 |
+
"rewards/accuracies": 0.918749988079071,
|
| 2152 |
+
"rewards/chosen": -4.6431074142456055,
|
| 2153 |
+
"rewards/margins": 7.463442325592041,
|
| 2154 |
+
"rewards/rejected": -12.106550216674805,
|
| 2155 |
+
"step": 1410
|
| 2156 |
+
},
|
| 2157 |
+
{
|
| 2158 |
+
"epoch": 1.970853573907009,
|
| 2159 |
+
"grad_norm": 18.974426521602602,
|
| 2160 |
+
"learning_rate": 2.9374783851240923e-10,
|
| 2161 |
+
"logits/chosen": 2.2130208015441895,
|
| 2162 |
+
"logits/rejected": 5.2726945877075195,
|
| 2163 |
+
"logps/chosen": -810.5875244140625,
|
| 2164 |
+
"logps/rejected": -1586.91552734375,
|
| 2165 |
+
"loss": 0.1663,
|
| 2166 |
+
"rewards/accuracies": 0.9156249761581421,
|
| 2167 |
+
"rewards/chosen": -4.543966770172119,
|
| 2168 |
+
"rewards/margins": 7.790687561035156,
|
| 2169 |
+
"rewards/rejected": -12.334654808044434,
|
| 2170 |
+
"step": 1420
|
| 2171 |
+
},
|
| 2172 |
+
{
|
| 2173 |
+
"epoch": 1.984732824427481,
|
| 2174 |
+
"grad_norm": 22.46417636362631,
|
| 2175 |
+
"learning_rate": 7.34477487716878e-11,
|
| 2176 |
+
"logits/chosen": 1.7181364297866821,
|
| 2177 |
+
"logits/rejected": 4.7832465171813965,
|
| 2178 |
+
"logps/chosen": -800.3704223632812,
|
| 2179 |
+
"logps/rejected": -1517.8531494140625,
|
| 2180 |
+
"loss": 0.1627,
|
| 2181 |
+
"rewards/accuracies": 0.9375,
|
| 2182 |
+
"rewards/chosen": -4.434643745422363,
|
| 2183 |
+
"rewards/margins": 7.220560550689697,
|
| 2184 |
+
"rewards/rejected": -11.655204772949219,
|
| 2185 |
+
"step": 1430
|
| 2186 |
+
},
|
| 2187 |
+
{
|
| 2188 |
+
"epoch": 1.9986120749479528,
|
| 2189 |
+
"grad_norm": 31.109687143592964,
|
| 2190 |
+
"learning_rate": 0.0,
|
| 2191 |
+
"logits/chosen": 2.243917942047119,
|
| 2192 |
+
"logits/rejected": 5.361481666564941,
|
| 2193 |
+
"logps/chosen": -816.5538940429688,
|
| 2194 |
+
"logps/rejected": -1620.42529296875,
|
| 2195 |
+
"loss": 0.1599,
|
| 2196 |
+
"rewards/accuracies": 0.925000011920929,
|
| 2197 |
+
"rewards/chosen": -4.733948707580566,
|
| 2198 |
+
"rewards/margins": 8.088279724121094,
|
| 2199 |
+
"rewards/rejected": -12.822227478027344,
|
| 2200 |
+
"step": 1440
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"epoch": 1.9986120749479528,
|
| 2204 |
+
"step": 1440,
|
| 2205 |
+
"total_flos": 0.0,
|
| 2206 |
+
"train_loss": 0.25082930790053476,
|
| 2207 |
+
"train_runtime": 42170.4729,
|
| 2208 |
+
"train_samples_per_second": 8.747,
|
| 2209 |
+
"train_steps_per_second": 0.034
|
| 2210 |
+
}
|
| 2211 |
+
],
|
| 2212 |
+
"logging_steps": 10,
|
| 2213 |
+
"max_steps": 1440,
|
| 2214 |
+
"num_input_tokens_seen": 0,
|
| 2215 |
+
"num_train_epochs": 2,
|
| 2216 |
+
"save_steps": 100,
|
| 2217 |
+
"stateful_callbacks": {
|
| 2218 |
+
"TrainerControl": {
|
| 2219 |
+
"args": {
|
| 2220 |
+
"should_epoch_stop": false,
|
| 2221 |
+
"should_evaluate": false,
|
| 2222 |
+
"should_log": false,
|
| 2223 |
+
"should_save": true,
|
| 2224 |
+
"should_training_stop": true
|
| 2225 |
+
},
|
| 2226 |
+
"attributes": {}
|
| 2227 |
+
}
|
| 2228 |
+
},
|
| 2229 |
+
"total_flos": 0.0,
|
| 2230 |
+
"train_batch_size": 8,
|
| 2231 |
+
"trial_name": null,
|
| 2232 |
+
"trial_params": null
|
| 2233 |
+
}
|