Commit
•
44fc118
1
Parent(s):
0b96062
Model save
Browse files- README.md +61 -0
- all_results.json +9 -0
- generation_config.json +7 -0
- train_results.json +9 -0
- trainer_state.json +1377 -0
README.md
ADDED
|
@@ -0,0 +1,61 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
license: apache-2.0
|
| 4 |
+
base_model: TinyLlama/TinyLlama-1.1B-Chat-v1.0
|
| 5 |
+
tags:
|
| 6 |
+
- trl
|
| 7 |
+
- dpo
|
| 8 |
+
- generated_from_trainer
|
| 9 |
+
model-index:
|
| 10 |
+
- name: tinyllama-1.1b-chat-v1.0-ui-math-dpo-2
|
| 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 |
+
# tinyllama-1.1b-chat-v1.0-ui-math-dpo-2
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [TinyLlama/TinyLlama-1.1B-Chat-v1.0](https://huggingface.co/TinyLlama/TinyLlama-1.1B-Chat-v1.0) on an unknown dataset.
|
| 20 |
+
|
| 21 |
+
## Model description
|
| 22 |
+
|
| 23 |
+
More information needed
|
| 24 |
+
|
| 25 |
+
## Intended uses & limitations
|
| 26 |
+
|
| 27 |
+
More information needed
|
| 28 |
+
|
| 29 |
+
## Training and evaluation data
|
| 30 |
+
|
| 31 |
+
More information needed
|
| 32 |
+
|
| 33 |
+
## Training procedure
|
| 34 |
+
|
| 35 |
+
### Training hyperparameters
|
| 36 |
+
|
| 37 |
+
The following hyperparameters were used during training:
|
| 38 |
+
- learning_rate: 5e-07
|
| 39 |
+
- train_batch_size: 8
|
| 40 |
+
- eval_batch_size: 8
|
| 41 |
+
- seed: 42
|
| 42 |
+
- distributed_type: multi-GPU
|
| 43 |
+
- num_devices: 8
|
| 44 |
+
- gradient_accumulation_steps: 4
|
| 45 |
+
- total_train_batch_size: 256
|
| 46 |
+
- total_eval_batch_size: 64
|
| 47 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 48 |
+
- lr_scheduler_type: cosine
|
| 49 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 50 |
+
- num_epochs: 2
|
| 51 |
+
|
| 52 |
+
### Training results
|
| 53 |
+
|
| 54 |
+
|
| 55 |
+
|
| 56 |
+
### Framework versions
|
| 57 |
+
|
| 58 |
+
- Transformers 4.44.1
|
| 59 |
+
- Pytorch 2.1.2+cu121
|
| 60 |
+
- Datasets 2.21.0
|
| 61 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.9977349943374858,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.32467431276022984,
|
| 5 |
+
"train_runtime": 5857.252,
|
| 6 |
+
"train_samples": 112998,
|
| 7 |
+
"train_samples_per_second": 38.584,
|
| 8 |
+
"train_steps_per_second": 0.151
|
| 9 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 1,
|
| 3 |
+
"eos_token_id": 2,
|
| 4 |
+
"max_length": 2048,
|
| 5 |
+
"pad_token_id": 0,
|
| 6 |
+
"transformers_version": "4.44.1"
|
| 7 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.9977349943374858,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.32467431276022984,
|
| 5 |
+
"train_runtime": 5857.252,
|
| 6 |
+
"train_samples": 112998,
|
| 7 |
+
"train_samples_per_second": 38.584,
|
| 8 |
+
"train_steps_per_second": 0.151
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1377 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.9977349943374858,
|
| 5 |
+
"eval_steps": 1000,
|
| 6 |
+
"global_step": 882,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0022650056625141564,
|
| 13 |
+
"grad_norm": 2.6474389252526,
|
| 14 |
+
"learning_rate": 5.617977528089887e-09,
|
| 15 |
+
"logits/chosen": -2.3315391540527344,
|
| 16 |
+
"logits/rejected": -2.3719687461853027,
|
| 17 |
+
"logps/chosen": -323.98822021484375,
|
| 18 |
+
"logps/rejected": -333.302734375,
|
| 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": 0.022650056625141562,
|
| 28 |
+
"grad_norm": 2.546226472747536,
|
| 29 |
+
"learning_rate": 5.617977528089887e-08,
|
| 30 |
+
"logits/chosen": -2.4010610580444336,
|
| 31 |
+
"logits/rejected": -2.3566555976867676,
|
| 32 |
+
"logps/chosen": -317.4289245605469,
|
| 33 |
+
"logps/rejected": -297.19677734375,
|
| 34 |
+
"loss": 0.6933,
|
| 35 |
+
"rewards/accuracies": 0.46875,
|
| 36 |
+
"rewards/chosen": 0.00016070896526798606,
|
| 37 |
+
"rewards/margins": -4.6764034777879715e-07,
|
| 38 |
+
"rewards/rejected": 0.00016117654740810394,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.045300113250283124,
|
| 43 |
+
"grad_norm": 2.575936847764113,
|
| 44 |
+
"learning_rate": 1.1235955056179774e-07,
|
| 45 |
+
"logits/chosen": -2.368352174758911,
|
| 46 |
+
"logits/rejected": -2.383195638656616,
|
| 47 |
+
"logps/chosen": -317.442138671875,
|
| 48 |
+
"logps/rejected": -310.83355712890625,
|
| 49 |
+
"loss": 0.693,
|
| 50 |
+
"rewards/accuracies": 0.5062500238418579,
|
| 51 |
+
"rewards/chosen": 0.0005755432066507638,
|
| 52 |
+
"rewards/margins": 0.0001175893921754323,
|
| 53 |
+
"rewards/rejected": 0.00045795380719937384,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.06795016987542468,
|
| 58 |
+
"grad_norm": 2.9029645784064435,
|
| 59 |
+
"learning_rate": 1.6853932584269663e-07,
|
| 60 |
+
"logits/chosen": -2.4023149013519287,
|
| 61 |
+
"logits/rejected": -2.3384761810302734,
|
| 62 |
+
"logps/chosen": -324.33111572265625,
|
| 63 |
+
"logps/rejected": -293.6396179199219,
|
| 64 |
+
"loss": 0.6924,
|
| 65 |
+
"rewards/accuracies": 0.59375,
|
| 66 |
+
"rewards/chosen": 0.0024513269308954477,
|
| 67 |
+
"rewards/margins": 0.001740488805808127,
|
| 68 |
+
"rewards/rejected": 0.000710838008671999,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.09060022650056625,
|
| 73 |
+
"grad_norm": 2.7631508582024225,
|
| 74 |
+
"learning_rate": 2.2471910112359549e-07,
|
| 75 |
+
"logits/chosen": -2.366269588470459,
|
| 76 |
+
"logits/rejected": -2.384061813354492,
|
| 77 |
+
"logps/chosen": -307.44244384765625,
|
| 78 |
+
"logps/rejected": -305.51422119140625,
|
| 79 |
+
"loss": 0.6905,
|
| 80 |
+
"rewards/accuracies": 0.737500011920929,
|
| 81 |
+
"rewards/chosen": 0.006844083778560162,
|
| 82 |
+
"rewards/margins": 0.005638611502945423,
|
| 83 |
+
"rewards/rejected": 0.0012054723920300603,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.11325028312570781,
|
| 88 |
+
"grad_norm": 2.4659439994932892,
|
| 89 |
+
"learning_rate": 2.8089887640449437e-07,
|
| 90 |
+
"logits/chosen": -2.337625741958618,
|
| 91 |
+
"logits/rejected": -2.3457820415496826,
|
| 92 |
+
"logps/chosen": -317.6352844238281,
|
| 93 |
+
"logps/rejected": -304.02239990234375,
|
| 94 |
+
"loss": 0.6859,
|
| 95 |
+
"rewards/accuracies": 0.793749988079071,
|
| 96 |
+
"rewards/chosen": 0.02052396535873413,
|
| 97 |
+
"rewards/margins": 0.015680011361837387,
|
| 98 |
+
"rewards/rejected": 0.004843952599912882,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.13590033975084936,
|
| 103 |
+
"grad_norm": 2.3504525836204753,
|
| 104 |
+
"learning_rate": 3.3707865168539325e-07,
|
| 105 |
+
"logits/chosen": -2.337041139602661,
|
| 106 |
+
"logits/rejected": -2.3470675945281982,
|
| 107 |
+
"logps/chosen": -307.0657653808594,
|
| 108 |
+
"logps/rejected": -308.83709716796875,
|
| 109 |
+
"loss": 0.6779,
|
| 110 |
+
"rewards/accuracies": 0.7906249761581421,
|
| 111 |
+
"rewards/chosen": 0.03320229798555374,
|
| 112 |
+
"rewards/margins": 0.029773065820336342,
|
| 113 |
+
"rewards/rejected": 0.003429233329370618,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.15855039637599094,
|
| 118 |
+
"grad_norm": 2.4436159820736068,
|
| 119 |
+
"learning_rate": 3.9325842696629214e-07,
|
| 120 |
+
"logits/chosen": -2.3073842525482178,
|
| 121 |
+
"logits/rejected": -2.2655694484710693,
|
| 122 |
+
"logps/chosen": -322.8226623535156,
|
| 123 |
+
"logps/rejected": -297.44964599609375,
|
| 124 |
+
"loss": 0.6641,
|
| 125 |
+
"rewards/accuracies": 0.793749988079071,
|
| 126 |
+
"rewards/chosen": 0.052168797701597214,
|
| 127 |
+
"rewards/margins": 0.05918798968195915,
|
| 128 |
+
"rewards/rejected": -0.007019191049039364,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.1812004530011325,
|
| 133 |
+
"grad_norm": 2.6664317328149236,
|
| 134 |
+
"learning_rate": 4.4943820224719097e-07,
|
| 135 |
+
"logits/chosen": -2.3002970218658447,
|
| 136 |
+
"logits/rejected": -2.2402286529541016,
|
| 137 |
+
"logps/chosen": -320.1070861816406,
|
| 138 |
+
"logps/rejected": -302.884033203125,
|
| 139 |
+
"loss": 0.6465,
|
| 140 |
+
"rewards/accuracies": 0.831250011920929,
|
| 141 |
+
"rewards/chosen": 0.03679288178682327,
|
| 142 |
+
"rewards/margins": 0.10391112416982651,
|
| 143 |
+
"rewards/rejected": -0.06711824238300323,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.20385050962627407,
|
| 148 |
+
"grad_norm": 3.2572528737706152,
|
| 149 |
+
"learning_rate": 4.999980381634755e-07,
|
| 150 |
+
"logits/chosen": -2.1139278411865234,
|
| 151 |
+
"logits/rejected": -2.1478025913238525,
|
| 152 |
+
"logps/chosen": -333.51519775390625,
|
| 153 |
+
"logps/rejected": -344.88470458984375,
|
| 154 |
+
"loss": 0.6106,
|
| 155 |
+
"rewards/accuracies": 0.831250011920929,
|
| 156 |
+
"rewards/chosen": -0.061306070536375046,
|
| 157 |
+
"rewards/margins": 0.20201241970062256,
|
| 158 |
+
"rewards/rejected": -0.2633184492588043,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.22650056625141562,
|
| 163 |
+
"grad_norm": 3.441455714530276,
|
| 164 |
+
"learning_rate": 4.99762655034665e-07,
|
| 165 |
+
"logits/chosen": -1.9943259954452515,
|
| 166 |
+
"logits/rejected": -2.025644302368164,
|
| 167 |
+
"logps/chosen": -347.1192321777344,
|
| 168 |
+
"logps/rejected": -368.75213623046875,
|
| 169 |
+
"loss": 0.5747,
|
| 170 |
+
"rewards/accuracies": 0.815625011920929,
|
| 171 |
+
"rewards/chosen": -0.25914034247398376,
|
| 172 |
+
"rewards/margins": 0.2930234670639038,
|
| 173 |
+
"rewards/rejected": -0.5521637797355652,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.2491506228765572,
|
| 178 |
+
"grad_norm": 3.756828982195148,
|
| 179 |
+
"learning_rate": 4.991353278599861e-07,
|
| 180 |
+
"logits/chosen": -1.9040400981903076,
|
| 181 |
+
"logits/rejected": -1.8980512619018555,
|
| 182 |
+
"logps/chosen": -368.0120544433594,
|
| 183 |
+
"logps/rejected": -400.9908752441406,
|
| 184 |
+
"loss": 0.5297,
|
| 185 |
+
"rewards/accuracies": 0.828125,
|
| 186 |
+
"rewards/chosen": -0.6502432227134705,
|
| 187 |
+
"rewards/margins": 0.4519161283969879,
|
| 188 |
+
"rewards/rejected": -1.1021593809127808,
|
| 189 |
+
"step": 110
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.2718006795016987,
|
| 193 |
+
"grad_norm": 4.624437428257384,
|
| 194 |
+
"learning_rate": 4.981170410826531e-07,
|
| 195 |
+
"logits/chosen": -1.8735542297363281,
|
| 196 |
+
"logits/rejected": -1.83624267578125,
|
| 197 |
+
"logps/chosen": -419.53680419921875,
|
| 198 |
+
"logps/rejected": -448.97528076171875,
|
| 199 |
+
"loss": 0.4874,
|
| 200 |
+
"rewards/accuracies": 0.793749988079071,
|
| 201 |
+
"rewards/chosen": -0.9317169189453125,
|
| 202 |
+
"rewards/margins": 0.6421880722045898,
|
| 203 |
+
"rewards/rejected": -1.5739049911499023,
|
| 204 |
+
"step": 120
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.2944507361268403,
|
| 208 |
+
"grad_norm": 4.949083580442793,
|
| 209 |
+
"learning_rate": 4.967093926654972e-07,
|
| 210 |
+
"logits/chosen": -1.8730270862579346,
|
| 211 |
+
"logits/rejected": -1.8644899129867554,
|
| 212 |
+
"logps/chosen": -421.25946044921875,
|
| 213 |
+
"logps/rejected": -489.58526611328125,
|
| 214 |
+
"loss": 0.4491,
|
| 215 |
+
"rewards/accuracies": 0.824999988079071,
|
| 216 |
+
"rewards/chosen": -0.9310008883476257,
|
| 217 |
+
"rewards/margins": 0.8616712689399719,
|
| 218 |
+
"rewards/rejected": -1.7926719188690186,
|
| 219 |
+
"step": 130
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.3171007927519819,
|
| 223 |
+
"grad_norm": 4.996031451766477,
|
| 224 |
+
"learning_rate": 4.949145915833377e-07,
|
| 225 |
+
"logits/chosen": -1.8482372760772705,
|
| 226 |
+
"logits/rejected": -1.809913992881775,
|
| 227 |
+
"logps/chosen": -437.0384826660156,
|
| 228 |
+
"logps/rejected": -518.9461669921875,
|
| 229 |
+
"loss": 0.4203,
|
| 230 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 231 |
+
"rewards/chosen": -1.2464931011199951,
|
| 232 |
+
"rewards/margins": 1.060370922088623,
|
| 233 |
+
"rewards/rejected": -2.306863784790039,
|
| 234 |
+
"step": 140
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.33975084937712347,
|
| 238 |
+
"grad_norm": 5.659599672667922,
|
| 239 |
+
"learning_rate": 4.92735454356513e-07,
|
| 240 |
+
"logits/chosen": -1.7960946559906006,
|
| 241 |
+
"logits/rejected": -1.747685194015503,
|
| 242 |
+
"logps/chosen": -468.9212341308594,
|
| 243 |
+
"logps/rejected": -572.489013671875,
|
| 244 |
+
"loss": 0.3953,
|
| 245 |
+
"rewards/accuracies": 0.831250011920929,
|
| 246 |
+
"rewards/chosen": -1.327953577041626,
|
| 247 |
+
"rewards/margins": 1.3009440898895264,
|
| 248 |
+
"rewards/rejected": -2.6288976669311523,
|
| 249 |
+
"step": 150
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.362400906002265,
|
| 253 |
+
"grad_norm": 6.071105217564194,
|
| 254 |
+
"learning_rate": 4.901754006310109e-07,
|
| 255 |
+
"logits/chosen": -1.8912861347198486,
|
| 256 |
+
"logits/rejected": -1.8290300369262695,
|
| 257 |
+
"logps/chosen": -469.5503845214844,
|
| 258 |
+
"logps/rejected": -571.6123657226562,
|
| 259 |
+
"loss": 0.3822,
|
| 260 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 261 |
+
"rewards/chosen": -1.3224334716796875,
|
| 262 |
+
"rewards/margins": 1.3983814716339111,
|
| 263 |
+
"rewards/rejected": -2.7208149433135986,
|
| 264 |
+
"step": 160
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.38505096262740657,
|
| 268 |
+
"grad_norm": 5.706084632255313,
|
| 269 |
+
"learning_rate": 4.872384478121341e-07,
|
| 270 |
+
"logits/chosen": -1.7244150638580322,
|
| 271 |
+
"logits/rejected": -1.7288404703140259,
|
| 272 |
+
"logps/chosen": -460.2561950683594,
|
| 273 |
+
"logps/rejected": -626.7166748046875,
|
| 274 |
+
"loss": 0.3686,
|
| 275 |
+
"rewards/accuracies": 0.8531249761581421,
|
| 276 |
+
"rewards/chosen": -1.572575569152832,
|
| 277 |
+
"rewards/margins": 1.626971960067749,
|
| 278 |
+
"rewards/rejected": -3.199547290802002,
|
| 279 |
+
"step": 170
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.40770101925254815,
|
| 283 |
+
"grad_norm": 6.231205543249148,
|
| 284 |
+
"learning_rate": 4.839292047601234e-07,
|
| 285 |
+
"logits/chosen": -1.714738130569458,
|
| 286 |
+
"logits/rejected": -1.655225157737732,
|
| 287 |
+
"logps/chosen": -453.4337463378906,
|
| 288 |
+
"logps/rejected": -580.6221923828125,
|
| 289 |
+
"loss": 0.3646,
|
| 290 |
+
"rewards/accuracies": 0.8656250238418579,
|
| 291 |
+
"rewards/chosen": -1.3668302297592163,
|
| 292 |
+
"rewards/margins": 1.5435349941253662,
|
| 293 |
+
"rewards/rejected": -2.910365343093872,
|
| 294 |
+
"step": 180
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.43035107587768967,
|
| 298 |
+
"grad_norm": 7.7415354302251185,
|
| 299 |
+
"learning_rate": 4.80252864557629e-07,
|
| 300 |
+
"logits/chosen": -1.5334300994873047,
|
| 301 |
+
"logits/rejected": -1.5589441061019897,
|
| 302 |
+
"logps/chosen": -473.0502014160156,
|
| 303 |
+
"logps/rejected": -634.1920776367188,
|
| 304 |
+
"loss": 0.3501,
|
| 305 |
+
"rewards/accuracies": 0.8218749761581421,
|
| 306 |
+
"rewards/chosen": -1.5327621698379517,
|
| 307 |
+
"rewards/margins": 1.8059412240982056,
|
| 308 |
+
"rewards/rejected": -3.338703155517578,
|
| 309 |
+
"step": 190
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.45300113250283125,
|
| 313 |
+
"grad_norm": 6.247718164321941,
|
| 314 |
+
"learning_rate": 4.7621519636038314e-07,
|
| 315 |
+
"logits/chosen": -1.6528962850570679,
|
| 316 |
+
"logits/rejected": -1.607163429260254,
|
| 317 |
+
"logps/chosen": -468.12823486328125,
|
| 318 |
+
"logps/rejected": -652.6515502929688,
|
| 319 |
+
"loss": 0.3528,
|
| 320 |
+
"rewards/accuracies": 0.871874988079071,
|
| 321 |
+
"rewards/chosen": -1.4182040691375732,
|
| 322 |
+
"rewards/margins": 2.0675435066223145,
|
| 323 |
+
"rewards/rejected": -3.4857475757598877,
|
| 324 |
+
"step": 200
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.47565118912797283,
|
| 328 |
+
"grad_norm": 5.866194755643268,
|
| 329 |
+
"learning_rate": 4.7182253634385947e-07,
|
| 330 |
+
"logits/chosen": -1.8014767169952393,
|
| 331 |
+
"logits/rejected": -1.7056528329849243,
|
| 332 |
+
"logps/chosen": -465.86004638671875,
|
| 333 |
+
"logps/rejected": -619.522705078125,
|
| 334 |
+
"loss": 0.3197,
|
| 335 |
+
"rewards/accuracies": 0.875,
|
| 336 |
+
"rewards/chosen": -1.308288335800171,
|
| 337 |
+
"rewards/margins": 1.8610426187515259,
|
| 338 |
+
"rewards/rejected": -3.1693310737609863,
|
| 339 |
+
"step": 210
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.4983012457531144,
|
| 343 |
+
"grad_norm": 8.540677017846228,
|
| 344 |
+
"learning_rate": 4.670817777601288e-07,
|
| 345 |
+
"logits/chosen": -1.6279010772705078,
|
| 346 |
+
"logits/rejected": -1.611676573753357,
|
| 347 |
+
"logps/chosen": -486.3113708496094,
|
| 348 |
+
"logps/rejected": -701.3948974609375,
|
| 349 |
+
"loss": 0.331,
|
| 350 |
+
"rewards/accuracies": 0.8843749761581421,
|
| 351 |
+
"rewards/chosen": -1.751829743385315,
|
| 352 |
+
"rewards/margins": 2.2083218097686768,
|
| 353 |
+
"rewards/rejected": -3.960151195526123,
|
| 354 |
+
"step": 220
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.5209513023782559,
|
| 358 |
+
"grad_norm": 7.180290608469894,
|
| 359 |
+
"learning_rate": 4.620003601205124e-07,
|
| 360 |
+
"logits/chosen": -1.618285894393921,
|
| 361 |
+
"logits/rejected": -1.5827783346176147,
|
| 362 |
+
"logps/chosen": -495.9786682128906,
|
| 363 |
+
"logps/rejected": -671.6422729492188,
|
| 364 |
+
"loss": 0.3079,
|
| 365 |
+
"rewards/accuracies": 0.840624988079071,
|
| 366 |
+
"rewards/chosen": -1.722399353981018,
|
| 367 |
+
"rewards/margins": 1.888711929321289,
|
| 368 |
+
"rewards/rejected": -3.6111114025115967,
|
| 369 |
+
"step": 230
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.5436013590033975,
|
| 373 |
+
"grad_norm": 7.812256996678949,
|
| 374 |
+
"learning_rate": 4.565862575210102e-07,
|
| 375 |
+
"logits/chosen": -1.5925363302230835,
|
| 376 |
+
"logits/rejected": -1.5041205883026123,
|
| 377 |
+
"logps/chosen": -467.281494140625,
|
| 378 |
+
"logps/rejected": -654.7891845703125,
|
| 379 |
+
"loss": 0.3183,
|
| 380 |
+
"rewards/accuracies": 0.846875011920929,
|
| 381 |
+
"rewards/chosen": -1.521933913230896,
|
| 382 |
+
"rewards/margins": 2.0333094596862793,
|
| 383 |
+
"rewards/rejected": -3.555243730545044,
|
| 384 |
+
"step": 240
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.5662514156285391,
|
| 388 |
+
"grad_norm": 7.655324105498245,
|
| 389 |
+
"learning_rate": 4.5084796612882256e-07,
|
| 390 |
+
"logits/chosen": -1.6674625873565674,
|
| 391 |
+
"logits/rejected": -1.6272201538085938,
|
| 392 |
+
"logps/chosen": -447.31683349609375,
|
| 393 |
+
"logps/rejected": -636.6031494140625,
|
| 394 |
+
"loss": 0.3193,
|
| 395 |
+
"rewards/accuracies": 0.8531249761581421,
|
| 396 |
+
"rewards/chosen": -1.515677809715271,
|
| 397 |
+
"rewards/margins": 1.9939563274383545,
|
| 398 |
+
"rewards/rejected": -3.509634494781494,
|
| 399 |
+
"step": 250
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.5889014722536806,
|
| 403 |
+
"grad_norm": 11.00110804956485,
|
| 404 |
+
"learning_rate": 4.447944908496042e-07,
|
| 405 |
+
"logits/chosen": -1.5731537342071533,
|
| 406 |
+
"logits/rejected": -1.5266633033752441,
|
| 407 |
+
"logps/chosen": -487.8212890625,
|
| 408 |
+
"logps/rejected": -684.2847900390625,
|
| 409 |
+
"loss": 0.294,
|
| 410 |
+
"rewards/accuracies": 0.8656250238418579,
|
| 411 |
+
"rewards/chosen": -1.6535365581512451,
|
| 412 |
+
"rewards/margins": 2.1172165870666504,
|
| 413 |
+
"rewards/rejected": -3.7707531452178955,
|
| 414 |
+
"step": 260
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.6115515288788222,
|
| 418 |
+
"grad_norm": 8.565085184192531,
|
| 419 |
+
"learning_rate": 4.3843533119637125e-07,
|
| 420 |
+
"logits/chosen": -1.6604524850845337,
|
| 421 |
+
"logits/rejected": -1.6133596897125244,
|
| 422 |
+
"logps/chosen": -491.6846618652344,
|
| 423 |
+
"logps/rejected": -732.2810668945312,
|
| 424 |
+
"loss": 0.2903,
|
| 425 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 426 |
+
"rewards/chosen": -1.610835075378418,
|
| 427 |
+
"rewards/margins": 2.487947940826416,
|
| 428 |
+
"rewards/rejected": -4.098783493041992,
|
| 429 |
+
"step": 270
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.6342015855039638,
|
| 433 |
+
"grad_norm": 9.092182327634628,
|
| 434 |
+
"learning_rate": 4.317804663822391e-07,
|
| 435 |
+
"logits/chosen": -1.5771735906600952,
|
| 436 |
+
"logits/rejected": -1.4461592435836792,
|
| 437 |
+
"logps/chosen": -494.58538818359375,
|
| 438 |
+
"logps/rejected": -696.6702880859375,
|
| 439 |
+
"loss": 0.2894,
|
| 440 |
+
"rewards/accuracies": 0.878125011920929,
|
| 441 |
+
"rewards/chosen": -1.75591242313385,
|
| 442 |
+
"rewards/margins": 2.3270115852355957,
|
| 443 |
+
"rewards/rejected": -4.082923412322998,
|
| 444 |
+
"step": 280
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.6568516421291053,
|
| 448 |
+
"grad_norm": 6.8364437103089495,
|
| 449 |
+
"learning_rate": 4.248403396603817e-07,
|
| 450 |
+
"logits/chosen": -1.662105917930603,
|
| 451 |
+
"logits/rejected": -1.6381728649139404,
|
| 452 |
+
"logps/chosen": -465.8875427246094,
|
| 453 |
+
"logps/rejected": -674.3609619140625,
|
| 454 |
+
"loss": 0.2922,
|
| 455 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 456 |
+
"rewards/chosen": -1.528600811958313,
|
| 457 |
+
"rewards/margins": 2.1931700706481934,
|
| 458 |
+
"rewards/rejected": -3.7217705249786377,
|
| 459 |
+
"step": 290
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.6795016987542469,
|
| 463 |
+
"grad_norm": 9.441509895706943,
|
| 464 |
+
"learning_rate": 4.176258419357899e-07,
|
| 465 |
+
"logits/chosen": -1.6376762390136719,
|
| 466 |
+
"logits/rejected": -1.5463725328445435,
|
| 467 |
+
"logps/chosen": -469.1526794433594,
|
| 468 |
+
"logps/rejected": -699.0280151367188,
|
| 469 |
+
"loss": 0.2994,
|
| 470 |
+
"rewards/accuracies": 0.8843749761581421,
|
| 471 |
+
"rewards/chosen": -1.5148029327392578,
|
| 472 |
+
"rewards/margins": 2.2971949577331543,
|
| 473 |
+
"rewards/rejected": -3.811997890472412,
|
| 474 |
+
"step": 300
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.7021517553793885,
|
| 478 |
+
"grad_norm": 9.112495750796707,
|
| 479 |
+
"learning_rate": 4.101482946745438e-07,
|
| 480 |
+
"logits/chosen": -1.5513279438018799,
|
| 481 |
+
"logits/rejected": -1.4326368570327759,
|
| 482 |
+
"logps/chosen": -492.14154052734375,
|
| 483 |
+
"logps/rejected": -711.12255859375,
|
| 484 |
+
"loss": 0.2852,
|
| 485 |
+
"rewards/accuracies": 0.862500011920929,
|
| 486 |
+
"rewards/chosen": -1.719791054725647,
|
| 487 |
+
"rewards/margins": 2.3883697986602783,
|
| 488 |
+
"rewards/rejected": -4.108160972595215,
|
| 489 |
+
"step": 310
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.72480181200453,
|
| 493 |
+
"grad_norm": 8.205308841823761,
|
| 494 |
+
"learning_rate": 4.024194321374212e-07,
|
| 495 |
+
"logits/chosen": -1.5063291788101196,
|
| 496 |
+
"logits/rejected": -1.4347339868545532,
|
| 497 |
+
"logps/chosen": -481.82080078125,
|
| 498 |
+
"logps/rejected": -725.50830078125,
|
| 499 |
+
"loss": 0.2888,
|
| 500 |
+
"rewards/accuracies": 0.875,
|
| 501 |
+
"rewards/chosen": -1.73996901512146,
|
| 502 |
+
"rewards/margins": 2.5435125827789307,
|
| 503 |
+
"rewards/rejected": -4.283481597900391,
|
| 504 |
+
"step": 320
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 0.7474518686296716,
|
| 508 |
+
"grad_norm": 8.205344156872085,
|
| 509 |
+
"learning_rate": 3.9445138296572113e-07,
|
| 510 |
+
"logits/chosen": -1.5704634189605713,
|
| 511 |
+
"logits/rejected": -1.4724892377853394,
|
| 512 |
+
"logps/chosen": -482.04742431640625,
|
| 513 |
+
"logps/rejected": -713.767578125,
|
| 514 |
+
"loss": 0.2815,
|
| 515 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 516 |
+
"rewards/chosen": -1.6026813983917236,
|
| 517 |
+
"rewards/margins": 2.5262672901153564,
|
| 518 |
+
"rewards/rejected": -4.128949165344238,
|
| 519 |
+
"step": 330
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.7701019252548131,
|
| 523 |
+
"grad_norm": 8.687044357645386,
|
| 524 |
+
"learning_rate": 3.8625665114819873e-07,
|
| 525 |
+
"logits/chosen": -1.6331688165664673,
|
| 526 |
+
"logits/rejected": -1.5526864528656006,
|
| 527 |
+
"logps/chosen": -502.4422912597656,
|
| 528 |
+
"logps/rejected": -728.0780029296875,
|
| 529 |
+
"loss": 0.2967,
|
| 530 |
+
"rewards/accuracies": 0.871874988079071,
|
| 531 |
+
"rewards/chosen": -1.7365747690200806,
|
| 532 |
+
"rewards/margins": 2.4770658016204834,
|
| 533 |
+
"rewards/rejected": -4.2136406898498535,
|
| 534 |
+
"step": 340
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.7927519818799547,
|
| 538 |
+
"grad_norm": 8.427795252063401,
|
| 539 |
+
"learning_rate": 3.778480963989811e-07,
|
| 540 |
+
"logits/chosen": -1.598920464515686,
|
| 541 |
+
"logits/rejected": -1.5631158351898193,
|
| 542 |
+
"logps/chosen": -502.58648681640625,
|
| 543 |
+
"logps/rejected": -721.0816650390625,
|
| 544 |
+
"loss": 0.2818,
|
| 545 |
+
"rewards/accuracies": 0.875,
|
| 546 |
+
"rewards/chosen": -1.7856614589691162,
|
| 547 |
+
"rewards/margins": 2.304830551147461,
|
| 548 |
+
"rewards/rejected": -4.090491771697998,
|
| 549 |
+
"step": 350
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.8154020385050963,
|
| 553 |
+
"grad_norm": 7.260269705482,
|
| 554 |
+
"learning_rate": 3.6923891397725475e-07,
|
| 555 |
+
"logits/chosen": -1.5732126235961914,
|
| 556 |
+
"logits/rejected": -1.4579116106033325,
|
| 557 |
+
"logps/chosen": -505.892822265625,
|
| 558 |
+
"logps/rejected": -708.0283813476562,
|
| 559 |
+
"loss": 0.273,
|
| 560 |
+
"rewards/accuracies": 0.893750011920929,
|
| 561 |
+
"rewards/chosen": -1.6417592763900757,
|
| 562 |
+
"rewards/margins": 2.4231035709381104,
|
| 563 |
+
"rewards/rejected": -4.0648627281188965,
|
| 564 |
+
"step": 360
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.8380520951302378,
|
| 568 |
+
"grad_norm": 10.479677653279909,
|
| 569 |
+
"learning_rate": 3.604426139803941e-07,
|
| 570 |
+
"logits/chosen": -1.439273715019226,
|
| 571 |
+
"logits/rejected": -1.3620045185089111,
|
| 572 |
+
"logps/chosen": -517.6815795898438,
|
| 573 |
+
"logps/rejected": -789.8757934570312,
|
| 574 |
+
"loss": 0.2764,
|
| 575 |
+
"rewards/accuracies": 0.840624988079071,
|
| 576 |
+
"rewards/chosen": -1.919734239578247,
|
| 577 |
+
"rewards/margins": 2.797135591506958,
|
| 578 |
+
"rewards/rejected": -4.716870307922363,
|
| 579 |
+
"step": 370
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.8607021517553793,
|
| 583 |
+
"grad_norm": 7.985186234740201,
|
| 584 |
+
"learning_rate": 3.5147300014302457e-07,
|
| 585 |
+
"logits/chosen": -1.559147596359253,
|
| 586 |
+
"logits/rejected": -1.436856746673584,
|
| 587 |
+
"logps/chosen": -494.31500244140625,
|
| 588 |
+
"logps/rejected": -769.4177856445312,
|
| 589 |
+
"loss": 0.2741,
|
| 590 |
+
"rewards/accuracies": 0.903124988079071,
|
| 591 |
+
"rewards/chosen": -1.6577142477035522,
|
| 592 |
+
"rewards/margins": 2.9566261768341064,
|
| 593 |
+
"rewards/rejected": -4.614340305328369,
|
| 594 |
+
"step": 380
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 0.883352208380521,
|
| 598 |
+
"grad_norm": 9.216176794426861,
|
| 599 |
+
"learning_rate": 3.423441481752911e-07,
|
| 600 |
+
"logits/chosen": -1.5555119514465332,
|
| 601 |
+
"logits/rejected": -1.4373959302902222,
|
| 602 |
+
"logps/chosen": -485.8768615722656,
|
| 603 |
+
"logps/rejected": -722.7447509765625,
|
| 604 |
+
"loss": 0.2552,
|
| 605 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 606 |
+
"rewards/chosen": -1.6954419612884521,
|
| 607 |
+
"rewards/margins": 2.5930514335632324,
|
| 608 |
+
"rewards/rejected": -4.2884931564331055,
|
| 609 |
+
"step": 390
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.9060022650056625,
|
| 613 |
+
"grad_norm": 11.159180818421824,
|
| 614 |
+
"learning_rate": 3.3307038367432443e-07,
|
| 615 |
+
"logits/chosen": -1.3249311447143555,
|
| 616 |
+
"logits/rejected": -1.2017321586608887,
|
| 617 |
+
"logps/chosen": -486.0079040527344,
|
| 618 |
+
"logps/rejected": -734.5474853515625,
|
| 619 |
+
"loss": 0.2917,
|
| 620 |
+
"rewards/accuracies": 0.890625,
|
| 621 |
+
"rewards/chosen": -1.7877933979034424,
|
| 622 |
+
"rewards/margins": 2.6582465171813965,
|
| 623 |
+
"rewards/rejected": -4.446040153503418,
|
| 624 |
+
"step": 400
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 0.928652321630804,
|
| 628 |
+
"grad_norm": 9.06036199110745,
|
| 629 |
+
"learning_rate": 3.2366625964356904e-07,
|
| 630 |
+
"logits/chosen": -1.3760263919830322,
|
| 631 |
+
"logits/rejected": -1.157128095626831,
|
| 632 |
+
"logps/chosen": -521.535888671875,
|
| 633 |
+
"logps/rejected": -756.6954345703125,
|
| 634 |
+
"loss": 0.2745,
|
| 635 |
+
"rewards/accuracies": 0.9156249761581421,
|
| 636 |
+
"rewards/chosen": -1.8339954614639282,
|
| 637 |
+
"rewards/margins": 2.690295696258545,
|
| 638 |
+
"rewards/rejected": -4.524291038513184,
|
| 639 |
+
"step": 410
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.9513023782559457,
|
| 643 |
+
"grad_norm": 8.613445152071886,
|
| 644 |
+
"learning_rate": 3.1414653365524824e-07,
|
| 645 |
+
"logits/chosen": -1.3222942352294922,
|
| 646 |
+
"logits/rejected": -1.1949504613876343,
|
| 647 |
+
"logps/chosen": -496.029296875,
|
| 648 |
+
"logps/rejected": -751.8590087890625,
|
| 649 |
+
"loss": 0.268,
|
| 650 |
+
"rewards/accuracies": 0.903124988079071,
|
| 651 |
+
"rewards/chosen": -1.8118371963500977,
|
| 652 |
+
"rewards/margins": 2.6409120559692383,
|
| 653 |
+
"rewards/rejected": -4.452749252319336,
|
| 654 |
+
"step": 420
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.9739524348810872,
|
| 658 |
+
"grad_norm": 10.126507731476897,
|
| 659 |
+
"learning_rate": 3.0452614469180916e-07,
|
| 660 |
+
"logits/chosen": -1.3294384479522705,
|
| 661 |
+
"logits/rejected": -1.1579396724700928,
|
| 662 |
+
"logps/chosen": -505.3882751464844,
|
| 663 |
+
"logps/rejected": -744.6370849609375,
|
| 664 |
+
"loss": 0.2817,
|
| 665 |
+
"rewards/accuracies": 0.8687499761581421,
|
| 666 |
+
"rewards/chosen": -1.895179033279419,
|
| 667 |
+
"rewards/margins": 2.613166332244873,
|
| 668 |
+
"rewards/rejected": -4.508345603942871,
|
| 669 |
+
"step": 430
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.9966024915062288,
|
| 673 |
+
"grad_norm": 8.359546887824873,
|
| 674 |
+
"learning_rate": 2.948201897026839e-07,
|
| 675 |
+
"logits/chosen": -1.2789385318756104,
|
| 676 |
+
"logits/rejected": -1.1186952590942383,
|
| 677 |
+
"logps/chosen": -518.8313598632812,
|
| 678 |
+
"logps/rejected": -757.8375244140625,
|
| 679 |
+
"loss": 0.2713,
|
| 680 |
+
"rewards/accuracies": 0.8687499761581421,
|
| 681 |
+
"rewards/chosen": -1.9324840307235718,
|
| 682 |
+
"rewards/margins": 2.687189817428589,
|
| 683 |
+
"rewards/rejected": -4.619673728942871,
|
| 684 |
+
"step": 440
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 1.0192525481313703,
|
| 688 |
+
"grad_norm": 9.916277053083935,
|
| 689 |
+
"learning_rate": 2.8504389991316105e-07,
|
| 690 |
+
"logits/chosen": -1.2456687688827515,
|
| 691 |
+
"logits/rejected": -1.116817831993103,
|
| 692 |
+
"logps/chosen": -517.7054443359375,
|
| 693 |
+
"logps/rejected": -768.4178466796875,
|
| 694 |
+
"loss": 0.2485,
|
| 695 |
+
"rewards/accuracies": 0.903124988079071,
|
| 696 |
+
"rewards/chosen": -2.0004186630249023,
|
| 697 |
+
"rewards/margins": 2.688814640045166,
|
| 698 |
+
"rewards/rejected": -4.689233303070068,
|
| 699 |
+
"step": 450
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 1.0419026047565119,
|
| 703 |
+
"grad_norm": 10.473177616855008,
|
| 704 |
+
"learning_rate": 2.7521261692254064e-07,
|
| 705 |
+
"logits/chosen": -1.201494574546814,
|
| 706 |
+
"logits/rejected": -1.1607218980789185,
|
| 707 |
+
"logps/chosen": -502.1883850097656,
|
| 708 |
+
"logps/rejected": -778.163818359375,
|
| 709 |
+
"loss": 0.2687,
|
| 710 |
+
"rewards/accuracies": 0.856249988079071,
|
| 711 |
+
"rewards/chosen": -1.8305113315582275,
|
| 712 |
+
"rewards/margins": 2.7049150466918945,
|
| 713 |
+
"rewards/rejected": -4.535426139831543,
|
| 714 |
+
"step": 460
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.0645526613816534,
|
| 718 |
+
"grad_norm": 8.161060651228674,
|
| 719 |
+
"learning_rate": 2.653417686290843e-07,
|
| 720 |
+
"logits/chosen": -1.3370718955993652,
|
| 721 |
+
"logits/rejected": -1.1073302030563354,
|
| 722 |
+
"logps/chosen": -504.89801025390625,
|
| 723 |
+
"logps/rejected": -745.27099609375,
|
| 724 |
+
"loss": 0.2668,
|
| 725 |
+
"rewards/accuracies": 0.8968750238418579,
|
| 726 |
+
"rewards/chosen": -1.8848193883895874,
|
| 727 |
+
"rewards/margins": 2.6601107120513916,
|
| 728 |
+
"rewards/rejected": -4.544930458068848,
|
| 729 |
+
"step": 470
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 1.087202718006795,
|
| 733 |
+
"grad_norm": 9.08329196731907,
|
| 734 |
+
"learning_rate": 2.5544684501953874e-07,
|
| 735 |
+
"logits/chosen": -1.3237013816833496,
|
| 736 |
+
"logits/rejected": -1.1105015277862549,
|
| 737 |
+
"logps/chosen": -522.7344970703125,
|
| 738 |
+
"logps/rejected": -795.3836669921875,
|
| 739 |
+
"loss": 0.2582,
|
| 740 |
+
"rewards/accuracies": 0.90625,
|
| 741 |
+
"rewards/chosen": -1.9063637256622314,
|
| 742 |
+
"rewards/margins": 2.994534492492676,
|
| 743 |
+
"rewards/rejected": -4.900897979736328,
|
| 744 |
+
"step": 480
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 1.1098527746319367,
|
| 748 |
+
"grad_norm": 9.239751540023942,
|
| 749 |
+
"learning_rate": 2.455433738612257e-07,
|
| 750 |
+
"logits/chosen": -1.2477200031280518,
|
| 751 |
+
"logits/rejected": -1.140298843383789,
|
| 752 |
+
"logps/chosen": -510.219970703125,
|
| 753 |
+
"logps/rejected": -790.1013793945312,
|
| 754 |
+
"loss": 0.2627,
|
| 755 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 756 |
+
"rewards/chosen": -1.9776588678359985,
|
| 757 |
+
"rewards/margins": 2.872147798538208,
|
| 758 |
+
"rewards/rejected": -4.849806785583496,
|
| 759 |
+
"step": 490
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 1.1325028312570782,
|
| 763 |
+
"grad_norm": 8.063198151261274,
|
| 764 |
+
"learning_rate": 2.356468963348451e-07,
|
| 765 |
+
"logits/chosen": -1.2390304803848267,
|
| 766 |
+
"logits/rejected": -1.0948108434677124,
|
| 767 |
+
"logps/chosen": -526.1553955078125,
|
| 768 |
+
"logps/rejected": -783.7991943359375,
|
| 769 |
+
"loss": 0.2513,
|
| 770 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 771 |
+
"rewards/chosen": -1.924363136291504,
|
| 772 |
+
"rewards/margins": 2.824939250946045,
|
| 773 |
+
"rewards/rejected": -4.749302864074707,
|
| 774 |
+
"step": 500
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 1.1551528878822197,
|
| 778 |
+
"grad_norm": 10.628914294455955,
|
| 779 |
+
"learning_rate": 2.257729426462276e-07,
|
| 780 |
+
"logits/chosen": -1.2929015159606934,
|
| 781 |
+
"logits/rejected": -1.1904685497283936,
|
| 782 |
+
"logps/chosen": -500.63934326171875,
|
| 783 |
+
"logps/rejected": -745.3140869140625,
|
| 784 |
+
"loss": 0.2745,
|
| 785 |
+
"rewards/accuracies": 0.878125011920929,
|
| 786 |
+
"rewards/chosen": -1.9067703485488892,
|
| 787 |
+
"rewards/margins": 2.5259783267974854,
|
| 788 |
+
"rewards/rejected": -4.432748317718506,
|
| 789 |
+
"step": 510
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 1.1778029445073612,
|
| 793 |
+
"grad_norm": 8.572156157412216,
|
| 794 |
+
"learning_rate": 2.1593700765531089e-07,
|
| 795 |
+
"logits/chosen": -1.3477250337600708,
|
| 796 |
+
"logits/rejected": -1.147517442703247,
|
| 797 |
+
"logps/chosen": -497.2181091308594,
|
| 798 |
+
"logps/rejected": -794.3870849609375,
|
| 799 |
+
"loss": 0.2556,
|
| 800 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 801 |
+
"rewards/chosen": -1.7573410272598267,
|
| 802 |
+
"rewards/margins": 3.1540889739990234,
|
| 803 |
+
"rewards/rejected": -4.9114298820495605,
|
| 804 |
+
"step": 520
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 1.2004530011325028,
|
| 808 |
+
"grad_norm": 9.357116476254854,
|
| 809 |
+
"learning_rate": 2.0615452656058264e-07,
|
| 810 |
+
"logits/chosen": -1.1733572483062744,
|
| 811 |
+
"logits/rejected": -1.077141523361206,
|
| 812 |
+
"logps/chosen": -533.3096923828125,
|
| 813 |
+
"logps/rejected": -809.8570556640625,
|
| 814 |
+
"loss": 0.2487,
|
| 815 |
+
"rewards/accuracies": 0.8968750238418579,
|
| 816 |
+
"rewards/chosen": -2.1453044414520264,
|
| 817 |
+
"rewards/margins": 2.8495564460754395,
|
| 818 |
+
"rewards/rejected": -4.994861602783203,
|
| 819 |
+
"step": 530
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 1.2231030577576445,
|
| 823 |
+
"grad_norm": 10.169756727358006,
|
| 824 |
+
"learning_rate": 1.964408506771477e-07,
|
| 825 |
+
"logits/chosen": -1.330486536026001,
|
| 826 |
+
"logits/rejected": -1.1011441946029663,
|
| 827 |
+
"logps/chosen": -514.3831787109375,
|
| 828 |
+
"logps/rejected": -743.8990478515625,
|
| 829 |
+
"loss": 0.2611,
|
| 830 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 831 |
+
"rewards/chosen": -1.9130659103393555,
|
| 832 |
+
"rewards/margins": 2.6177687644958496,
|
| 833 |
+
"rewards/rejected": -4.530834674835205,
|
| 834 |
+
"step": 540
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 1.245753114382786,
|
| 838 |
+
"grad_norm": 10.039033828119221,
|
| 839 |
+
"learning_rate": 1.868112233464317e-07,
|
| 840 |
+
"logits/chosen": -1.2120153903961182,
|
| 841 |
+
"logits/rejected": -1.0327503681182861,
|
| 842 |
+
"logps/chosen": -521.9749755859375,
|
| 843 |
+
"logps/rejected": -802.6920166015625,
|
| 844 |
+
"loss": 0.2457,
|
| 845 |
+
"rewards/accuracies": 0.903124988079071,
|
| 846 |
+
"rewards/chosen": -1.9761244058609009,
|
| 847 |
+
"rewards/margins": 2.9549593925476074,
|
| 848 |
+
"rewards/rejected": -4.931083679199219,
|
| 849 |
+
"step": 550
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 1.2684031710079275,
|
| 853 |
+
"grad_norm": 10.38214991301508,
|
| 854 |
+
"learning_rate": 1.7728075601532256e-07,
|
| 855 |
+
"logits/chosen": -1.1885329484939575,
|
| 856 |
+
"logits/rejected": -0.9879345893859863,
|
| 857 |
+
"logps/chosen": -522.0006103515625,
|
| 858 |
+
"logps/rejected": -811.2237548828125,
|
| 859 |
+
"loss": 0.24,
|
| 860 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 861 |
+
"rewards/chosen": -2.0510880947113037,
|
| 862 |
+
"rewards/margins": 2.9444658756256104,
|
| 863 |
+
"rewards/rejected": -4.995553970336914,
|
| 864 |
+
"step": 560
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 1.291053227633069,
|
| 868 |
+
"grad_norm": 8.356041939902939,
|
| 869 |
+
"learning_rate": 1.678644045222913e-07,
|
| 870 |
+
"logits/chosen": -1.2759056091308594,
|
| 871 |
+
"logits/rejected": -1.119461178779602,
|
| 872 |
+
"logps/chosen": -509.73583984375,
|
| 873 |
+
"logps/rejected": -767.29443359375,
|
| 874 |
+
"loss": 0.262,
|
| 875 |
+
"rewards/accuracies": 0.887499988079071,
|
| 876 |
+
"rewards/chosen": -1.9916547536849976,
|
| 877 |
+
"rewards/margins": 2.671016216278076,
|
| 878 |
+
"rewards/rejected": -4.662671089172363,
|
| 879 |
+
"step": 570
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 1.3137032842582106,
|
| 883 |
+
"grad_norm": 8.912049656879141,
|
| 884 |
+
"learning_rate": 1.585769456277027e-07,
|
| 885 |
+
"logits/chosen": -1.2154964208602905,
|
| 886 |
+
"logits/rejected": -1.0624061822891235,
|
| 887 |
+
"logps/chosen": -511.436767578125,
|
| 888 |
+
"logps/rejected": -811.2041015625,
|
| 889 |
+
"loss": 0.2244,
|
| 890 |
+
"rewards/accuracies": 0.918749988079071,
|
| 891 |
+
"rewards/chosen": -1.986825704574585,
|
| 892 |
+
"rewards/margins": 3.094810962677002,
|
| 893 |
+
"rewards/rejected": -5.081636905670166,
|
| 894 |
+
"step": 580
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 1.3363533408833521,
|
| 898 |
+
"grad_norm": 9.754438029355793,
|
| 899 |
+
"learning_rate": 1.4943295382514728e-07,
|
| 900 |
+
"logits/chosen": -1.162213921546936,
|
| 901 |
+
"logits/rejected": -1.0503453016281128,
|
| 902 |
+
"logps/chosen": -510.4007263183594,
|
| 903 |
+
"logps/rejected": -791.3707275390625,
|
| 904 |
+
"loss": 0.2591,
|
| 905 |
+
"rewards/accuracies": 0.8843749761581421,
|
| 906 |
+
"rewards/chosen": -1.8774373531341553,
|
| 907 |
+
"rewards/margins": 2.9028286933898926,
|
| 908 |
+
"rewards/rejected": -4.780265808105469,
|
| 909 |
+
"step": 590
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 1.3590033975084936,
|
| 913 |
+
"grad_norm": 10.049386128170934,
|
| 914 |
+
"learning_rate": 1.404467784701844e-07,
|
| 915 |
+
"logits/chosen": -1.2329612970352173,
|
| 916 |
+
"logits/rejected": -1.0405888557434082,
|
| 917 |
+
"logps/chosen": -513.1143798828125,
|
| 918 |
+
"logps/rejected": -783.7535400390625,
|
| 919 |
+
"loss": 0.2439,
|
| 920 |
+
"rewards/accuracies": 0.921875,
|
| 921 |
+
"rewards/chosen": -1.8017826080322266,
|
| 922 |
+
"rewards/margins": 3.078281879425049,
|
| 923 |
+
"rewards/rejected": -4.880064487457275,
|
| 924 |
+
"step": 600
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 1.3816534541336354,
|
| 928 |
+
"grad_norm": 10.351517688722195,
|
| 929 |
+
"learning_rate": 1.3163252126238522e-07,
|
| 930 |
+
"logits/chosen": -1.2502057552337646,
|
| 931 |
+
"logits/rejected": -1.0325191020965576,
|
| 932 |
+
"logps/chosen": -521.5572509765625,
|
| 933 |
+
"logps/rejected": -792.7130737304688,
|
| 934 |
+
"loss": 0.2364,
|
| 935 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 936 |
+
"rewards/chosen": -2.0930302143096924,
|
| 937 |
+
"rewards/margins": 2.9488086700439453,
|
| 938 |
+
"rewards/rejected": -5.041838645935059,
|
| 939 |
+
"step": 610
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 1.404303510758777,
|
| 943 |
+
"grad_norm": 10.201760293762261,
|
| 944 |
+
"learning_rate": 1.2300401411601636e-07,
|
| 945 |
+
"logits/chosen": -1.2752196788787842,
|
| 946 |
+
"logits/rejected": -1.0628340244293213,
|
| 947 |
+
"logps/chosen": -522.7730712890625,
|
| 948 |
+
"logps/rejected": -773.3001708984375,
|
| 949 |
+
"loss": 0.2389,
|
| 950 |
+
"rewards/accuracies": 0.921875,
|
| 951 |
+
"rewards/chosen": -1.9185714721679688,
|
| 952 |
+
"rewards/margins": 2.8828213214874268,
|
| 953 |
+
"rewards/rejected": -4.801392555236816,
|
| 954 |
+
"step": 620
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 1.4269535673839184,
|
| 958 |
+
"grad_norm": 8.835921463685338,
|
| 959 |
+
"learning_rate": 1.145747974540856e-07,
|
| 960 |
+
"logits/chosen": -1.2582758665084839,
|
| 961 |
+
"logits/rejected": -1.0210275650024414,
|
| 962 |
+
"logps/chosen": -551.0906982421875,
|
| 963 |
+
"logps/rejected": -785.31982421875,
|
| 964 |
+
"loss": 0.2562,
|
| 965 |
+
"rewards/accuracies": 0.8687499761581421,
|
| 966 |
+
"rewards/chosen": -2.2444424629211426,
|
| 967 |
+
"rewards/margins": 2.7261741161346436,
|
| 968 |
+
"rewards/rejected": -4.970616817474365,
|
| 969 |
+
"step": 630
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 1.44960362400906,
|
| 973 |
+
"grad_norm": 12.991915284453638,
|
| 974 |
+
"learning_rate": 1.0635809895981699e-07,
|
| 975 |
+
"logits/chosen": -1.2509222030639648,
|
| 976 |
+
"logits/rejected": -1.0769033432006836,
|
| 977 |
+
"logps/chosen": -552.90087890625,
|
| 978 |
+
"logps/rejected": -806.5215454101562,
|
| 979 |
+
"loss": 0.2494,
|
| 980 |
+
"rewards/accuracies": 0.887499988079071,
|
| 981 |
+
"rewards/chosen": -2.2703638076782227,
|
| 982 |
+
"rewards/margins": 2.754547119140625,
|
| 983 |
+
"rewards/rejected": -5.024910926818848,
|
| 984 |
+
"step": 640
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 1.4722536806342017,
|
| 988 |
+
"grad_norm": 10.504984052416424,
|
| 989 |
+
"learning_rate": 9.836681281889678e-08,
|
| 990 |
+
"logits/chosen": -1.34755277633667,
|
| 991 |
+
"logits/rejected": -1.1464807987213135,
|
| 992 |
+
"logps/chosen": -515.4022216796875,
|
| 993 |
+
"logps/rejected": -784.6029052734375,
|
| 994 |
+
"loss": 0.2408,
|
| 995 |
+
"rewards/accuracies": 0.918749988079071,
|
| 996 |
+
"rewards/chosen": -1.8913652896881104,
|
| 997 |
+
"rewards/margins": 2.9281973838806152,
|
| 998 |
+
"rewards/rejected": -4.819562911987305,
|
| 999 |
+
"step": 650
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 1.4949037372593432,
|
| 1003 |
+
"grad_norm": 10.125922089796893,
|
| 1004 |
+
"learning_rate": 9.061347948506629e-08,
|
| 1005 |
+
"logits/chosen": -1.2215931415557861,
|
| 1006 |
+
"logits/rejected": -1.1074914932250977,
|
| 1007 |
+
"logps/chosen": -523.8232421875,
|
| 1008 |
+
"logps/rejected": -797.5764770507812,
|
| 1009 |
+
"loss": 0.2312,
|
| 1010 |
+
"rewards/accuracies": 0.878125011920929,
|
| 1011 |
+
"rewards/chosen": -1.9787216186523438,
|
| 1012 |
+
"rewards/margins": 2.8579554557800293,
|
| 1013 |
+
"rewards/rejected": -4.836677551269531,
|
| 1014 |
+
"step": 660
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 1.5175537938844847,
|
| 1018 |
+
"grad_norm": 13.328368593276167,
|
| 1019 |
+
"learning_rate": 8.311026600081394e-08,
|
| 1020 |
+
"logits/chosen": -1.11947762966156,
|
| 1021 |
+
"logits/rejected": -0.9734872579574585,
|
| 1022 |
+
"logps/chosen": -534.2103271484375,
|
| 1023 |
+
"logps/rejected": -830.9747924804688,
|
| 1024 |
+
"loss": 0.2318,
|
| 1025 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1026 |
+
"rewards/chosen": -2.1272778511047363,
|
| 1027 |
+
"rewards/margins": 3.069934844970703,
|
| 1028 |
+
"rewards/rejected": -5.1972126960754395,
|
| 1029 |
+
"step": 670
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 1.5402038505096263,
|
| 1033 |
+
"grad_norm": 9.729514763537226,
|
| 1034 |
+
"learning_rate": 7.586894690404894e-08,
|
| 1035 |
+
"logits/chosen": -1.217233657836914,
|
| 1036 |
+
"logits/rejected": -0.9491454362869263,
|
| 1037 |
+
"logps/chosen": -547.2406005859375,
|
| 1038 |
+
"logps/rejected": -827.4415283203125,
|
| 1039 |
+
"loss": 0.2277,
|
| 1040 |
+
"rewards/accuracies": 0.921875,
|
| 1041 |
+
"rewards/chosen": -2.2959611415863037,
|
| 1042 |
+
"rewards/margins": 3.175316095352173,
|
| 1043 |
+
"rewards/rejected": -5.471277713775635,
|
| 1044 |
+
"step": 680
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 1.5628539071347678,
|
| 1048 |
+
"grad_norm": 9.576592993059183,
|
| 1049 |
+
"learning_rate": 6.890088575071954e-08,
|
| 1050 |
+
"logits/chosen": -1.187366247177124,
|
| 1051 |
+
"logits/rejected": -1.0213922262191772,
|
| 1052 |
+
"logps/chosen": -545.3994750976562,
|
| 1053 |
+
"logps/rejected": -804.3179321289062,
|
| 1054 |
+
"loss": 0.2423,
|
| 1055 |
+
"rewards/accuracies": 0.8843749761581421,
|
| 1056 |
+
"rewards/chosen": -2.1425087451934814,
|
| 1057 |
+
"rewards/margins": 2.924528121948242,
|
| 1058 |
+
"rewards/rejected": -5.067037105560303,
|
| 1059 |
+
"step": 690
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 1.5855039637599093,
|
| 1063 |
+
"grad_norm": 12.126314956409692,
|
| 1064 |
+
"learning_rate": 6.221701728237008e-08,
|
| 1065 |
+
"logits/chosen": -1.1499079465866089,
|
| 1066 |
+
"logits/rejected": -0.9196650385856628,
|
| 1067 |
+
"logps/chosen": -524.6936645507812,
|
| 1068 |
+
"logps/rejected": -810.4720458984375,
|
| 1069 |
+
"loss": 0.248,
|
| 1070 |
+
"rewards/accuracies": 0.8843749761581421,
|
| 1071 |
+
"rewards/chosen": -2.0638883113861084,
|
| 1072 |
+
"rewards/margins": 3.0613222122192383,
|
| 1073 |
+
"rewards/rejected": -5.125210762023926,
|
| 1074 |
+
"step": 700
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 1.6081540203850508,
|
| 1078 |
+
"grad_norm": 11.134418963577295,
|
| 1079 |
+
"learning_rate": 5.582783026662349e-08,
|
| 1080 |
+
"logits/chosen": -1.0319141149520874,
|
| 1081 |
+
"logits/rejected": -0.956072986125946,
|
| 1082 |
+
"logps/chosen": -526.4638671875,
|
| 1083 |
+
"logps/rejected": -814.6434326171875,
|
| 1084 |
+
"loss": 0.26,
|
| 1085 |
+
"rewards/accuracies": 0.875,
|
| 1086 |
+
"rewards/chosen": -2.0331358909606934,
|
| 1087 |
+
"rewards/margins": 2.952976703643799,
|
| 1088 |
+
"rewards/rejected": -4.986112117767334,
|
| 1089 |
+
"step": 710
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 1.6308040770101924,
|
| 1093 |
+
"grad_norm": 9.315625565902986,
|
| 1094 |
+
"learning_rate": 4.9743351037514534e-08,
|
| 1095 |
+
"logits/chosen": -1.129042148590088,
|
| 1096 |
+
"logits/rejected": -0.8932101130485535,
|
| 1097 |
+
"logps/chosen": -548.4468994140625,
|
| 1098 |
+
"logps/rejected": -836.2429809570312,
|
| 1099 |
+
"loss": 0.2368,
|
| 1100 |
+
"rewards/accuracies": 0.890625,
|
| 1101 |
+
"rewards/chosen": -2.1513266563415527,
|
| 1102 |
+
"rewards/margins": 3.1787352561950684,
|
| 1103 |
+
"rewards/rejected": -5.330061912536621,
|
| 1104 |
+
"step": 720
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 1.6534541336353341,
|
| 1108 |
+
"grad_norm": 9.92739535657428,
|
| 1109 |
+
"learning_rate": 4.397312776150439e-08,
|
| 1110 |
+
"logits/chosen": -1.048117995262146,
|
| 1111 |
+
"logits/rejected": -0.967863917350769,
|
| 1112 |
+
"logps/chosen": -528.8873901367188,
|
| 1113 |
+
"logps/rejected": -818.2623291015625,
|
| 1114 |
+
"loss": 0.2489,
|
| 1115 |
+
"rewards/accuracies": 0.875,
|
| 1116 |
+
"rewards/chosen": -2.2309556007385254,
|
| 1117 |
+
"rewards/margins": 2.885885715484619,
|
| 1118 |
+
"rewards/rejected": -5.116840839385986,
|
| 1119 |
+
"step": 730
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 1.6761041902604756,
|
| 1123 |
+
"grad_norm": 11.55426000236745,
|
| 1124 |
+
"learning_rate": 3.852621545386761e-08,
|
| 1125 |
+
"logits/chosen": -1.1159693002700806,
|
| 1126 |
+
"logits/rejected": -0.8810169100761414,
|
| 1127 |
+
"logps/chosen": -534.227294921875,
|
| 1128 |
+
"logps/rejected": -810.526611328125,
|
| 1129 |
+
"loss": 0.2553,
|
| 1130 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 1131 |
+
"rewards/chosen": -2.1146936416625977,
|
| 1132 |
+
"rewards/margins": 3.039085626602173,
|
| 1133 |
+
"rewards/rejected": -5.153779029846191,
|
| 1134 |
+
"step": 740
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 1.6987542468856174,
|
| 1138 |
+
"grad_norm": 11.882365137452437,
|
| 1139 |
+
"learning_rate": 3.341116176896447e-08,
|
| 1140 |
+
"logits/chosen": -1.1606030464172363,
|
| 1141 |
+
"logits/rejected": -0.969256579875946,
|
| 1142 |
+
"logps/chosen": -527.7787475585938,
|
| 1143 |
+
"logps/rejected": -783.6763916015625,
|
| 1144 |
+
"loss": 0.2362,
|
| 1145 |
+
"rewards/accuracies": 0.903124988079071,
|
| 1146 |
+
"rewards/chosen": -2.098135471343994,
|
| 1147 |
+
"rewards/margins": 2.7795679569244385,
|
| 1148 |
+
"rewards/rejected": -4.877703666687012,
|
| 1149 |
+
"step": 750
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 1.721404303510759,
|
| 1153 |
+
"grad_norm": 11.932846657725108,
|
| 1154 |
+
"learning_rate": 2.863599358669755e-08,
|
| 1155 |
+
"logits/chosen": -1.1670676469802856,
|
| 1156 |
+
"logits/rejected": -0.9956063032150269,
|
| 1157 |
+
"logps/chosen": -504.7257385253906,
|
| 1158 |
+
"logps/rejected": -799.7178955078125,
|
| 1159 |
+
"loss": 0.2366,
|
| 1160 |
+
"rewards/accuracies": 0.918749988079071,
|
| 1161 |
+
"rewards/chosen": -2.0211801528930664,
|
| 1162 |
+
"rewards/margins": 3.071319818496704,
|
| 1163 |
+
"rewards/rejected": -5.09250020980835,
|
| 1164 |
+
"step": 760
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 1.7440543601359004,
|
| 1168 |
+
"grad_norm": 11.068137379022545,
|
| 1169 |
+
"learning_rate": 2.4208204416201665e-08,
|
| 1170 |
+
"logits/chosen": -1.1682803630828857,
|
| 1171 |
+
"logits/rejected": -1.0027186870574951,
|
| 1172 |
+
"logps/chosen": -526.065673828125,
|
| 1173 |
+
"logps/rejected": -813.3223876953125,
|
| 1174 |
+
"loss": 0.239,
|
| 1175 |
+
"rewards/accuracies": 0.909375011920929,
|
| 1176 |
+
"rewards/chosen": -2.0436623096466064,
|
| 1177 |
+
"rewards/margins": 3.140903949737549,
|
| 1178 |
+
"rewards/rejected": -5.184566020965576,
|
| 1179 |
+
"step": 770
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 1.766704416761042,
|
| 1183 |
+
"grad_norm": 12.478180187301291,
|
| 1184 |
+
"learning_rate": 2.0134742636534945e-08,
|
| 1185 |
+
"logits/chosen": -1.0638221502304077,
|
| 1186 |
+
"logits/rejected": -0.9631050229072571,
|
| 1187 |
+
"logps/chosen": -504.27154541015625,
|
| 1188 |
+
"logps/rejected": -821.7930908203125,
|
| 1189 |
+
"loss": 0.2399,
|
| 1190 |
+
"rewards/accuracies": 0.9281250238418579,
|
| 1191 |
+
"rewards/chosen": -1.9072799682617188,
|
| 1192 |
+
"rewards/margins": 3.2128396034240723,
|
| 1193 |
+
"rewards/rejected": -5.120119571685791,
|
| 1194 |
+
"step": 780
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 1.7893544733861835,
|
| 1198 |
+
"grad_norm": 10.797469509942093,
|
| 1199 |
+
"learning_rate": 1.6422000592823642e-08,
|
| 1200 |
+
"logits/chosen": -1.1970932483673096,
|
| 1201 |
+
"logits/rejected": -1.005079984664917,
|
| 1202 |
+
"logps/chosen": -508.31494140625,
|
| 1203 |
+
"logps/rejected": -812.4329223632812,
|
| 1204 |
+
"loss": 0.2332,
|
| 1205 |
+
"rewards/accuracies": 0.940625011920929,
|
| 1206 |
+
"rewards/chosen": -1.9276374578475952,
|
| 1207 |
+
"rewards/margins": 3.216580867767334,
|
| 1208 |
+
"rewards/rejected": -5.144218444824219,
|
| 1209 |
+
"step": 790
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 1.812004530011325,
|
| 1213 |
+
"grad_norm": 9.201913655606754,
|
| 1214 |
+
"learning_rate": 1.3075804564971981e-08,
|
| 1215 |
+
"logits/chosen": -1.165398120880127,
|
| 1216 |
+
"logits/rejected": -0.946398913860321,
|
| 1217 |
+
"logps/chosen": -515.598876953125,
|
| 1218 |
+
"logps/rejected": -806.0333251953125,
|
| 1219 |
+
"loss": 0.2595,
|
| 1220 |
+
"rewards/accuracies": 0.90625,
|
| 1221 |
+
"rewards/chosen": -2.098996639251709,
|
| 1222 |
+
"rewards/margins": 3.0318830013275146,
|
| 1223 |
+
"rewards/rejected": -5.130879878997803,
|
| 1224 |
+
"step": 800
|
| 1225 |
+
},
|
| 1226 |
+
{
|
| 1227 |
+
"epoch": 1.8346545866364665,
|
| 1228 |
+
"grad_norm": 12.171909373979952,
|
| 1229 |
+
"learning_rate": 1.01014056246794e-08,
|
| 1230 |
+
"logits/chosen": -1.14596426486969,
|
| 1231 |
+
"logits/rejected": -0.9247859716415405,
|
| 1232 |
+
"logps/chosen": -521.8255615234375,
|
| 1233 |
+
"logps/rejected": -803.5557861328125,
|
| 1234 |
+
"loss": 0.2352,
|
| 1235 |
+
"rewards/accuracies": 0.918749988079071,
|
| 1236 |
+
"rewards/chosen": -2.031855344772339,
|
| 1237 |
+
"rewards/margins": 3.0747625827789307,
|
| 1238 |
+
"rewards/rejected": -5.1066179275512695,
|
| 1239 |
+
"step": 810
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 1.857304643261608,
|
| 1243 |
+
"grad_norm": 10.898179869470376,
|
| 1244 |
+
"learning_rate": 7.503471395112065e-09,
|
| 1245 |
+
"logits/chosen": -1.1780989170074463,
|
| 1246 |
+
"logits/rejected": -0.931725025177002,
|
| 1247 |
+
"logps/chosen": -521.3743896484375,
|
| 1248 |
+
"logps/rejected": -795.4444580078125,
|
| 1249 |
+
"loss": 0.2545,
|
| 1250 |
+
"rewards/accuracies": 0.8968750238418579,
|
| 1251 |
+
"rewards/chosen": -2.1411118507385254,
|
| 1252 |
+
"rewards/margins": 2.875601291656494,
|
| 1253 |
+
"rewards/rejected": -5.0167131423950195,
|
| 1254 |
+
"step": 820
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 1.8799546998867496,
|
| 1258 |
+
"grad_norm": 8.213548605406768,
|
| 1259 |
+
"learning_rate": 5.286078726160548e-09,
|
| 1260 |
+
"logits/chosen": -1.0937467813491821,
|
| 1261 |
+
"logits/rejected": -0.9241981506347656,
|
| 1262 |
+
"logps/chosen": -538.4014282226562,
|
| 1263 |
+
"logps/rejected": -784.3555297851562,
|
| 1264 |
+
"loss": 0.2322,
|
| 1265 |
+
"rewards/accuracies": 0.9156249761581421,
|
| 1266 |
+
"rewards/chosen": -2.020725727081299,
|
| 1267 |
+
"rewards/margins": 2.831569194793701,
|
| 1268 |
+
"rewards/rejected": -4.852294921875,
|
| 1269 |
+
"step": 830
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 1.9026047565118913,
|
| 1273 |
+
"grad_norm": 11.652554680737152,
|
| 1274 |
+
"learning_rate": 3.4527072967777404e-09,
|
| 1275 |
+
"logits/chosen": -1.1555739641189575,
|
| 1276 |
+
"logits/rejected": -0.9933372735977173,
|
| 1277 |
+
"logps/chosen": -534.9871215820312,
|
| 1278 |
+
"logps/rejected": -795.6361083984375,
|
| 1279 |
+
"loss": 0.2605,
|
| 1280 |
+
"rewards/accuracies": 0.878125011920929,
|
| 1281 |
+
"rewards/chosen": -2.151045083999634,
|
| 1282 |
+
"rewards/margins": 2.7527434825897217,
|
| 1283 |
+
"rewards/rejected": -4.9037885665893555,
|
| 1284 |
+
"step": 840
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 1.9252548131370328,
|
| 1288 |
+
"grad_norm": 11.308556042176512,
|
| 1289 |
+
"learning_rate": 2.0062341544370143e-09,
|
| 1290 |
+
"logits/chosen": -1.1897794008255005,
|
| 1291 |
+
"logits/rejected": -0.9474241137504578,
|
| 1292 |
+
"logps/chosen": -540.0062255859375,
|
| 1293 |
+
"logps/rejected": -806.772216796875,
|
| 1294 |
+
"loss": 0.2696,
|
| 1295 |
+
"rewards/accuracies": 0.8656250238418579,
|
| 1296 |
+
"rewards/chosen": -2.1642906665802,
|
| 1297 |
+
"rewards/margins": 2.9361448287963867,
|
| 1298 |
+
"rewards/rejected": -5.100435733795166,
|
| 1299 |
+
"step": 850
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 1.9479048697621744,
|
| 1303 |
+
"grad_norm": 10.396927511013917,
|
| 1304 |
+
"learning_rate": 9.489292002793913e-10,
|
| 1305 |
+
"logits/chosen": -1.1064279079437256,
|
| 1306 |
+
"logits/rejected": -0.8940499424934387,
|
| 1307 |
+
"logps/chosen": -526.8607177734375,
|
| 1308 |
+
"logps/rejected": -789.2191772460938,
|
| 1309 |
+
"loss": 0.2429,
|
| 1310 |
+
"rewards/accuracies": 0.887499988079071,
|
| 1311 |
+
"rewards/chosen": -2.0915961265563965,
|
| 1312 |
+
"rewards/margins": 2.8657567501068115,
|
| 1313 |
+
"rewards/rejected": -4.957352638244629,
|
| 1314 |
+
"step": 860
|
| 1315 |
+
},
|
| 1316 |
+
{
|
| 1317 |
+
"epoch": 1.970554926387316,
|
| 1318 |
+
"grad_norm": 8.728874761020926,
|
| 1319 |
+
"learning_rate": 2.824516270348576e-10,
|
| 1320 |
+
"logits/chosen": -1.2010364532470703,
|
| 1321 |
+
"logits/rejected": -1.0010019540786743,
|
| 1322 |
+
"logps/chosen": -522.8780517578125,
|
| 1323 |
+
"logps/rejected": -835.6373901367188,
|
| 1324 |
+
"loss": 0.2245,
|
| 1325 |
+
"rewards/accuracies": 0.9281250238418579,
|
| 1326 |
+
"rewards/chosen": -1.9398930072784424,
|
| 1327 |
+
"rewards/margins": 3.3346550464630127,
|
| 1328 |
+
"rewards/rejected": -5.274548053741455,
|
| 1329 |
+
"step": 870
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 1.9932049830124576,
|
| 1333 |
+
"grad_norm": 8.736000347816738,
|
| 1334 |
+
"learning_rate": 7.847315307535352e-12,
|
| 1335 |
+
"logits/chosen": -1.1201304197311401,
|
| 1336 |
+
"logits/rejected": -0.9259728193283081,
|
| 1337 |
+
"logps/chosen": -526.8204345703125,
|
| 1338 |
+
"logps/rejected": -830.1511840820312,
|
| 1339 |
+
"loss": 0.2423,
|
| 1340 |
+
"rewards/accuracies": 0.9375,
|
| 1341 |
+
"rewards/chosen": -2.0089852809906006,
|
| 1342 |
+
"rewards/margins": 3.2325141429901123,
|
| 1343 |
+
"rewards/rejected": -5.241499423980713,
|
| 1344 |
+
"step": 880
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 1.9977349943374858,
|
| 1348 |
+
"step": 882,
|
| 1349 |
+
"total_flos": 0.0,
|
| 1350 |
+
"train_loss": 0.32467431276022984,
|
| 1351 |
+
"train_runtime": 5857.252,
|
| 1352 |
+
"train_samples_per_second": 38.584,
|
| 1353 |
+
"train_steps_per_second": 0.151
|
| 1354 |
+
}
|
| 1355 |
+
],
|
| 1356 |
+
"logging_steps": 10,
|
| 1357 |
+
"max_steps": 882,
|
| 1358 |
+
"num_input_tokens_seen": 0,
|
| 1359 |
+
"num_train_epochs": 2,
|
| 1360 |
+
"save_steps": 100,
|
| 1361 |
+
"stateful_callbacks": {
|
| 1362 |
+
"TrainerControl": {
|
| 1363 |
+
"args": {
|
| 1364 |
+
"should_epoch_stop": false,
|
| 1365 |
+
"should_evaluate": false,
|
| 1366 |
+
"should_log": false,
|
| 1367 |
+
"should_save": true,
|
| 1368 |
+
"should_training_stop": true
|
| 1369 |
+
},
|
| 1370 |
+
"attributes": {}
|
| 1371 |
+
}
|
| 1372 |
+
},
|
| 1373 |
+
"total_flos": 0.0,
|
| 1374 |
+
"train_batch_size": 8,
|
| 1375 |
+
"trial_name": null,
|
| 1376 |
+
"trial_params": null
|
| 1377 |
+
}
|