Model save
Browse files- README.md +60 -0
- all_results.json +9 -0
- generation_config.json +6 -0
- train_results.json +9 -0
- trainer_state.json +987 -0
README.md
ADDED
|
@@ -0,0 +1,60 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
base_model: allenai/tulu-2-7b
|
| 4 |
+
tags:
|
| 5 |
+
- trl
|
| 6 |
+
- dpo
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
model-index:
|
| 9 |
+
- name: tulu-2-7b-hh-dpo
|
| 10 |
+
results: []
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 14 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 15 |
+
|
| 16 |
+
# tulu-2-7b-hh-dpo
|
| 17 |
+
|
| 18 |
+
This model is a fine-tuned version of [allenai/tulu-2-7b](https://huggingface.co/allenai/tulu-2-7b) on an unknown dataset.
|
| 19 |
+
|
| 20 |
+
## Model description
|
| 21 |
+
|
| 22 |
+
More information needed
|
| 23 |
+
|
| 24 |
+
## Intended uses & limitations
|
| 25 |
+
|
| 26 |
+
More information needed
|
| 27 |
+
|
| 28 |
+
## Training and evaluation data
|
| 29 |
+
|
| 30 |
+
More information needed
|
| 31 |
+
|
| 32 |
+
## Training procedure
|
| 33 |
+
|
| 34 |
+
### Training hyperparameters
|
| 35 |
+
|
| 36 |
+
The following hyperparameters were used during training:
|
| 37 |
+
- learning_rate: 5e-07
|
| 38 |
+
- train_batch_size: 8
|
| 39 |
+
- eval_batch_size: 8
|
| 40 |
+
- seed: 42
|
| 41 |
+
- distributed_type: multi-GPU
|
| 42 |
+
- num_devices: 8
|
| 43 |
+
- gradient_accumulation_steps: 4
|
| 44 |
+
- total_train_batch_size: 256
|
| 45 |
+
- total_eval_batch_size: 64
|
| 46 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 47 |
+
- lr_scheduler_type: cosine
|
| 48 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 49 |
+
- num_epochs: 1
|
| 50 |
+
|
| 51 |
+
### Training results
|
| 52 |
+
|
| 53 |
+
|
| 54 |
+
|
| 55 |
+
### Framework versions
|
| 56 |
+
|
| 57 |
+
- Transformers 4.44.1
|
| 58 |
+
- Pytorch 2.1.2+cu121
|
| 59 |
+
- Datasets 2.21.0
|
| 60 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9996020692399522,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.6662861807331159,
|
| 5 |
+
"train_runtime": 15158.4407,
|
| 6 |
+
"train_samples": 160800,
|
| 7 |
+
"train_samples_per_second": 10.608,
|
| 8 |
+
"train_steps_per_second": 0.041
|
| 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": 0.9996020692399522,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.6662861807331159,
|
| 5 |
+
"train_runtime": 15158.4407,
|
| 6 |
+
"train_samples": 160800,
|
| 7 |
+
"train_samples_per_second": 10.608,
|
| 8 |
+
"train_steps_per_second": 0.041
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,987 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9996020692399522,
|
| 5 |
+
"eval_steps": 1000,
|
| 6 |
+
"global_step": 628,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0015917230401910067,
|
| 13 |
+
"grad_norm": 1.083263489032715,
|
| 14 |
+
"learning_rate": 7.936507936507936e-09,
|
| 15 |
+
"logits/chosen": -1.870174765586853,
|
| 16 |
+
"logits/rejected": -1.8014098405838013,
|
| 17 |
+
"logps/chosen": -158.94805908203125,
|
| 18 |
+
"logps/rejected": -194.90118408203125,
|
| 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": 0.01591723040191007,
|
| 28 |
+
"grad_norm": 1.7486374056371567,
|
| 29 |
+
"learning_rate": 7.936507936507936e-08,
|
| 30 |
+
"logits/chosen": -1.8175890445709229,
|
| 31 |
+
"logits/rejected": -1.8403409719467163,
|
| 32 |
+
"logps/chosen": -188.36404418945312,
|
| 33 |
+
"logps/rejected": -179.35687255859375,
|
| 34 |
+
"loss": 0.6932,
|
| 35 |
+
"rewards/accuracies": 0.3576388955116272,
|
| 36 |
+
"rewards/chosen": -0.0004040856729261577,
|
| 37 |
+
"rewards/margins": -0.0008076421800069511,
|
| 38 |
+
"rewards/rejected": 0.00040355639066547155,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.03183446080382014,
|
| 43 |
+
"grad_norm": 1.2288534081464715,
|
| 44 |
+
"learning_rate": 1.5873015873015872e-07,
|
| 45 |
+
"logits/chosen": -1.9721990823745728,
|
| 46 |
+
"logits/rejected": -1.980015516281128,
|
| 47 |
+
"logps/chosen": -197.92391967773438,
|
| 48 |
+
"logps/rejected": -207.22412109375,
|
| 49 |
+
"loss": 0.6932,
|
| 50 |
+
"rewards/accuracies": 0.5,
|
| 51 |
+
"rewards/chosen": 8.275737855001353e-06,
|
| 52 |
+
"rewards/margins": -0.0001316081325057894,
|
| 53 |
+
"rewards/rejected": 0.00013988386490382254,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.0477516912057302,
|
| 58 |
+
"grad_norm": 0.9705773937715004,
|
| 59 |
+
"learning_rate": 2.3809523809523806e-07,
|
| 60 |
+
"logits/chosen": -1.8532848358154297,
|
| 61 |
+
"logits/rejected": -1.8753681182861328,
|
| 62 |
+
"logps/chosen": -192.031982421875,
|
| 63 |
+
"logps/rejected": -195.5606231689453,
|
| 64 |
+
"loss": 0.6931,
|
| 65 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 66 |
+
"rewards/chosen": 0.0007524025277234614,
|
| 67 |
+
"rewards/margins": 0.000668005901388824,
|
| 68 |
+
"rewards/rejected": 8.439658995484933e-05,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.06366892160764027,
|
| 73 |
+
"grad_norm": 1.3879715707434512,
|
| 74 |
+
"learning_rate": 3.1746031746031743e-07,
|
| 75 |
+
"logits/chosen": -1.9114805459976196,
|
| 76 |
+
"logits/rejected": -1.9526256322860718,
|
| 77 |
+
"logps/chosen": -198.54586791992188,
|
| 78 |
+
"logps/rejected": -190.65158081054688,
|
| 79 |
+
"loss": 0.6931,
|
| 80 |
+
"rewards/accuracies": 0.515625,
|
| 81 |
+
"rewards/chosen": 0.0004263836017344147,
|
| 82 |
+
"rewards/margins": 0.00022551305301021785,
|
| 83 |
+
"rewards/rejected": 0.0002008705196203664,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.07958615200955034,
|
| 88 |
+
"grad_norm": 0.9223665834941889,
|
| 89 |
+
"learning_rate": 3.968253968253968e-07,
|
| 90 |
+
"logits/chosen": -1.9061291217803955,
|
| 91 |
+
"logits/rejected": -1.963629126548767,
|
| 92 |
+
"logps/chosen": -190.56007385253906,
|
| 93 |
+
"logps/rejected": -193.8275146484375,
|
| 94 |
+
"loss": 0.693,
|
| 95 |
+
"rewards/accuracies": 0.512499988079071,
|
| 96 |
+
"rewards/chosen": 0.0014020069502294064,
|
| 97 |
+
"rewards/margins": 0.00029031833400949836,
|
| 98 |
+
"rewards/rejected": 0.001111688674427569,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.0955033824114604,
|
| 103 |
+
"grad_norm": 1.6880319723067283,
|
| 104 |
+
"learning_rate": 4.761904761904761e-07,
|
| 105 |
+
"logits/chosen": -1.9793567657470703,
|
| 106 |
+
"logits/rejected": -2.001311779022217,
|
| 107 |
+
"logps/chosen": -196.06846618652344,
|
| 108 |
+
"logps/rejected": -188.52914428710938,
|
| 109 |
+
"loss": 0.6929,
|
| 110 |
+
"rewards/accuracies": 0.528124988079071,
|
| 111 |
+
"rewards/chosen": 0.00374850956723094,
|
| 112 |
+
"rewards/margins": 0.0002347224799450487,
|
| 113 |
+
"rewards/rejected": 0.003513787407428026,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.11142061281337047,
|
| 118 |
+
"grad_norm": 1.2950341728538266,
|
| 119 |
+
"learning_rate": 4.998106548810311e-07,
|
| 120 |
+
"logits/chosen": -1.9034782648086548,
|
| 121 |
+
"logits/rejected": -1.9826198816299438,
|
| 122 |
+
"logps/chosen": -196.86940002441406,
|
| 123 |
+
"logps/rejected": -177.41644287109375,
|
| 124 |
+
"loss": 0.6926,
|
| 125 |
+
"rewards/accuracies": 0.5718749761581421,
|
| 126 |
+
"rewards/chosen": 0.005933411885052919,
|
| 127 |
+
"rewards/margins": 0.001384641625918448,
|
| 128 |
+
"rewards/rejected": 0.004548769909888506,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.12733784321528055,
|
| 133 |
+
"grad_norm": 2.34883059858935,
|
| 134 |
+
"learning_rate": 4.988839406031596e-07,
|
| 135 |
+
"logits/chosen": -1.9306480884552002,
|
| 136 |
+
"logits/rejected": -1.9404945373535156,
|
| 137 |
+
"logps/chosen": -176.39120483398438,
|
| 138 |
+
"logps/rejected": -193.53347778320312,
|
| 139 |
+
"loss": 0.6922,
|
| 140 |
+
"rewards/accuracies": 0.550000011920929,
|
| 141 |
+
"rewards/chosen": 0.010212738066911697,
|
| 142 |
+
"rewards/margins": 0.0014647066127508879,
|
| 143 |
+
"rewards/rejected": 0.008748031221330166,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.14325507361719061,
|
| 148 |
+
"grad_norm": 1.8215899265671025,
|
| 149 |
+
"learning_rate": 4.971879403278432e-07,
|
| 150 |
+
"logits/chosen": -1.8801124095916748,
|
| 151 |
+
"logits/rejected": -1.9323303699493408,
|
| 152 |
+
"logps/chosen": -189.4450225830078,
|
| 153 |
+
"logps/rejected": -190.84414672851562,
|
| 154 |
+
"loss": 0.6922,
|
| 155 |
+
"rewards/accuracies": 0.543749988079071,
|
| 156 |
+
"rewards/chosen": 0.011725414544343948,
|
| 157 |
+
"rewards/margins": 0.0013827414950355887,
|
| 158 |
+
"rewards/rejected": 0.01034267246723175,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.15917230401910068,
|
| 163 |
+
"grad_norm": 1.208023660425821,
|
| 164 |
+
"learning_rate": 4.947278962947386e-07,
|
| 165 |
+
"logits/chosen": -1.8471660614013672,
|
| 166 |
+
"logits/rejected": -1.8932346105575562,
|
| 167 |
+
"logps/chosen": -193.34718322753906,
|
| 168 |
+
"logps/rejected": -196.87457275390625,
|
| 169 |
+
"loss": 0.6915,
|
| 170 |
+
"rewards/accuracies": 0.581250011920929,
|
| 171 |
+
"rewards/chosen": 0.015522779896855354,
|
| 172 |
+
"rewards/margins": 0.002660988597199321,
|
| 173 |
+
"rewards/rejected": 0.01286179106682539,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.17508953442101075,
|
| 178 |
+
"grad_norm": 1.1987469741900727,
|
| 179 |
+
"learning_rate": 4.915114123589732e-07,
|
| 180 |
+
"logits/chosen": -1.8427928686141968,
|
| 181 |
+
"logits/rejected": -1.9131923913955688,
|
| 182 |
+
"logps/chosen": -177.29983520507812,
|
| 183 |
+
"logps/rejected": -178.62013244628906,
|
| 184 |
+
"loss": 0.6911,
|
| 185 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 186 |
+
"rewards/chosen": 0.019433852285146713,
|
| 187 |
+
"rewards/margins": 0.003971750847995281,
|
| 188 |
+
"rewards/rejected": 0.015462102368474007,
|
| 189 |
+
"step": 110
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.1910067648229208,
|
| 193 |
+
"grad_norm": 1.3513440295821657,
|
| 194 |
+
"learning_rate": 4.875484304880629e-07,
|
| 195 |
+
"logits/chosen": -1.852002501487732,
|
| 196 |
+
"logits/rejected": -1.8977711200714111,
|
| 197 |
+
"logps/chosen": -196.73301696777344,
|
| 198 |
+
"logps/rejected": -198.54873657226562,
|
| 199 |
+
"loss": 0.6905,
|
| 200 |
+
"rewards/accuracies": 0.59375,
|
| 201 |
+
"rewards/chosen": 0.02165614813566208,
|
| 202 |
+
"rewards/margins": 0.006273286882787943,
|
| 203 |
+
"rewards/rejected": 0.015382861718535423,
|
| 204 |
+
"step": 120
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.20692399522483088,
|
| 208 |
+
"grad_norm": 0.919308465194322,
|
| 209 |
+
"learning_rate": 4.828512000318616e-07,
|
| 210 |
+
"logits/chosen": -1.8836482763290405,
|
| 211 |
+
"logits/rejected": -1.9497747421264648,
|
| 212 |
+
"logps/chosen": -197.17617797851562,
|
| 213 |
+
"logps/rejected": -199.19927978515625,
|
| 214 |
+
"loss": 0.6895,
|
| 215 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 216 |
+
"rewards/chosen": 0.026965487748384476,
|
| 217 |
+
"rewards/margins": 0.008914025500416756,
|
| 218 |
+
"rewards/rejected": 0.01805146224796772,
|
| 219 |
+
"step": 130
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.22284122562674094,
|
| 223 |
+
"grad_norm": 1.3141295336399639,
|
| 224 |
+
"learning_rate": 4.774342398605221e-07,
|
| 225 |
+
"logits/chosen": -1.9415794610977173,
|
| 226 |
+
"logits/rejected": -1.9936542510986328,
|
| 227 |
+
"logps/chosen": -191.3373565673828,
|
| 228 |
+
"logps/rejected": -182.86744689941406,
|
| 229 |
+
"loss": 0.6894,
|
| 230 |
+
"rewards/accuracies": 0.565625011920929,
|
| 231 |
+
"rewards/chosen": 0.02900194190442562,
|
| 232 |
+
"rewards/margins": 0.007812701165676117,
|
| 233 |
+
"rewards/rejected": 0.021189238876104355,
|
| 234 |
+
"step": 140
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.238758456028651,
|
| 238 |
+
"grad_norm": 1.0426885097254153,
|
| 239 |
+
"learning_rate": 4.713142934875005e-07,
|
| 240 |
+
"logits/chosen": -1.9385217428207397,
|
| 241 |
+
"logits/rejected": -2.025717258453369,
|
| 242 |
+
"logps/chosen": -201.0162811279297,
|
| 243 |
+
"logps/rejected": -189.43167114257812,
|
| 244 |
+
"loss": 0.6882,
|
| 245 |
+
"rewards/accuracies": 0.606249988079071,
|
| 246 |
+
"rewards/chosen": 0.030758211389183998,
|
| 247 |
+
"rewards/margins": 0.014848137274384499,
|
| 248 |
+
"rewards/rejected": 0.01591007225215435,
|
| 249 |
+
"step": 150
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.2546756864305611,
|
| 253 |
+
"grad_norm": 0.9993599317482367,
|
| 254 |
+
"learning_rate": 4.64510277316316e-07,
|
| 255 |
+
"logits/chosen": -1.883013367652893,
|
| 256 |
+
"logits/rejected": -1.9087369441986084,
|
| 257 |
+
"logps/chosen": -177.09432983398438,
|
| 258 |
+
"logps/rejected": -181.0247802734375,
|
| 259 |
+
"loss": 0.6875,
|
| 260 |
+
"rewards/accuracies": 0.581250011920929,
|
| 261 |
+
"rewards/chosen": 0.019768565893173218,
|
| 262 |
+
"rewards/margins": 0.01374894194304943,
|
| 263 |
+
"rewards/rejected": 0.006019626744091511,
|
| 264 |
+
"step": 160
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.27059291683247116,
|
| 268 |
+
"grad_norm": 6.045830822367488,
|
| 269 |
+
"learning_rate": 4.570432221710314e-07,
|
| 270 |
+
"logits/chosen": -1.8615553379058838,
|
| 271 |
+
"logits/rejected": -1.9155362844467163,
|
| 272 |
+
"logps/chosen": -191.3380126953125,
|
| 273 |
+
"logps/rejected": -196.19644165039062,
|
| 274 |
+
"loss": 0.6866,
|
| 275 |
+
"rewards/accuracies": 0.574999988079071,
|
| 276 |
+
"rewards/chosen": 0.01975223794579506,
|
| 277 |
+
"rewards/margins": 0.01370079256594181,
|
| 278 |
+
"rewards/rejected": 0.006051443982869387,
|
| 279 |
+
"step": 170
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.28651014723438123,
|
| 283 |
+
"grad_norm": 3.6863233560878217,
|
| 284 |
+
"learning_rate": 4.4893620829118124e-07,
|
| 285 |
+
"logits/chosen": -1.8656437397003174,
|
| 286 |
+
"logits/rejected": -1.9271419048309326,
|
| 287 |
+
"logps/chosen": -197.45355224609375,
|
| 288 |
+
"logps/rejected": -187.43008422851562,
|
| 289 |
+
"loss": 0.6861,
|
| 290 |
+
"rewards/accuracies": 0.621874988079071,
|
| 291 |
+
"rewards/chosen": 0.011210992000997066,
|
| 292 |
+
"rewards/margins": 0.020882535725831985,
|
| 293 |
+
"rewards/rejected": -0.009671543724834919,
|
| 294 |
+
"step": 180
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.3024273776362913,
|
| 298 |
+
"grad_norm": 1.0448028144351018,
|
| 299 |
+
"learning_rate": 4.40214293992074e-07,
|
| 300 |
+
"logits/chosen": -1.9013402462005615,
|
| 301 |
+
"logits/rejected": -1.9314658641815186,
|
| 302 |
+
"logps/chosen": -189.4186553955078,
|
| 303 |
+
"logps/rejected": -181.21237182617188,
|
| 304 |
+
"loss": 0.685,
|
| 305 |
+
"rewards/accuracies": 0.6031249761581421,
|
| 306 |
+
"rewards/chosen": -0.005181929562240839,
|
| 307 |
+
"rewards/margins": 0.016439586877822876,
|
| 308 |
+
"rewards/rejected": -0.021621517837047577,
|
| 309 |
+
"step": 190
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.31834460803820136,
|
| 313 |
+
"grad_norm": 2.39968390976015,
|
| 314 |
+
"learning_rate": 4.3090443821097566e-07,
|
| 315 |
+
"logits/chosen": -1.7716095447540283,
|
| 316 |
+
"logits/rejected": -1.8057435750961304,
|
| 317 |
+
"logps/chosen": -191.8214111328125,
|
| 318 |
+
"logps/rejected": -202.6089630126953,
|
| 319 |
+
"loss": 0.6834,
|
| 320 |
+
"rewards/accuracies": 0.612500011920929,
|
| 321 |
+
"rewards/chosen": -0.008436818607151508,
|
| 322 |
+
"rewards/margins": 0.02256094664335251,
|
| 323 |
+
"rewards/rejected": -0.030997764319181442,
|
| 324 |
+
"step": 200
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.3342618384401114,
|
| 328 |
+
"grad_norm": 1.2956793625076484,
|
| 329 |
+
"learning_rate": 4.210354171785795e-07,
|
| 330 |
+
"logits/chosen": -1.7898409366607666,
|
| 331 |
+
"logits/rejected": -1.8095554113388062,
|
| 332 |
+
"logps/chosen": -187.91336059570312,
|
| 333 |
+
"logps/rejected": -191.72274780273438,
|
| 334 |
+
"loss": 0.6822,
|
| 335 |
+
"rewards/accuracies": 0.6031249761581421,
|
| 336 |
+
"rewards/chosen": -0.006862046662718058,
|
| 337 |
+
"rewards/margins": 0.026659756898880005,
|
| 338 |
+
"rewards/rejected": -0.033521804958581924,
|
| 339 |
+
"step": 210
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.3501790688420215,
|
| 343 |
+
"grad_norm": 2.0041098363033623,
|
| 344 |
+
"learning_rate": 4.1063773547332584e-07,
|
| 345 |
+
"logits/chosen": -1.7102569341659546,
|
| 346 |
+
"logits/rejected": -1.7631139755249023,
|
| 347 |
+
"logps/chosen": -192.54238891601562,
|
| 348 |
+
"logps/rejected": -196.4750213623047,
|
| 349 |
+
"loss": 0.6811,
|
| 350 |
+
"rewards/accuracies": 0.515625,
|
| 351 |
+
"rewards/chosen": -0.02755718305706978,
|
| 352 |
+
"rewards/margins": 0.021966466680169106,
|
| 353 |
+
"rewards/rejected": -0.04952365159988403,
|
| 354 |
+
"step": 220
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.36609629924393156,
|
| 358 |
+
"grad_norm": 1.9332381756111658,
|
| 359 |
+
"learning_rate": 3.997435317334988e-07,
|
| 360 |
+
"logits/chosen": -1.7369827032089233,
|
| 361 |
+
"logits/rejected": -1.7770426273345947,
|
| 362 |
+
"logps/chosen": -191.87298583984375,
|
| 363 |
+
"logps/rejected": -193.9591522216797,
|
| 364 |
+
"loss": 0.6817,
|
| 365 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 366 |
+
"rewards/chosen": -0.043014999479055405,
|
| 367 |
+
"rewards/margins": 0.020200857892632484,
|
| 368 |
+
"rewards/rejected": -0.06321585923433304,
|
| 369 |
+
"step": 230
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.3820135296458416,
|
| 373 |
+
"grad_norm": 1.8903913985972478,
|
| 374 |
+
"learning_rate": 3.8838647931853684e-07,
|
| 375 |
+
"logits/chosen": -1.7956278324127197,
|
| 376 |
+
"logits/rejected": -1.7797701358795166,
|
| 377 |
+
"logps/chosen": -191.063720703125,
|
| 378 |
+
"logps/rejected": -194.3412322998047,
|
| 379 |
+
"loss": 0.6807,
|
| 380 |
+
"rewards/accuracies": 0.596875011920929,
|
| 381 |
+
"rewards/chosen": -0.04468019679188728,
|
| 382 |
+
"rewards/margins": 0.024412112310528755,
|
| 383 |
+
"rewards/rejected": -0.06909231096506119,
|
| 384 |
+
"step": 240
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.3979307600477517,
|
| 388 |
+
"grad_norm": 2.209986978975659,
|
| 389 |
+
"learning_rate": 3.7660168222660824e-07,
|
| 390 |
+
"logits/chosen": -1.7686593532562256,
|
| 391 |
+
"logits/rejected": -1.784973382949829,
|
| 392 |
+
"logps/chosen": -210.4496612548828,
|
| 393 |
+
"logps/rejected": -210.5813751220703,
|
| 394 |
+
"loss": 0.6764,
|
| 395 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 396 |
+
"rewards/chosen": -0.06448944658041,
|
| 397 |
+
"rewards/margins": 0.04242020100355148,
|
| 398 |
+
"rewards/rejected": -0.10690965503454208,
|
| 399 |
+
"step": 250
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.41384799044966175,
|
| 403 |
+
"grad_norm": 2.7492197743164577,
|
| 404 |
+
"learning_rate": 3.6442556659016475e-07,
|
| 405 |
+
"logits/chosen": -1.6692039966583252,
|
| 406 |
+
"logits/rejected": -1.7762839794158936,
|
| 407 |
+
"logps/chosen": -204.4394073486328,
|
| 408 |
+
"logps/rejected": -207.07705688476562,
|
| 409 |
+
"loss": 0.6753,
|
| 410 |
+
"rewards/accuracies": 0.590624988079071,
|
| 411 |
+
"rewards/chosen": -0.07524862885475159,
|
| 412 |
+
"rewards/margins": 0.02850642427802086,
|
| 413 |
+
"rewards/rejected": -0.10375505685806274,
|
| 414 |
+
"step": 260
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.4297652208515718,
|
| 418 |
+
"grad_norm": 2.266511648349345,
|
| 419 |
+
"learning_rate": 3.5189576808485404e-07,
|
| 420 |
+
"logits/chosen": -1.7136598825454712,
|
| 421 |
+
"logits/rejected": -1.710269570350647,
|
| 422 |
+
"logps/chosen": -208.0347137451172,
|
| 423 |
+
"logps/rejected": -209.95751953125,
|
| 424 |
+
"loss": 0.6737,
|
| 425 |
+
"rewards/accuracies": 0.578125,
|
| 426 |
+
"rewards/chosen": -0.08503381162881851,
|
| 427 |
+
"rewards/margins": 0.0269694272428751,
|
| 428 |
+
"rewards/rejected": -0.11200324445962906,
|
| 429 |
+
"step": 270
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.4456824512534819,
|
| 433 |
+
"grad_norm": 2.5291958520679354,
|
| 434 |
+
"learning_rate": 3.390510155998023e-07,
|
| 435 |
+
"logits/chosen": -1.628941297531128,
|
| 436 |
+
"logits/rejected": -1.6508426666259766,
|
| 437 |
+
"logps/chosen": -214.1870574951172,
|
| 438 |
+
"logps/rejected": -212.94284057617188,
|
| 439 |
+
"loss": 0.6707,
|
| 440 |
+
"rewards/accuracies": 0.6468750238418579,
|
| 441 |
+
"rewards/chosen": -0.12327191978693008,
|
| 442 |
+
"rewards/margins": 0.07120023667812347,
|
| 443 |
+
"rewards/rejected": -0.19447214901447296,
|
| 444 |
+
"step": 280
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.46159968165539195,
|
| 448 |
+
"grad_norm": 2.574405785518012,
|
| 449 |
+
"learning_rate": 3.2593101152883795e-07,
|
| 450 |
+
"logits/chosen": -1.6216236352920532,
|
| 451 |
+
"logits/rejected": -1.6766021251678467,
|
| 452 |
+
"logps/chosen": -216.7034149169922,
|
| 453 |
+
"logps/rejected": -219.9063262939453,
|
| 454 |
+
"loss": 0.6732,
|
| 455 |
+
"rewards/accuracies": 0.621874988079071,
|
| 456 |
+
"rewards/chosen": -0.16276732087135315,
|
| 457 |
+
"rewards/margins": 0.05662418156862259,
|
| 458 |
+
"rewards/rejected": -0.21939150989055634,
|
| 459 |
+
"step": 290
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.477516912057302,
|
| 463 |
+
"grad_norm": 2.6453554421569514,
|
| 464 |
+
"learning_rate": 3.125763090526674e-07,
|
| 465 |
+
"logits/chosen": -1.5658096075057983,
|
| 466 |
+
"logits/rejected": -1.609926462173462,
|
| 467 |
+
"logps/chosen": -212.8961944580078,
|
| 468 |
+
"logps/rejected": -213.0232696533203,
|
| 469 |
+
"loss": 0.6673,
|
| 470 |
+
"rewards/accuracies": 0.596875011920929,
|
| 471 |
+
"rewards/chosen": -0.14756593108177185,
|
| 472 |
+
"rewards/margins": 0.06441614031791687,
|
| 473 |
+
"rewards/rejected": -0.21198205649852753,
|
| 474 |
+
"step": 300
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.4934341424592121,
|
| 478 |
+
"grad_norm": 4.170174585407687,
|
| 479 |
+
"learning_rate": 2.9902818679131775e-07,
|
| 480 |
+
"logits/chosen": -1.5784204006195068,
|
| 481 |
+
"logits/rejected": -1.5990747213363647,
|
| 482 |
+
"logps/chosen": -216.91799926757812,
|
| 483 |
+
"logps/rejected": -221.1397705078125,
|
| 484 |
+
"loss": 0.665,
|
| 485 |
+
"rewards/accuracies": 0.590624988079071,
|
| 486 |
+
"rewards/chosen": -0.20333552360534668,
|
| 487 |
+
"rewards/margins": 0.05764692276716232,
|
| 488 |
+
"rewards/rejected": -0.2609824538230896,
|
| 489 |
+
"step": 310
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.5093513728611222,
|
| 493 |
+
"grad_norm": 2.5979901565172714,
|
| 494 |
+
"learning_rate": 2.8532852121428733e-07,
|
| 495 |
+
"logits/chosen": -1.5708476305007935,
|
| 496 |
+
"logits/rejected": -1.610568642616272,
|
| 497 |
+
"logps/chosen": -212.8919219970703,
|
| 498 |
+
"logps/rejected": -209.1988983154297,
|
| 499 |
+
"loss": 0.6602,
|
| 500 |
+
"rewards/accuracies": 0.6031249761581421,
|
| 501 |
+
"rewards/chosen": -0.21002519130706787,
|
| 502 |
+
"rewards/margins": 0.07976146787405014,
|
| 503 |
+
"rewards/rejected": -0.2897866368293762,
|
| 504 |
+
"step": 320
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 0.5252686032630323,
|
| 508 |
+
"grad_norm": 3.0268583107868676,
|
| 509 |
+
"learning_rate": 2.715196572027789e-07,
|
| 510 |
+
"logits/chosen": -1.5595386028289795,
|
| 511 |
+
"logits/rejected": -1.579420804977417,
|
| 512 |
+
"logps/chosen": -208.55126953125,
|
| 513 |
+
"logps/rejected": -228.11337280273438,
|
| 514 |
+
"loss": 0.6561,
|
| 515 |
+
"rewards/accuracies": 0.6343749761581421,
|
| 516 |
+
"rewards/chosen": -0.19552794098854065,
|
| 517 |
+
"rewards/margins": 0.11572189629077911,
|
| 518 |
+
"rewards/rejected": -0.31124982237815857,
|
| 519 |
+
"step": 330
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.5411858336649423,
|
| 523 |
+
"grad_norm": 2.608251205411858,
|
| 524 |
+
"learning_rate": 2.5764427716409815e-07,
|
| 525 |
+
"logits/chosen": -1.5596858263015747,
|
| 526 |
+
"logits/rejected": -1.5745502710342407,
|
| 527 |
+
"logps/chosen": -222.6412811279297,
|
| 528 |
+
"logps/rejected": -227.70651245117188,
|
| 529 |
+
"loss": 0.6544,
|
| 530 |
+
"rewards/accuracies": 0.640625,
|
| 531 |
+
"rewards/chosen": -0.21032941341400146,
|
| 532 |
+
"rewards/margins": 0.09801909327507019,
|
| 533 |
+
"rewards/rejected": -0.30834850668907166,
|
| 534 |
+
"step": 340
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.5571030640668524,
|
| 538 |
+
"grad_norm": 2.3596476657059666,
|
| 539 |
+
"learning_rate": 2.4374526910277886e-07,
|
| 540 |
+
"logits/chosen": -1.4910805225372314,
|
| 541 |
+
"logits/rejected": -1.4995691776275635,
|
| 542 |
+
"logps/chosen": -219.46536254882812,
|
| 543 |
+
"logps/rejected": -235.89620971679688,
|
| 544 |
+
"loss": 0.6613,
|
| 545 |
+
"rewards/accuracies": 0.625,
|
| 546 |
+
"rewards/chosen": -0.2539146840572357,
|
| 547 |
+
"rewards/margins": 0.09661410748958588,
|
| 548 |
+
"rewards/rejected": -0.3505287766456604,
|
| 549 |
+
"step": 350
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.5730202944687625,
|
| 553 |
+
"grad_norm": 2.721512353744307,
|
| 554 |
+
"learning_rate": 2.2986559405621886e-07,
|
| 555 |
+
"logits/chosen": -1.5120307207107544,
|
| 556 |
+
"logits/rejected": -1.5819119215011597,
|
| 557 |
+
"logps/chosen": -236.832763671875,
|
| 558 |
+
"logps/rejected": -235.5454559326172,
|
| 559 |
+
"loss": 0.6668,
|
| 560 |
+
"rewards/accuracies": 0.565625011920929,
|
| 561 |
+
"rewards/chosen": -0.3000979721546173,
|
| 562 |
+
"rewards/margins": 0.07621388137340546,
|
| 563 |
+
"rewards/rejected": -0.37631186842918396,
|
| 564 |
+
"step": 360
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.5889375248706725,
|
| 568 |
+
"grad_norm": 3.020330440823815,
|
| 569 |
+
"learning_rate": 2.160481533045751e-07,
|
| 570 |
+
"logits/chosen": -1.4668768644332886,
|
| 571 |
+
"logits/rejected": -1.5363471508026123,
|
| 572 |
+
"logps/chosen": -218.35946655273438,
|
| 573 |
+
"logps/rejected": -224.10208129882812,
|
| 574 |
+
"loss": 0.6492,
|
| 575 |
+
"rewards/accuracies": 0.6156250238418579,
|
| 576 |
+
"rewards/chosen": -0.2719998359680176,
|
| 577 |
+
"rewards/margins": 0.12549051642417908,
|
| 578 |
+
"rewards/rejected": -0.39749038219451904,
|
| 579 |
+
"step": 370
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.6048547552725826,
|
| 583 |
+
"grad_norm": 2.9446946400840974,
|
| 584 |
+
"learning_rate": 2.0233565576536564e-07,
|
| 585 |
+
"logits/chosen": -1.485190749168396,
|
| 586 |
+
"logits/rejected": -1.5274041891098022,
|
| 587 |
+
"logps/chosen": -222.81765747070312,
|
| 588 |
+
"logps/rejected": -222.657470703125,
|
| 589 |
+
"loss": 0.6538,
|
| 590 |
+
"rewards/accuracies": 0.59375,
|
| 591 |
+
"rewards/chosen": -0.26167741417884827,
|
| 592 |
+
"rewards/margins": 0.08558806031942368,
|
| 593 |
+
"rewards/rejected": -0.34726545214653015,
|
| 594 |
+
"step": 380
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 0.6207719856744927,
|
| 598 |
+
"grad_norm": 3.1883509652873223,
|
| 599 |
+
"learning_rate": 1.887704859826528e-07,
|
| 600 |
+
"logits/chosen": -1.4792619943618774,
|
| 601 |
+
"logits/rejected": -1.5055263042449951,
|
| 602 |
+
"logps/chosen": -221.5879364013672,
|
| 603 |
+
"logps/rejected": -239.82571411132812,
|
| 604 |
+
"loss": 0.6495,
|
| 605 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 606 |
+
"rewards/chosen": -0.2849220633506775,
|
| 607 |
+
"rewards/margins": 0.13855159282684326,
|
| 608 |
+
"rewards/rejected": -0.42347365617752075,
|
| 609 |
+
"step": 390
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.6366892160764027,
|
| 613 |
+
"grad_norm": 2.8954609378442657,
|
| 614 |
+
"learning_rate": 1.7539457311884675e-07,
|
| 615 |
+
"logits/chosen": -1.4238282442092896,
|
| 616 |
+
"logits/rejected": -1.423081398010254,
|
| 617 |
+
"logps/chosen": -229.7063446044922,
|
| 618 |
+
"logps/rejected": -235.14236450195312,
|
| 619 |
+
"loss": 0.6522,
|
| 620 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 621 |
+
"rewards/chosen": -0.2554658353328705,
|
| 622 |
+
"rewards/margins": 0.14818701148033142,
|
| 623 |
+
"rewards/rejected": -0.4036528468132019,
|
| 624 |
+
"step": 400
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 0.6526064464783128,
|
| 628 |
+
"grad_norm": 3.203997856966496,
|
| 629 |
+
"learning_rate": 1.6224926135406693e-07,
|
| 630 |
+
"logits/chosen": -1.460460901260376,
|
| 631 |
+
"logits/rejected": -1.4777477979660034,
|
| 632 |
+
"logps/chosen": -236.0177764892578,
|
| 633 |
+
"logps/rejected": -240.2278289794922,
|
| 634 |
+
"loss": 0.6525,
|
| 635 |
+
"rewards/accuracies": 0.596875011920929,
|
| 636 |
+
"rewards/chosen": -0.32773512601852417,
|
| 637 |
+
"rewards/margins": 0.08220280706882477,
|
| 638 |
+
"rewards/rejected": -0.4099380075931549,
|
| 639 |
+
"step": 410
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.6685236768802229,
|
| 643 |
+
"grad_norm": 2.8800206835573094,
|
| 644 |
+
"learning_rate": 1.4937518209365108e-07,
|
| 645 |
+
"logits/chosen": -1.5116090774536133,
|
| 646 |
+
"logits/rejected": -1.5359185934066772,
|
| 647 |
+
"logps/chosen": -248.60263061523438,
|
| 648 |
+
"logps/rejected": -245.75967407226562,
|
| 649 |
+
"loss": 0.6527,
|
| 650 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 651 |
+
"rewards/chosen": -0.3213188052177429,
|
| 652 |
+
"rewards/margins": 0.10633780807256699,
|
| 653 |
+
"rewards/rejected": -0.4276565909385681,
|
| 654 |
+
"step": 420
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.6844409072821329,
|
| 658 |
+
"grad_norm": 3.29734083048723,
|
| 659 |
+
"learning_rate": 1.3681212837880977e-07,
|
| 660 |
+
"logits/chosen": -1.4253356456756592,
|
| 661 |
+
"logits/rejected": -1.444923758506775,
|
| 662 |
+
"logps/chosen": -220.8928985595703,
|
| 663 |
+
"logps/rejected": -238.1382293701172,
|
| 664 |
+
"loss": 0.6502,
|
| 665 |
+
"rewards/accuracies": 0.621874988079071,
|
| 666 |
+
"rewards/chosen": -0.3292413353919983,
|
| 667 |
+
"rewards/margins": 0.10124459117650986,
|
| 668 |
+
"rewards/rejected": -0.43048587441444397,
|
| 669 |
+
"step": 430
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.700358137684043,
|
| 673 |
+
"grad_norm": 3.2357462634957908,
|
| 674 |
+
"learning_rate": 1.2459893188861613e-07,
|
| 675 |
+
"logits/chosen": -1.3886950016021729,
|
| 676 |
+
"logits/rejected": -1.4186146259307861,
|
| 677 |
+
"logps/chosen": -229.12069702148438,
|
| 678 |
+
"logps/rejected": -241.5669708251953,
|
| 679 |
+
"loss": 0.65,
|
| 680 |
+
"rewards/accuracies": 0.6656249761581421,
|
| 681 |
+
"rewards/chosen": -0.31468960642814636,
|
| 682 |
+
"rewards/margins": 0.14469651877880096,
|
| 683 |
+
"rewards/rejected": -0.4593861699104309,
|
| 684 |
+
"step": 440
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 0.716275368085953,
|
| 688 |
+
"grad_norm": 3.664386493120931,
|
| 689 |
+
"learning_rate": 1.1277334291351145e-07,
|
| 690 |
+
"logits/chosen": -1.4633034467697144,
|
| 691 |
+
"logits/rejected": -1.4953584671020508,
|
| 692 |
+
"logps/chosen": -229.84976196289062,
|
| 693 |
+
"logps/rejected": -245.46762084960938,
|
| 694 |
+
"loss": 0.6451,
|
| 695 |
+
"rewards/accuracies": 0.643750011920929,
|
| 696 |
+
"rewards/chosen": -0.35235971212387085,
|
| 697 |
+
"rewards/margins": 0.1600349247455597,
|
| 698 |
+
"rewards/rejected": -0.5123946666717529,
|
| 699 |
+
"step": 450
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.7321925984878631,
|
| 703 |
+
"grad_norm": 3.054029038543547,
|
| 704 |
+
"learning_rate": 1.0137191367132078e-07,
|
| 705 |
+
"logits/chosen": -1.466225504875183,
|
| 706 |
+
"logits/rejected": -1.5026957988739014,
|
| 707 |
+
"logps/chosen": -227.568359375,
|
| 708 |
+
"logps/rejected": -251.8888397216797,
|
| 709 |
+
"loss": 0.6449,
|
| 710 |
+
"rewards/accuracies": 0.6468750238418579,
|
| 711 |
+
"rewards/chosen": -0.2990425229072571,
|
| 712 |
+
"rewards/margins": 0.16379448771476746,
|
| 713 |
+
"rewards/rejected": -0.46283698081970215,
|
| 714 |
+
"step": 460
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 0.7481098288897732,
|
| 718 |
+
"grad_norm": 3.604867452453406,
|
| 719 |
+
"learning_rate": 9.042988532644249e-08,
|
| 720 |
+
"logits/chosen": -1.447538137435913,
|
| 721 |
+
"logits/rejected": -1.5032737255096436,
|
| 722 |
+
"logps/chosen": -240.51644897460938,
|
| 723 |
+
"logps/rejected": -250.5240020751953,
|
| 724 |
+
"loss": 0.6527,
|
| 725 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 726 |
+
"rewards/chosen": -0.36844930052757263,
|
| 727 |
+
"rewards/margins": 0.1114746555685997,
|
| 728 |
+
"rewards/rejected": -0.47992390394210815,
|
| 729 |
+
"step": 470
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.7640270592916832,
|
| 733 |
+
"grad_norm": 3.100922028441707,
|
| 734 |
+
"learning_rate": 7.998107906142839e-08,
|
| 735 |
+
"logits/chosen": -1.4335956573486328,
|
| 736 |
+
"logits/rejected": -1.468783974647522,
|
| 737 |
+
"logps/chosen": -220.85147094726562,
|
| 738 |
+
"logps/rejected": -234.76596069335938,
|
| 739 |
+
"loss": 0.6462,
|
| 740 |
+
"rewards/accuracies": 0.640625,
|
| 741 |
+
"rewards/chosen": -0.3198398947715759,
|
| 742 |
+
"rewards/margins": 0.13547857105731964,
|
| 743 |
+
"rewards/rejected": -0.45531851053237915,
|
| 744 |
+
"step": 480
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.7799442896935933,
|
| 748 |
+
"grad_norm": 3.0897060153169447,
|
| 749 |
+
"learning_rate": 7.005779153764682e-08,
|
| 750 |
+
"logits/chosen": -1.460636854171753,
|
| 751 |
+
"logits/rejected": -1.481178641319275,
|
| 752 |
+
"logps/chosen": -217.1626434326172,
|
| 753 |
+
"logps/rejected": -241.839111328125,
|
| 754 |
+
"loss": 0.6465,
|
| 755 |
+
"rewards/accuracies": 0.628125011920929,
|
| 756 |
+
"rewards/chosen": -0.33777886629104614,
|
| 757 |
+
"rewards/margins": 0.1417325884103775,
|
| 758 |
+
"rewards/rejected": -0.47951143980026245,
|
| 759 |
+
"step": 490
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.7958615200955034,
|
| 763 |
+
"grad_norm": 3.515426969892236,
|
| 764 |
+
"learning_rate": 6.069069506815325e-08,
|
| 765 |
+
"logits/chosen": -1.423098087310791,
|
| 766 |
+
"logits/rejected": -1.4111202955245972,
|
| 767 |
+
"logps/chosen": -228.85360717773438,
|
| 768 |
+
"logps/rejected": -247.3131561279297,
|
| 769 |
+
"loss": 0.6459,
|
| 770 |
+
"rewards/accuracies": 0.6468750238418579,
|
| 771 |
+
"rewards/chosen": -0.3209570348262787,
|
| 772 |
+
"rewards/margins": 0.18556544184684753,
|
| 773 |
+
"rewards/rejected": -0.506522536277771,
|
| 774 |
+
"step": 500
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 0.8117787504974134,
|
| 778 |
+
"grad_norm": 3.168259746643005,
|
| 779 |
+
"learning_rate": 5.190874281132851e-08,
|
| 780 |
+
"logits/chosen": -1.3531607389450073,
|
| 781 |
+
"logits/rejected": -1.4315484762191772,
|
| 782 |
+
"logps/chosen": -228.17135620117188,
|
| 783 |
+
"logps/rejected": -237.56796264648438,
|
| 784 |
+
"loss": 0.6446,
|
| 785 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 786 |
+
"rewards/chosen": -0.3634713888168335,
|
| 787 |
+
"rewards/margins": 0.11672137677669525,
|
| 788 |
+
"rewards/rejected": -0.48019275069236755,
|
| 789 |
+
"step": 510
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.8276959808993235,
|
| 793 |
+
"grad_norm": 3.547783875081257,
|
| 794 |
+
"learning_rate": 4.373907927832513e-08,
|
| 795 |
+
"logits/chosen": -1.4724462032318115,
|
| 796 |
+
"logits/rejected": -1.4711410999298096,
|
| 797 |
+
"logps/chosen": -216.9486541748047,
|
| 798 |
+
"logps/rejected": -229.16531372070312,
|
| 799 |
+
"loss": 0.6497,
|
| 800 |
+
"rewards/accuracies": 0.574999988079071,
|
| 801 |
+
"rewards/chosen": -0.34610506892204285,
|
| 802 |
+
"rewards/margins": 0.10790134966373444,
|
| 803 |
+
"rewards/rejected": -0.4540063738822937,
|
| 804 |
+
"step": 520
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 0.8436132113012336,
|
| 808 |
+
"grad_norm": 3.0690519001773744,
|
| 809 |
+
"learning_rate": 3.620695643093924e-08,
|
| 810 |
+
"logits/chosen": -1.436292290687561,
|
| 811 |
+
"logits/rejected": -1.4620921611785889,
|
| 812 |
+
"logps/chosen": -220.2271728515625,
|
| 813 |
+
"logps/rejected": -242.4230499267578,
|
| 814 |
+
"loss": 0.6444,
|
| 815 |
+
"rewards/accuracies": 0.671875,
|
| 816 |
+
"rewards/chosen": -0.2857658565044403,
|
| 817 |
+
"rewards/margins": 0.16418084502220154,
|
| 818 |
+
"rewards/rejected": -0.44994670152664185,
|
| 819 |
+
"step": 530
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 0.8595304417031436,
|
| 823 |
+
"grad_norm": 3.5137530293056676,
|
| 824 |
+
"learning_rate": 2.9335655629243645e-08,
|
| 825 |
+
"logits/chosen": -1.4339349269866943,
|
| 826 |
+
"logits/rejected": -1.4756128787994385,
|
| 827 |
+
"logps/chosen": -225.1971893310547,
|
| 828 |
+
"logps/rejected": -240.7228546142578,
|
| 829 |
+
"loss": 0.6474,
|
| 830 |
+
"rewards/accuracies": 0.628125011920929,
|
| 831 |
+
"rewards/chosen": -0.32935741543769836,
|
| 832 |
+
"rewards/margins": 0.12228095531463623,
|
| 833 |
+
"rewards/rejected": -0.4516383707523346,
|
| 834 |
+
"step": 540
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 0.8754476721050537,
|
| 838 |
+
"grad_norm": 3.303113230393075,
|
| 839 |
+
"learning_rate": 2.31464156702382e-08,
|
| 840 |
+
"logits/chosen": -1.4354156255722046,
|
| 841 |
+
"logits/rejected": -1.4688793420791626,
|
| 842 |
+
"logps/chosen": -228.87588500976562,
|
| 843 |
+
"logps/rejected": -227.47872924804688,
|
| 844 |
+
"loss": 0.6503,
|
| 845 |
+
"rewards/accuracies": 0.609375,
|
| 846 |
+
"rewards/chosen": -0.31748342514038086,
|
| 847 |
+
"rewards/margins": 0.12047611176967621,
|
| 848 |
+
"rewards/rejected": -0.4379595220088959,
|
| 849 |
+
"step": 550
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.8913649025069638,
|
| 853 |
+
"grad_norm": 3.6187583437530138,
|
| 854 |
+
"learning_rate": 1.7658367139945228e-08,
|
| 855 |
+
"logits/chosen": -1.484116554260254,
|
| 856 |
+
"logits/rejected": -1.4669333696365356,
|
| 857 |
+
"logps/chosen": -220.8059844970703,
|
| 858 |
+
"logps/rejected": -235.5103759765625,
|
| 859 |
+
"loss": 0.6422,
|
| 860 |
+
"rewards/accuracies": 0.578125,
|
| 861 |
+
"rewards/chosen": -0.34964197874069214,
|
| 862 |
+
"rewards/margins": 0.1103706806898117,
|
| 863 |
+
"rewards/rejected": -0.46001267433166504,
|
| 864 |
+
"step": 560
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.9072821329088738,
|
| 868 |
+
"grad_norm": 2.946862169915585,
|
| 869 |
+
"learning_rate": 1.2888473281864597e-08,
|
| 870 |
+
"logits/chosen": -1.3717347383499146,
|
| 871 |
+
"logits/rejected": -1.4307266473770142,
|
| 872 |
+
"logps/chosen": -230.9296112060547,
|
| 873 |
+
"logps/rejected": -248.7758026123047,
|
| 874 |
+
"loss": 0.6373,
|
| 875 |
+
"rewards/accuracies": 0.6468750238418579,
|
| 876 |
+
"rewards/chosen": -0.3366944193840027,
|
| 877 |
+
"rewards/margins": 0.13169880211353302,
|
| 878 |
+
"rewards/rejected": -0.4683932363986969,
|
| 879 |
+
"step": 570
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 0.9231993633107839,
|
| 883 |
+
"grad_norm": 3.986364311332544,
|
| 884 |
+
"learning_rate": 8.851477564560061e-09,
|
| 885 |
+
"logits/chosen": -1.4669787883758545,
|
| 886 |
+
"logits/rejected": -1.5305414199829102,
|
| 887 |
+
"logps/chosen": -227.6433563232422,
|
| 888 |
+
"logps/rejected": -243.227294921875,
|
| 889 |
+
"loss": 0.6434,
|
| 890 |
+
"rewards/accuracies": 0.6468750238418579,
|
| 891 |
+
"rewards/chosen": -0.32894089818000793,
|
| 892 |
+
"rewards/margins": 0.15033873915672302,
|
| 893 |
+
"rewards/rejected": -0.47927966713905334,
|
| 894 |
+
"step": 580
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 0.939116593712694,
|
| 898 |
+
"grad_norm": 3.4745032456527394,
|
| 899 |
+
"learning_rate": 5.559858110443016e-09,
|
| 900 |
+
"logits/chosen": -1.418752908706665,
|
| 901 |
+
"logits/rejected": -1.463607668876648,
|
| 902 |
+
"logps/chosen": -237.24252319335938,
|
| 903 |
+
"logps/rejected": -252.40145874023438,
|
| 904 |
+
"loss": 0.6384,
|
| 905 |
+
"rewards/accuracies": 0.625,
|
| 906 |
+
"rewards/chosen": -0.3376900553703308,
|
| 907 |
+
"rewards/margins": 0.15840700268745422,
|
| 908 |
+
"rewards/rejected": -0.4960970878601074,
|
| 909 |
+
"step": 590
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 0.955033824114604,
|
| 913 |
+
"grad_norm": 6.105907684391068,
|
| 914 |
+
"learning_rate": 3.023789126611137e-09,
|
| 915 |
+
"logits/chosen": -1.4318501949310303,
|
| 916 |
+
"logits/rejected": -1.4236361980438232,
|
| 917 |
+
"logps/chosen": -224.8287811279297,
|
| 918 |
+
"logps/rejected": -240.1597442626953,
|
| 919 |
+
"loss": 0.6398,
|
| 920 |
+
"rewards/accuracies": 0.699999988079071,
|
| 921 |
+
"rewards/chosen": -0.32297495007514954,
|
| 922 |
+
"rewards/margins": 0.1879177987575531,
|
| 923 |
+
"rewards/rejected": -0.5108927488327026,
|
| 924 |
+
"step": 600
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 0.9709510545165141,
|
| 928 |
+
"grad_norm": 7.921761427791219,
|
| 929 |
+
"learning_rate": 1.2511094569571668e-09,
|
| 930 |
+
"logits/chosen": -1.4697223901748657,
|
| 931 |
+
"logits/rejected": -1.5091297626495361,
|
| 932 |
+
"logps/chosen": -227.2742156982422,
|
| 933 |
+
"logps/rejected": -238.57974243164062,
|
| 934 |
+
"loss": 0.6414,
|
| 935 |
+
"rewards/accuracies": 0.596875011920929,
|
| 936 |
+
"rewards/chosen": -0.3365384042263031,
|
| 937 |
+
"rewards/margins": 0.1458342969417572,
|
| 938 |
+
"rewards/rejected": -0.4823727011680603,
|
| 939 |
+
"step": 610
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 0.9868682849184242,
|
| 943 |
+
"grad_norm": 3.1578814752663407,
|
| 944 |
+
"learning_rate": 2.4729835275189016e-10,
|
| 945 |
+
"logits/chosen": -1.4319922924041748,
|
| 946 |
+
"logits/rejected": -1.4167181253433228,
|
| 947 |
+
"logps/chosen": -224.8016357421875,
|
| 948 |
+
"logps/rejected": -247.70339965820312,
|
| 949 |
+
"loss": 0.6404,
|
| 950 |
+
"rewards/accuracies": 0.6343749761581421,
|
| 951 |
+
"rewards/chosen": -0.36974093317985535,
|
| 952 |
+
"rewards/margins": 0.14402151107788086,
|
| 953 |
+
"rewards/rejected": -0.5137624740600586,
|
| 954 |
+
"step": 620
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.9996020692399522,
|
| 958 |
+
"step": 628,
|
| 959 |
+
"total_flos": 0.0,
|
| 960 |
+
"train_loss": 0.6662861807331159,
|
| 961 |
+
"train_runtime": 15158.4407,
|
| 962 |
+
"train_samples_per_second": 10.608,
|
| 963 |
+
"train_steps_per_second": 0.041
|
| 964 |
+
}
|
| 965 |
+
],
|
| 966 |
+
"logging_steps": 10,
|
| 967 |
+
"max_steps": 628,
|
| 968 |
+
"num_input_tokens_seen": 0,
|
| 969 |
+
"num_train_epochs": 1,
|
| 970 |
+
"save_steps": 100,
|
| 971 |
+
"stateful_callbacks": {
|
| 972 |
+
"TrainerControl": {
|
| 973 |
+
"args": {
|
| 974 |
+
"should_epoch_stop": false,
|
| 975 |
+
"should_evaluate": false,
|
| 976 |
+
"should_log": false,
|
| 977 |
+
"should_save": true,
|
| 978 |
+
"should_training_stop": true
|
| 979 |
+
},
|
| 980 |
+
"attributes": {}
|
| 981 |
+
}
|
| 982 |
+
},
|
| 983 |
+
"total_flos": 0.0,
|
| 984 |
+
"train_batch_size": 8,
|
| 985 |
+
"trial_name": null,
|
| 986 |
+
"trial_params": null
|
| 987 |
+
}
|