Commit
•
5d26a1b
1
Parent(s):
85d3677
Model save
Browse files- README.md +61 -0
- all_results.json +9 -0
- generation_config.json +6 -0
- train_results.json +9 -0
- trainer_state.json +1137 -0
README.md
ADDED
|
@@ -0,0 +1,61 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
license: apache-2.0
|
| 4 |
+
base_model: alignment-handbook/zephyr-7b-sft-full
|
| 5 |
+
tags:
|
| 6 |
+
- trl
|
| 7 |
+
- dpo
|
| 8 |
+
- generated_from_trainer
|
| 9 |
+
model-index:
|
| 10 |
+
- name: uf-rlced-conifer-zephyr-7b-dpo-full
|
| 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 |
+
# uf-rlced-conifer-zephyr-7b-dpo-full
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on an unknown dataset.
|
| 20 |
+
|
| 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: 1
|
| 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": 0.9993060374739764,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.32248947852187687,
|
| 5 |
+
"train_runtime": 20768.7862,
|
| 6 |
+
"train_samples": 184443,
|
| 7 |
+
"train_samples_per_second": 8.881,
|
| 8 |
+
"train_steps_per_second": 0.035
|
| 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": 0.9993060374739764,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.32248947852187687,
|
| 5 |
+
"train_runtime": 20768.7862,
|
| 6 |
+
"train_samples": 184443,
|
| 7 |
+
"train_samples_per_second": 8.881,
|
| 8 |
+
"train_steps_per_second": 0.035
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1137 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9993060374739764,
|
| 5 |
+
"eval_steps": 1000,
|
| 6 |
+
"global_step": 720,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0013879250520471894,
|
| 13 |
+
"grad_norm": 7.102522482374833,
|
| 14 |
+
"learning_rate": 6.9444444444444435e-09,
|
| 15 |
+
"logits/chosen": -2.6390600204467773,
|
| 16 |
+
"logits/rejected": -2.659458637237549,
|
| 17 |
+
"logps/chosen": -349.352783203125,
|
| 18 |
+
"logps/rejected": -372.3662109375,
|
| 19 |
+
"loss": 0.6931,
|
| 20 |
+
"rewards/accuracies": 0.0,
|
| 21 |
+
"rewards/chosen": 0.0,
|
| 22 |
+
"rewards/margins": 0.0,
|
| 23 |
+
"rewards/rejected": 0.0,
|
| 24 |
+
"step": 1
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.013879250520471894,
|
| 28 |
+
"grad_norm": 7.0553818387235,
|
| 29 |
+
"learning_rate": 6.944444444444444e-08,
|
| 30 |
+
"logits/chosen": -2.701545238494873,
|
| 31 |
+
"logits/rejected": -2.6816134452819824,
|
| 32 |
+
"logps/chosen": -346.0178527832031,
|
| 33 |
+
"logps/rejected": -351.35076904296875,
|
| 34 |
+
"loss": 0.6931,
|
| 35 |
+
"rewards/accuracies": 0.4618055522441864,
|
| 36 |
+
"rewards/chosen": 0.0002754588203970343,
|
| 37 |
+
"rewards/margins": 0.0005549925845116377,
|
| 38 |
+
"rewards/rejected": -0.0002795337641146034,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.027758501040943788,
|
| 43 |
+
"grad_norm": 7.149876435606926,
|
| 44 |
+
"learning_rate": 1.3888888888888888e-07,
|
| 45 |
+
"logits/chosen": -2.7141003608703613,
|
| 46 |
+
"logits/rejected": -2.635436773300171,
|
| 47 |
+
"logps/chosen": -331.3048400878906,
|
| 48 |
+
"logps/rejected": -326.703369140625,
|
| 49 |
+
"loss": 0.6922,
|
| 50 |
+
"rewards/accuracies": 0.574999988079071,
|
| 51 |
+
"rewards/chosen": 0.0008915923535823822,
|
| 52 |
+
"rewards/margins": 0.00229089823551476,
|
| 53 |
+
"rewards/rejected": -0.0013993059983476996,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.041637751561415685,
|
| 58 |
+
"grad_norm": 6.6581229172191145,
|
| 59 |
+
"learning_rate": 2.0833333333333333e-07,
|
| 60 |
+
"logits/chosen": -2.712292194366455,
|
| 61 |
+
"logits/rejected": -2.695763349533081,
|
| 62 |
+
"logps/chosen": -306.89031982421875,
|
| 63 |
+
"logps/rejected": -346.85418701171875,
|
| 64 |
+
"loss": 0.6868,
|
| 65 |
+
"rewards/accuracies": 0.734375,
|
| 66 |
+
"rewards/chosen": 0.008129820227622986,
|
| 67 |
+
"rewards/margins": 0.011882511898875237,
|
| 68 |
+
"rewards/rejected": -0.003752692136913538,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.055517002081887576,
|
| 73 |
+
"grad_norm": 7.5041216352968245,
|
| 74 |
+
"learning_rate": 2.7777777777777776e-07,
|
| 75 |
+
"logits/chosen": -2.7055609226226807,
|
| 76 |
+
"logits/rejected": -2.694276809692383,
|
| 77 |
+
"logps/chosen": -325.8503723144531,
|
| 78 |
+
"logps/rejected": -350.899658203125,
|
| 79 |
+
"loss": 0.6716,
|
| 80 |
+
"rewards/accuracies": 0.828125,
|
| 81 |
+
"rewards/chosen": 0.03933032229542732,
|
| 82 |
+
"rewards/margins": 0.04708551615476608,
|
| 83 |
+
"rewards/rejected": -0.007755190134048462,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.06939625260235947,
|
| 88 |
+
"grad_norm": 8.186890991027267,
|
| 89 |
+
"learning_rate": 3.472222222222222e-07,
|
| 90 |
+
"logits/chosen": -2.708888292312622,
|
| 91 |
+
"logits/rejected": -2.6669304370880127,
|
| 92 |
+
"logps/chosen": -343.9357604980469,
|
| 93 |
+
"logps/rejected": -363.4876403808594,
|
| 94 |
+
"loss": 0.6338,
|
| 95 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 96 |
+
"rewards/chosen": 0.05323966592550278,
|
| 97 |
+
"rewards/margins": 0.13971659541130066,
|
| 98 |
+
"rewards/rejected": -0.08647692948579788,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.08327550312283137,
|
| 103 |
+
"grad_norm": 10.573196863586034,
|
| 104 |
+
"learning_rate": 4.1666666666666667e-07,
|
| 105 |
+
"logits/chosen": -2.7201571464538574,
|
| 106 |
+
"logits/rejected": -2.666919469833374,
|
| 107 |
+
"logps/chosen": -341.60662841796875,
|
| 108 |
+
"logps/rejected": -394.83935546875,
|
| 109 |
+
"loss": 0.5595,
|
| 110 |
+
"rewards/accuracies": 0.8031250238418579,
|
| 111 |
+
"rewards/chosen": -0.09441100060939789,
|
| 112 |
+
"rewards/margins": 0.3541037440299988,
|
| 113 |
+
"rewards/rejected": -0.4485146999359131,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.09715475364330327,
|
| 118 |
+
"grad_norm": 12.834200963453847,
|
| 119 |
+
"learning_rate": 4.861111111111111e-07,
|
| 120 |
+
"logits/chosen": -2.795583724975586,
|
| 121 |
+
"logits/rejected": -2.743293285369873,
|
| 122 |
+
"logps/chosen": -421.5907287597656,
|
| 123 |
+
"logps/rejected": -483.6188049316406,
|
| 124 |
+
"loss": 0.4814,
|
| 125 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 126 |
+
"rewards/chosen": -0.7086803913116455,
|
| 127 |
+
"rewards/margins": 0.6790022253990173,
|
| 128 |
+
"rewards/rejected": -1.3876826763153076,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.11103400416377515,
|
| 133 |
+
"grad_norm": 19.870733060862168,
|
| 134 |
+
"learning_rate": 4.998119881260575e-07,
|
| 135 |
+
"logits/chosen": -2.7935426235198975,
|
| 136 |
+
"logits/rejected": -2.7506136894226074,
|
| 137 |
+
"logps/chosen": -449.7024841308594,
|
| 138 |
+
"logps/rejected": -549.8660888671875,
|
| 139 |
+
"loss": 0.4425,
|
| 140 |
+
"rewards/accuracies": 0.753125011920929,
|
| 141 |
+
"rewards/chosen": -1.2756296396255493,
|
| 142 |
+
"rewards/margins": 0.9220088124275208,
|
| 143 |
+
"rewards/rejected": -2.1976382732391357,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.12491325468424705,
|
| 148 |
+
"grad_norm": 13.159418295939288,
|
| 149 |
+
"learning_rate": 4.990486745229364e-07,
|
| 150 |
+
"logits/chosen": -2.780831813812256,
|
| 151 |
+
"logits/rejected": -2.759742259979248,
|
| 152 |
+
"logps/chosen": -451.6048278808594,
|
| 153 |
+
"logps/rejected": -602.498046875,
|
| 154 |
+
"loss": 0.4078,
|
| 155 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 156 |
+
"rewards/chosen": -1.2828444242477417,
|
| 157 |
+
"rewards/margins": 1.3387337923049927,
|
| 158 |
+
"rewards/rejected": -2.6215782165527344,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.13879250520471895,
|
| 163 |
+
"grad_norm": 19.37754158826291,
|
| 164 |
+
"learning_rate": 4.977001008412112e-07,
|
| 165 |
+
"logits/chosen": -2.5731492042541504,
|
| 166 |
+
"logits/rejected": -2.49765682220459,
|
| 167 |
+
"logps/chosen": -508.98223876953125,
|
| 168 |
+
"logps/rejected": -693.7402954101562,
|
| 169 |
+
"loss": 0.3795,
|
| 170 |
+
"rewards/accuracies": 0.8218749761581421,
|
| 171 |
+
"rewards/chosen": -1.520322322845459,
|
| 172 |
+
"rewards/margins": 1.8594118356704712,
|
| 173 |
+
"rewards/rejected": -3.3797340393066406,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.15267175572519084,
|
| 178 |
+
"grad_norm": 15.404464079078549,
|
| 179 |
+
"learning_rate": 4.957694362057149e-07,
|
| 180 |
+
"logits/chosen": -2.0857653617858887,
|
| 181 |
+
"logits/rejected": -1.8319637775421143,
|
| 182 |
+
"logps/chosen": -497.1195373535156,
|
| 183 |
+
"logps/rejected": -708.096923828125,
|
| 184 |
+
"loss": 0.3493,
|
| 185 |
+
"rewards/accuracies": 0.793749988079071,
|
| 186 |
+
"rewards/chosen": -1.465522289276123,
|
| 187 |
+
"rewards/margins": 2.147498369216919,
|
| 188 |
+
"rewards/rejected": -3.613020658493042,
|
| 189 |
+
"step": 110
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.16655100624566274,
|
| 193 |
+
"grad_norm": 19.947763315181913,
|
| 194 |
+
"learning_rate": 4.932612176449559e-07,
|
| 195 |
+
"logits/chosen": -1.201900839805603,
|
| 196 |
+
"logits/rejected": -0.7779704928398132,
|
| 197 |
+
"logps/chosen": -505.0912170410156,
|
| 198 |
+
"logps/rejected": -777.6538696289062,
|
| 199 |
+
"loss": 0.3487,
|
| 200 |
+
"rewards/accuracies": 0.8218749761581421,
|
| 201 |
+
"rewards/chosen": -1.800119400024414,
|
| 202 |
+
"rewards/margins": 2.440018892288208,
|
| 203 |
+
"rewards/rejected": -4.240138053894043,
|
| 204 |
+
"step": 120
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.18043025676613464,
|
| 208 |
+
"grad_norm": 14.14456268527709,
|
| 209 |
+
"learning_rate": 4.901813394291801e-07,
|
| 210 |
+
"logits/chosen": -0.6655147075653076,
|
| 211 |
+
"logits/rejected": -0.07147368043661118,
|
| 212 |
+
"logps/chosen": -448.736572265625,
|
| 213 |
+
"logps/rejected": -710.1648559570312,
|
| 214 |
+
"loss": 0.332,
|
| 215 |
+
"rewards/accuracies": 0.862500011920929,
|
| 216 |
+
"rewards/chosen": -1.2649247646331787,
|
| 217 |
+
"rewards/margins": 2.5081429481506348,
|
| 218 |
+
"rewards/rejected": -3.7730679512023926,
|
| 219 |
+
"step": 130
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.19430950728660654,
|
| 223 |
+
"grad_norm": 14.401041246804821,
|
| 224 |
+
"learning_rate": 4.865370392189376e-07,
|
| 225 |
+
"logits/chosen": 0.013273833319544792,
|
| 226 |
+
"logits/rejected": 0.503928542137146,
|
| 227 |
+
"logps/chosen": -503.90087890625,
|
| 228 |
+
"logps/rejected": -738.6141357421875,
|
| 229 |
+
"loss": 0.3286,
|
| 230 |
+
"rewards/accuracies": 0.824999988079071,
|
| 231 |
+
"rewards/chosen": -1.7297884225845337,
|
| 232 |
+
"rewards/margins": 2.2531497478485107,
|
| 233 |
+
"rewards/rejected": -3.982938051223755,
|
| 234 |
+
"step": 140
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.2081887578070784,
|
| 238 |
+
"grad_norm": 17.814943595433935,
|
| 239 |
+
"learning_rate": 4.823368810567056e-07,
|
| 240 |
+
"logits/chosen": 0.13247263431549072,
|
| 241 |
+
"logits/rejected": 0.7856736183166504,
|
| 242 |
+
"logps/chosen": -530.7867431640625,
|
| 243 |
+
"logps/rejected": -812.2960205078125,
|
| 244 |
+
"loss": 0.3307,
|
| 245 |
+
"rewards/accuracies": 0.8531249761581421,
|
| 246 |
+
"rewards/chosen": -1.9312444925308228,
|
| 247 |
+
"rewards/margins": 2.786510705947876,
|
| 248 |
+
"rewards/rejected": -4.71775484085083,
|
| 249 |
+
"step": 150
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.2220680083275503,
|
| 253 |
+
"grad_norm": 26.180081364951256,
|
| 254 |
+
"learning_rate": 4.775907352415367e-07,
|
| 255 |
+
"logits/chosen": -0.29394057393074036,
|
| 256 |
+
"logits/rejected": 0.6625006794929504,
|
| 257 |
+
"logps/chosen": -522.6262817382812,
|
| 258 |
+
"logps/rejected": -792.3927612304688,
|
| 259 |
+
"loss": 0.3259,
|
| 260 |
+
"rewards/accuracies": 0.8343750238418579,
|
| 261 |
+
"rewards/chosen": -1.8135877847671509,
|
| 262 |
+
"rewards/margins": 2.6192572116851807,
|
| 263 |
+
"rewards/rejected": -4.432845115661621,
|
| 264 |
+
"step": 160
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.2359472588480222,
|
| 268 |
+
"grad_norm": 14.51543214840684,
|
| 269 |
+
"learning_rate": 4.723097551340265e-07,
|
| 270 |
+
"logits/chosen": -0.6154422760009766,
|
| 271 |
+
"logits/rejected": 0.7039809226989746,
|
| 272 |
+
"logps/chosen": -582.3150024414062,
|
| 273 |
+
"logps/rejected": -870.4035034179688,
|
| 274 |
+
"loss": 0.327,
|
| 275 |
+
"rewards/accuracies": 0.824999988079071,
|
| 276 |
+
"rewards/chosen": -2.0321500301361084,
|
| 277 |
+
"rewards/margins": 2.800929546356201,
|
| 278 |
+
"rewards/rejected": -4.833079814910889,
|
| 279 |
+
"step": 170
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.2498265093684941,
|
| 283 |
+
"grad_norm": 16.28764148948749,
|
| 284 |
+
"learning_rate": 4.6650635094610966e-07,
|
| 285 |
+
"logits/chosen": -0.5453392267227173,
|
| 286 |
+
"logits/rejected": 1.0312373638153076,
|
| 287 |
+
"logps/chosen": -532.6235961914062,
|
| 288 |
+
"logps/rejected": -847.57666015625,
|
| 289 |
+
"loss": 0.3092,
|
| 290 |
+
"rewards/accuracies": 0.856249988079071,
|
| 291 |
+
"rewards/chosen": -1.9564176797866821,
|
| 292 |
+
"rewards/margins": 3.06388783454895,
|
| 293 |
+
"rewards/rejected": -5.020305633544922,
|
| 294 |
+
"step": 180
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.263705759888966,
|
| 298 |
+
"grad_norm": 16.022768213210764,
|
| 299 |
+
"learning_rate": 4.6019416057727577e-07,
|
| 300 |
+
"logits/chosen": -0.6275483965873718,
|
| 301 |
+
"logits/rejected": 0.9134941101074219,
|
| 302 |
+
"logps/chosen": -546.0833129882812,
|
| 303 |
+
"logps/rejected": -847.2735595703125,
|
| 304 |
+
"loss": 0.3035,
|
| 305 |
+
"rewards/accuracies": 0.871874988079071,
|
| 306 |
+
"rewards/chosen": -1.9999825954437256,
|
| 307 |
+
"rewards/margins": 2.9674413204193115,
|
| 308 |
+
"rewards/rejected": -4.967423915863037,
|
| 309 |
+
"step": 190
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.2775850104094379,
|
| 313 |
+
"grad_norm": 17.512075327743087,
|
| 314 |
+
"learning_rate": 4.5338801756574185e-07,
|
| 315 |
+
"logits/chosen": -0.4514436721801758,
|
| 316 |
+
"logits/rejected": 1.0739384889602661,
|
| 317 |
+
"logps/chosen": -560.406494140625,
|
| 318 |
+
"logps/rejected": -897.5866088867188,
|
| 319 |
+
"loss": 0.306,
|
| 320 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 321 |
+
"rewards/chosen": -2.1923580169677734,
|
| 322 |
+
"rewards/margins": 3.19802188873291,
|
| 323 |
+
"rewards/rejected": -5.390379905700684,
|
| 324 |
+
"step": 200
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.2914642609299098,
|
| 328 |
+
"grad_norm": 20.682964186544258,
|
| 329 |
+
"learning_rate": 4.461039162298939e-07,
|
| 330 |
+
"logits/chosen": 0.5358384847640991,
|
| 331 |
+
"logits/rejected": 2.0921194553375244,
|
| 332 |
+
"logps/chosen": -565.9454956054688,
|
| 333 |
+
"logps/rejected": -872.009765625,
|
| 334 |
+
"loss": 0.3124,
|
| 335 |
+
"rewards/accuracies": 0.8687499761581421,
|
| 336 |
+
"rewards/chosen": -2.333399534225464,
|
| 337 |
+
"rewards/margins": 3.0657191276550293,
|
| 338 |
+
"rewards/rejected": -5.399118900299072,
|
| 339 |
+
"step": 210
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.3053435114503817,
|
| 343 |
+
"grad_norm": 20.51613275970851,
|
| 344 |
+
"learning_rate": 4.3835897408191513e-07,
|
| 345 |
+
"logits/chosen": 0.03184742480516434,
|
| 346 |
+
"logits/rejected": 1.515594244003296,
|
| 347 |
+
"logps/chosen": -496.56744384765625,
|
| 348 |
+
"logps/rejected": -793.4273071289062,
|
| 349 |
+
"loss": 0.2886,
|
| 350 |
+
"rewards/accuracies": 0.8343750238418579,
|
| 351 |
+
"rewards/chosen": -1.7265708446502686,
|
| 352 |
+
"rewards/margins": 2.774601459503174,
|
| 353 |
+
"rewards/rejected": -4.501172065734863,
|
| 354 |
+
"step": 220
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.3192227619708536,
|
| 358 |
+
"grad_norm": 15.952868528864848,
|
| 359 |
+
"learning_rate": 4.301713916019286e-07,
|
| 360 |
+
"logits/chosen": -0.11686134338378906,
|
| 361 |
+
"logits/rejected": 1.3415229320526123,
|
| 362 |
+
"logps/chosen": -533.4548950195312,
|
| 363 |
+
"logps/rejected": -846.3487548828125,
|
| 364 |
+
"loss": 0.2971,
|
| 365 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 366 |
+
"rewards/chosen": -1.9538986682891846,
|
| 367 |
+
"rewards/margins": 3.150516986846924,
|
| 368 |
+
"rewards/rejected": -5.104415416717529,
|
| 369 |
+
"step": 230
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.3331020124913255,
|
| 373 |
+
"grad_norm": 22.620955439230894,
|
| 374 |
+
"learning_rate": 4.2156040946718343e-07,
|
| 375 |
+
"logits/chosen": 0.23749442398548126,
|
| 376 |
+
"logits/rejected": 1.675614595413208,
|
| 377 |
+
"logps/chosen": -543.6336059570312,
|
| 378 |
+
"logps/rejected": -936.5950317382812,
|
| 379 |
+
"loss": 0.2868,
|
| 380 |
+
"rewards/accuracies": 0.875,
|
| 381 |
+
"rewards/chosen": -2.0718159675598145,
|
| 382 |
+
"rewards/margins": 3.8275322914123535,
|
| 383 |
+
"rewards/rejected": -5.899348258972168,
|
| 384 |
+
"step": 240
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.3469812630117974,
|
| 388 |
+
"grad_norm": 20.46893017144138,
|
| 389 |
+
"learning_rate": 4.125462633367959e-07,
|
| 390 |
+
"logits/chosen": 0.0811614841222763,
|
| 391 |
+
"logits/rejected": 1.8465839624404907,
|
| 392 |
+
"logps/chosen": -561.4341430664062,
|
| 393 |
+
"logps/rejected": -961.2340087890625,
|
| 394 |
+
"loss": 0.2916,
|
| 395 |
+
"rewards/accuracies": 0.8531249761581421,
|
| 396 |
+
"rewards/chosen": -2.1782994270324707,
|
| 397 |
+
"rewards/margins": 3.972548246383667,
|
| 398 |
+
"rewards/rejected": -6.150847911834717,
|
| 399 |
+
"step": 250
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.3608605135322693,
|
| 403 |
+
"grad_norm": 14.598720710083828,
|
| 404 |
+
"learning_rate": 4.031501362983007e-07,
|
| 405 |
+
"logits/chosen": 0.12005454301834106,
|
| 406 |
+
"logits/rejected": 1.7045762538909912,
|
| 407 |
+
"logps/chosen": -555.3475952148438,
|
| 408 |
+
"logps/rejected": -890.2991333007812,
|
| 409 |
+
"loss": 0.3008,
|
| 410 |
+
"rewards/accuracies": 0.846875011920929,
|
| 411 |
+
"rewards/chosen": -2.2171506881713867,
|
| 412 |
+
"rewards/margins": 3.3155696392059326,
|
| 413 |
+
"rewards/rejected": -5.53272008895874,
|
| 414 |
+
"step": 260
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.3747397640527412,
|
| 418 |
+
"grad_norm": 16.026833129762366,
|
| 419 |
+
"learning_rate": 3.933941090877615e-07,
|
| 420 |
+
"logits/chosen": 0.3684059679508209,
|
| 421 |
+
"logits/rejected": 1.7908920049667358,
|
| 422 |
+
"logps/chosen": -571.7980346679688,
|
| 423 |
+
"logps/rejected": -938.0518798828125,
|
| 424 |
+
"loss": 0.2825,
|
| 425 |
+
"rewards/accuracies": 0.84375,
|
| 426 |
+
"rewards/chosen": -2.356633424758911,
|
| 427 |
+
"rewards/margins": 3.5530152320861816,
|
| 428 |
+
"rewards/rejected": -5.9096479415893555,
|
| 429 |
+
"step": 270
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.3886190145732131,
|
| 433 |
+
"grad_norm": 17.23246443508795,
|
| 434 |
+
"learning_rate": 3.833011082004228e-07,
|
| 435 |
+
"logits/chosen": 0.3943202495574951,
|
| 436 |
+
"logits/rejected": 2.144474506378174,
|
| 437 |
+
"logps/chosen": -606.2532958984375,
|
| 438 |
+
"logps/rejected": -1000.9134521484375,
|
| 439 |
+
"loss": 0.2855,
|
| 440 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 441 |
+
"rewards/chosen": -2.7293572425842285,
|
| 442 |
+
"rewards/margins": 3.9710164070129395,
|
| 443 |
+
"rewards/rejected": -6.70037317276001,
|
| 444 |
+
"step": 280
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.4024982650936849,
|
| 448 |
+
"grad_norm": 14.280408221009226,
|
| 449 |
+
"learning_rate": 3.728948520138426e-07,
|
| 450 |
+
"logits/chosen": 0.1209539920091629,
|
| 451 |
+
"logits/rejected": 1.5210872888565063,
|
| 452 |
+
"logps/chosen": -600.26611328125,
|
| 453 |
+
"logps/rejected": -961.9885864257812,
|
| 454 |
+
"loss": 0.3001,
|
| 455 |
+
"rewards/accuracies": 0.8656250238418579,
|
| 456 |
+
"rewards/chosen": -2.5707249641418457,
|
| 457 |
+
"rewards/margins": 3.2284767627716064,
|
| 458 |
+
"rewards/rejected": -5.799201965332031,
|
| 459 |
+
"step": 290
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.4163775156141568,
|
| 463 |
+
"grad_norm": 14.385360493039851,
|
| 464 |
+
"learning_rate": 3.6219979505011555e-07,
|
| 465 |
+
"logits/chosen": -0.4804894030094147,
|
| 466 |
+
"logits/rejected": 1.5072767734527588,
|
| 467 |
+
"logps/chosen": -552.1934814453125,
|
| 468 |
+
"logps/rejected": -922.3473510742188,
|
| 469 |
+
"loss": 0.2839,
|
| 470 |
+
"rewards/accuracies": 0.856249988079071,
|
| 471 |
+
"rewards/chosen": -2.130495071411133,
|
| 472 |
+
"rewards/margins": 3.7109711170196533,
|
| 473 |
+
"rewards/rejected": -5.841465950012207,
|
| 474 |
+
"step": 300
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.4302567661346287,
|
| 478 |
+
"grad_norm": 17.33261427588973,
|
| 479 |
+
"learning_rate": 3.512410705081684e-07,
|
| 480 |
+
"logits/chosen": 0.34214338660240173,
|
| 481 |
+
"logits/rejected": 2.0437850952148438,
|
| 482 |
+
"logps/chosen": -627.4119262695312,
|
| 483 |
+
"logps/rejected": -1092.1920166015625,
|
| 484 |
+
"loss": 0.2926,
|
| 485 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 486 |
+
"rewards/chosen": -2.9209723472595215,
|
| 487 |
+
"rewards/margins": 4.4133782386779785,
|
| 488 |
+
"rewards/rejected": -7.3343505859375,
|
| 489 |
+
"step": 310
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.4441360166551006,
|
| 493 |
+
"grad_norm": 14.401848583946974,
|
| 494 |
+
"learning_rate": 3.400444312011776e-07,
|
| 495 |
+
"logits/chosen": 0.3555607795715332,
|
| 496 |
+
"logits/rejected": 2.2223145961761475,
|
| 497 |
+
"logps/chosen": -568.3477783203125,
|
| 498 |
+
"logps/rejected": -927.1591796875,
|
| 499 |
+
"loss": 0.2822,
|
| 500 |
+
"rewards/accuracies": 0.8531249761581421,
|
| 501 |
+
"rewards/chosen": -2.276163101196289,
|
| 502 |
+
"rewards/margins": 3.52215313911438,
|
| 503 |
+
"rewards/rejected": -5.79831600189209,
|
| 504 |
+
"step": 320
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 0.4580152671755725,
|
| 508 |
+
"grad_norm": 18.609028672434732,
|
| 509 |
+
"learning_rate": 3.286361890379034e-07,
|
| 510 |
+
"logits/chosen": 0.6069589853286743,
|
| 511 |
+
"logits/rejected": 2.4949374198913574,
|
| 512 |
+
"logps/chosen": -575.1094970703125,
|
| 513 |
+
"logps/rejected": -997.7161865234375,
|
| 514 |
+
"loss": 0.2857,
|
| 515 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 516 |
+
"rewards/chosen": -2.511615753173828,
|
| 517 |
+
"rewards/margins": 4.006805896759033,
|
| 518 |
+
"rewards/rejected": -6.518421173095703,
|
| 519 |
+
"step": 330
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.4718945176960444,
|
| 523 |
+
"grad_norm": 18.17335174994865,
|
| 524 |
+
"learning_rate": 3.1704315319015936e-07,
|
| 525 |
+
"logits/chosen": 0.4636520445346832,
|
| 526 |
+
"logits/rejected": 2.732633113861084,
|
| 527 |
+
"logps/chosen": -559.9303588867188,
|
| 528 |
+
"logps/rejected": -1051.6278076171875,
|
| 529 |
+
"loss": 0.2647,
|
| 530 |
+
"rewards/accuracies": 0.8968750238418579,
|
| 531 |
+
"rewards/chosen": -2.1130199432373047,
|
| 532 |
+
"rewards/margins": 4.71722936630249,
|
| 533 |
+
"rewards/rejected": -6.830248832702637,
|
| 534 |
+
"step": 340
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.4857737682165163,
|
| 538 |
+
"grad_norm": 21.753884723720933,
|
| 539 |
+
"learning_rate": 3.052925670917219e-07,
|
| 540 |
+
"logits/chosen": 1.2968404293060303,
|
| 541 |
+
"logits/rejected": 3.403611421585083,
|
| 542 |
+
"logps/chosen": -615.7195434570312,
|
| 543 |
+
"logps/rejected": -1081.2171630859375,
|
| 544 |
+
"loss": 0.2829,
|
| 545 |
+
"rewards/accuracies": 0.828125,
|
| 546 |
+
"rewards/chosen": -2.809459686279297,
|
| 547 |
+
"rewards/margins": 4.458012104034424,
|
| 548 |
+
"rewards/rejected": -7.267472267150879,
|
| 549 |
+
"step": 350
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.4996530187369882,
|
| 553 |
+
"grad_norm": 19.099053383072434,
|
| 554 |
+
"learning_rate": 2.934120444167326e-07,
|
| 555 |
+
"logits/chosen": 1.4352385997772217,
|
| 556 |
+
"logits/rejected": 3.273483991622925,
|
| 557 |
+
"logps/chosen": -581.668701171875,
|
| 558 |
+
"logps/rejected": -1042.809814453125,
|
| 559 |
+
"loss": 0.2704,
|
| 560 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 561 |
+
"rewards/chosen": -2.5318121910095215,
|
| 562 |
+
"rewards/margins": 4.3836283683776855,
|
| 563 |
+
"rewards/rejected": -6.915441036224365,
|
| 564 |
+
"step": 360
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.5135322692574601,
|
| 568 |
+
"grad_norm": 24.65293117421233,
|
| 569 |
+
"learning_rate": 2.814295041880407e-07,
|
| 570 |
+
"logits/chosen": 1.0912240743637085,
|
| 571 |
+
"logits/rejected": 3.1343836784362793,
|
| 572 |
+
"logps/chosen": -557.176513671875,
|
| 573 |
+
"logps/rejected": -982.60107421875,
|
| 574 |
+
"loss": 0.2701,
|
| 575 |
+
"rewards/accuracies": 0.890625,
|
| 576 |
+
"rewards/chosen": -2.306490898132324,
|
| 577 |
+
"rewards/margins": 4.240821361541748,
|
| 578 |
+
"rewards/rejected": -6.5473127365112305,
|
| 579 |
+
"step": 370
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.527411519777932,
|
| 583 |
+
"grad_norm": 16.00175200334935,
|
| 584 |
+
"learning_rate": 2.6937310516798275e-07,
|
| 585 |
+
"logits/chosen": 0.5683895349502563,
|
| 586 |
+
"logits/rejected": 2.84633469581604,
|
| 587 |
+
"logps/chosen": -578.7667236328125,
|
| 588 |
+
"logps/rejected": -935.6932373046875,
|
| 589 |
+
"loss": 0.2745,
|
| 590 |
+
"rewards/accuracies": 0.8125,
|
| 591 |
+
"rewards/chosen": -2.3604445457458496,
|
| 592 |
+
"rewards/margins": 3.500983476638794,
|
| 593 |
+
"rewards/rejected": -5.861428260803223,
|
| 594 |
+
"step": 380
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 0.5412907702984039,
|
| 598 |
+
"grad_norm": 20.233593507922254,
|
| 599 |
+
"learning_rate": 2.5727117968577785e-07,
|
| 600 |
+
"logits/chosen": 0.8911641240119934,
|
| 601 |
+
"logits/rejected": 3.5523343086242676,
|
| 602 |
+
"logps/chosen": -587.8340454101562,
|
| 603 |
+
"logps/rejected": -973.2986450195312,
|
| 604 |
+
"loss": 0.2781,
|
| 605 |
+
"rewards/accuracies": 0.875,
|
| 606 |
+
"rewards/chosen": -2.5025112628936768,
|
| 607 |
+
"rewards/margins": 3.980858325958252,
|
| 608 |
+
"rewards/rejected": -6.483368873596191,
|
| 609 |
+
"step": 390
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.5551700208188758,
|
| 613 |
+
"grad_norm": 20.17598650271236,
|
| 614 |
+
"learning_rate": 2.4515216705704393e-07,
|
| 615 |
+
"logits/chosen": 1.0148788690567017,
|
| 616 |
+
"logits/rejected": 3.686418056488037,
|
| 617 |
+
"logps/chosen": -623.1129760742188,
|
| 618 |
+
"logps/rejected": -1018.2747802734375,
|
| 619 |
+
"loss": 0.2823,
|
| 620 |
+
"rewards/accuracies": 0.862500011920929,
|
| 621 |
+
"rewards/chosen": -2.7866435050964355,
|
| 622 |
+
"rewards/margins": 4.11606502532959,
|
| 623 |
+
"rewards/rejected": -6.902708530426025,
|
| 624 |
+
"step": 400
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 0.5690492713393477,
|
| 628 |
+
"grad_norm": 15.785292476024692,
|
| 629 |
+
"learning_rate": 2.330445467518977e-07,
|
| 630 |
+
"logits/chosen": 0.4552311897277832,
|
| 631 |
+
"logits/rejected": 2.6864824295043945,
|
| 632 |
+
"logps/chosen": -626.0252685546875,
|
| 633 |
+
"logps/rejected": -1022.4189453125,
|
| 634 |
+
"loss": 0.2698,
|
| 635 |
+
"rewards/accuracies": 0.8531249761581421,
|
| 636 |
+
"rewards/chosen": -2.769418239593506,
|
| 637 |
+
"rewards/margins": 3.919069290161133,
|
| 638 |
+
"rewards/rejected": -6.6884870529174805,
|
| 639 |
+
"step": 410
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.5829285218598196,
|
| 643 |
+
"grad_norm": 24.14459841167646,
|
| 644 |
+
"learning_rate": 2.209767714686924e-07,
|
| 645 |
+
"logits/chosen": 1.0538936853408813,
|
| 646 |
+
"logits/rejected": 3.6474509239196777,
|
| 647 |
+
"logps/chosen": -675.6682739257812,
|
| 648 |
+
"logps/rejected": -1114.697021484375,
|
| 649 |
+
"loss": 0.2865,
|
| 650 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 651 |
+
"rewards/chosen": -3.275700092315674,
|
| 652 |
+
"rewards/margins": 4.374134063720703,
|
| 653 |
+
"rewards/rejected": -7.649834632873535,
|
| 654 |
+
"step": 420
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.5968077723802915,
|
| 658 |
+
"grad_norm": 15.264745793489116,
|
| 659 |
+
"learning_rate": 2.0897720027066897e-07,
|
| 660 |
+
"logits/chosen": 1.596185326576233,
|
| 661 |
+
"logits/rejected": 4.331744194030762,
|
| 662 |
+
"logps/chosen": -679.60888671875,
|
| 663 |
+
"logps/rejected": -1198.2357177734375,
|
| 664 |
+
"loss": 0.277,
|
| 665 |
+
"rewards/accuracies": 0.890625,
|
| 666 |
+
"rewards/chosen": -3.4800593852996826,
|
| 667 |
+
"rewards/margins": 5.010531902313232,
|
| 668 |
+
"rewards/rejected": -8.490591049194336,
|
| 669 |
+
"step": 430
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.6106870229007634,
|
| 673 |
+
"grad_norm": 19.482670778404138,
|
| 674 |
+
"learning_rate": 1.970740319426474e-07,
|
| 675 |
+
"logits/chosen": 2.1650781631469727,
|
| 676 |
+
"logits/rejected": 4.562286853790283,
|
| 677 |
+
"logps/chosen": -713.8223266601562,
|
| 678 |
+
"logps/rejected": -1141.03369140625,
|
| 679 |
+
"loss": 0.2843,
|
| 680 |
+
"rewards/accuracies": 0.8656250238418579,
|
| 681 |
+
"rewards/chosen": -3.699204683303833,
|
| 682 |
+
"rewards/margins": 4.300296783447266,
|
| 683 |
+
"rewards/rejected": -7.9995012283325195,
|
| 684 |
+
"step": 440
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 0.6245662734212353,
|
| 688 |
+
"grad_norm": 16.787688600188858,
|
| 689 |
+
"learning_rate": 1.8529523872436977e-07,
|
| 690 |
+
"logits/chosen": 3.045531749725342,
|
| 691 |
+
"logits/rejected": 4.9512128829956055,
|
| 692 |
+
"logps/chosen": -796.2913818359375,
|
| 693 |
+
"logps/rejected": -1225.3551025390625,
|
| 694 |
+
"loss": 0.2792,
|
| 695 |
+
"rewards/accuracies": 0.828125,
|
| 696 |
+
"rewards/chosen": -4.586967468261719,
|
| 697 |
+
"rewards/margins": 4.1703386306762695,
|
| 698 |
+
"rewards/rejected": -8.757307052612305,
|
| 699 |
+
"step": 450
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.6384455239417072,
|
| 703 |
+
"grad_norm": 16.212273972183578,
|
| 704 |
+
"learning_rate": 1.7366850057622172e-07,
|
| 705 |
+
"logits/chosen": 2.1972670555114746,
|
| 706 |
+
"logits/rejected": 3.9832711219787598,
|
| 707 |
+
"logps/chosen": -759.0075073242188,
|
| 708 |
+
"logps/rejected": -1123.007080078125,
|
| 709 |
+
"loss": 0.2771,
|
| 710 |
+
"rewards/accuracies": 0.828125,
|
| 711 |
+
"rewards/chosen": -4.106420040130615,
|
| 712 |
+
"rewards/margins": 3.6798558235168457,
|
| 713 |
+
"rewards/rejected": -7.786275386810303,
|
| 714 |
+
"step": 460
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 0.6523247744621791,
|
| 718 |
+
"grad_norm": 18.354213726936635,
|
| 719 |
+
"learning_rate": 1.622211401318028e-07,
|
| 720 |
+
"logits/chosen": 1.4993724822998047,
|
| 721 |
+
"logits/rejected": 4.176512241363525,
|
| 722 |
+
"logps/chosen": -691.9547119140625,
|
| 723 |
+
"logps/rejected": -1159.3623046875,
|
| 724 |
+
"loss": 0.2635,
|
| 725 |
+
"rewards/accuracies": 0.856249988079071,
|
| 726 |
+
"rewards/chosen": -3.5886549949645996,
|
| 727 |
+
"rewards/margins": 4.674525260925293,
|
| 728 |
+
"rewards/rejected": -8.26318073272705,
|
| 729 |
+
"step": 470
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.666204024982651,
|
| 733 |
+
"grad_norm": 17.687786590391063,
|
| 734 |
+
"learning_rate": 1.5098005849021078e-07,
|
| 735 |
+
"logits/chosen": 1.456290602684021,
|
| 736 |
+
"logits/rejected": 4.025675296783447,
|
| 737 |
+
"logps/chosen": -683.2501220703125,
|
| 738 |
+
"logps/rejected": -1227.657470703125,
|
| 739 |
+
"loss": 0.2727,
|
| 740 |
+
"rewards/accuracies": 0.903124988079071,
|
| 741 |
+
"rewards/chosen": -3.4943652153015137,
|
| 742 |
+
"rewards/margins": 5.14384651184082,
|
| 743 |
+
"rewards/rejected": -8.638211250305176,
|
| 744 |
+
"step": 480
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.6800832755031229,
|
| 748 |
+
"grad_norm": 15.302939524453814,
|
| 749 |
+
"learning_rate": 1.3997167199892385e-07,
|
| 750 |
+
"logits/chosen": 0.9439892768859863,
|
| 751 |
+
"logits/rejected": 3.5333430767059326,
|
| 752 |
+
"logps/chosen": -721.109375,
|
| 753 |
+
"logps/rejected": -1133.033203125,
|
| 754 |
+
"loss": 0.2586,
|
| 755 |
+
"rewards/accuracies": 0.862500011920929,
|
| 756 |
+
"rewards/chosen": -3.779620409011841,
|
| 757 |
+
"rewards/margins": 4.077832221984863,
|
| 758 |
+
"rewards/rejected": -7.857451438903809,
|
| 759 |
+
"step": 490
|
| 760 |
+
},
|
| 761 |
+
{
|
| 762 |
+
"epoch": 0.6939625260235948,
|
| 763 |
+
"grad_norm": 20.366996379564934,
|
| 764 |
+
"learning_rate": 1.2922185017584036e-07,
|
| 765 |
+
"logits/chosen": 1.6919466257095337,
|
| 766 |
+
"logits/rejected": 4.416836261749268,
|
| 767 |
+
"logps/chosen": -755.7452392578125,
|
| 768 |
+
"logps/rejected": -1292.0657958984375,
|
| 769 |
+
"loss": 0.2637,
|
| 770 |
+
"rewards/accuracies": 0.890625,
|
| 771 |
+
"rewards/chosen": -4.0984649658203125,
|
| 772 |
+
"rewards/margins": 5.336091041564941,
|
| 773 |
+
"rewards/rejected": -9.434555053710938,
|
| 774 |
+
"step": 500
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 0.7078417765440667,
|
| 778 |
+
"grad_norm": 19.07449137750698,
|
| 779 |
+
"learning_rate": 1.1875585491635998e-07,
|
| 780 |
+
"logits/chosen": 1.5969302654266357,
|
| 781 |
+
"logits/rejected": 3.998643398284912,
|
| 782 |
+
"logps/chosen": -720.2608032226562,
|
| 783 |
+
"logps/rejected": -1152.5308837890625,
|
| 784 |
+
"loss": 0.2657,
|
| 785 |
+
"rewards/accuracies": 0.824999988079071,
|
| 786 |
+
"rewards/chosen": -3.8675122261047363,
|
| 787 |
+
"rewards/margins": 4.360492706298828,
|
| 788 |
+
"rewards/rejected": -8.228004455566406,
|
| 789 |
+
"step": 510
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 0.7217210270645386,
|
| 793 |
+
"grad_norm": 20.85771539055888,
|
| 794 |
+
"learning_rate": 1.0859828112836539e-07,
|
| 795 |
+
"logits/chosen": 1.2433149814605713,
|
| 796 |
+
"logits/rejected": 4.097987174987793,
|
| 797 |
+
"logps/chosen": -671.2435913085938,
|
| 798 |
+
"logps/rejected": -1159.519287109375,
|
| 799 |
+
"loss": 0.2748,
|
| 800 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 801 |
+
"rewards/chosen": -3.430670976638794,
|
| 802 |
+
"rewards/margins": 4.710363388061523,
|
| 803 |
+
"rewards/rejected": -8.141035079956055,
|
| 804 |
+
"step": 520
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 0.7356002775850105,
|
| 808 |
+
"grad_norm": 16.162845971874738,
|
| 809 |
+
"learning_rate": 9.877299893461455e-08,
|
| 810 |
+
"logits/chosen": 1.8318910598754883,
|
| 811 |
+
"logits/rejected": 4.480563163757324,
|
| 812 |
+
"logps/chosen": -694.7840576171875,
|
| 813 |
+
"logps/rejected": -1228.274658203125,
|
| 814 |
+
"loss": 0.2565,
|
| 815 |
+
"rewards/accuracies": 0.878125011920929,
|
| 816 |
+
"rewards/chosen": -3.6290040016174316,
|
| 817 |
+
"rewards/margins": 5.196530818939209,
|
| 818 |
+
"rewards/rejected": -8.82553482055664,
|
| 819 |
+
"step": 530
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 0.7494795281054824,
|
| 823 |
+
"grad_norm": 22.751190521542522,
|
| 824 |
+
"learning_rate": 8.930309757836516e-08,
|
| 825 |
+
"logits/chosen": 1.7889764308929443,
|
| 826 |
+
"logits/rejected": 4.209009647369385,
|
| 827 |
+
"logps/chosen": -711.76123046875,
|
| 828 |
+
"logps/rejected": -1204.97265625,
|
| 829 |
+
"loss": 0.2757,
|
| 830 |
+
"rewards/accuracies": 0.871874988079071,
|
| 831 |
+
"rewards/chosen": -3.8144354820251465,
|
| 832 |
+
"rewards/margins": 4.804244041442871,
|
| 833 |
+
"rewards/rejected": -8.618680000305176,
|
| 834 |
+
"step": 540
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 0.7633587786259542,
|
| 838 |
+
"grad_norm": 18.413955181298185,
|
| 839 |
+
"learning_rate": 8.021083116405173e-08,
|
| 840 |
+
"logits/chosen": 1.8144102096557617,
|
| 841 |
+
"logits/rejected": 4.302700519561768,
|
| 842 |
+
"logps/chosen": -712.6237182617188,
|
| 843 |
+
"logps/rejected": -1180.5711669921875,
|
| 844 |
+
"loss": 0.2783,
|
| 845 |
+
"rewards/accuracies": 0.856249988079071,
|
| 846 |
+
"rewards/chosen": -3.794156312942505,
|
| 847 |
+
"rewards/margins": 4.554933071136475,
|
| 848 |
+
"rewards/rejected": -8.349088668823242,
|
| 849 |
+
"step": 550
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.7772380291464261,
|
| 853 |
+
"grad_norm": 15.840889959483494,
|
| 854 |
+
"learning_rate": 7.151756636052527e-08,
|
| 855 |
+
"logits/chosen": 1.8780730962753296,
|
| 856 |
+
"logits/rejected": 3.9335429668426514,
|
| 857 |
+
"logps/chosen": -710.3377075195312,
|
| 858 |
+
"logps/rejected": -1109.950439453125,
|
| 859 |
+
"loss": 0.2613,
|
| 860 |
+
"rewards/accuracies": 0.84375,
|
| 861 |
+
"rewards/chosen": -3.773430585861206,
|
| 862 |
+
"rewards/margins": 3.8713042736053467,
|
| 863 |
+
"rewards/rejected": -7.6447343826293945,
|
| 864 |
+
"step": 560
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.7911172796668979,
|
| 868 |
+
"grad_norm": 17.548413688482757,
|
| 869 |
+
"learning_rate": 6.324373218975104e-08,
|
| 870 |
+
"logits/chosen": 1.9608227014541626,
|
| 871 |
+
"logits/rejected": 4.276215076446533,
|
| 872 |
+
"logps/chosen": -716.494140625,
|
| 873 |
+
"logps/rejected": -1132.06298828125,
|
| 874 |
+
"loss": 0.2786,
|
| 875 |
+
"rewards/accuracies": 0.862500011920929,
|
| 876 |
+
"rewards/chosen": -3.817418336868286,
|
| 877 |
+
"rewards/margins": 4.1049346923828125,
|
| 878 |
+
"rewards/rejected": -7.9223527908325195,
|
| 879 |
+
"step": 570
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 0.8049965301873698,
|
| 883 |
+
"grad_norm": 15.393822405492056,
|
| 884 |
+
"learning_rate": 5.5408772018959996e-08,
|
| 885 |
+
"logits/chosen": 1.5480540990829468,
|
| 886 |
+
"logits/rejected": 4.223781585693359,
|
| 887 |
+
"logps/chosen": -728.512451171875,
|
| 888 |
+
"logps/rejected": -1185.4339599609375,
|
| 889 |
+
"loss": 0.2625,
|
| 890 |
+
"rewards/accuracies": 0.875,
|
| 891 |
+
"rewards/chosen": -3.6450915336608887,
|
| 892 |
+
"rewards/margins": 4.617922306060791,
|
| 893 |
+
"rewards/rejected": -8.26301383972168,
|
| 894 |
+
"step": 580
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 0.8188757807078417,
|
| 898 |
+
"grad_norm": 15.931503102859638,
|
| 899 |
+
"learning_rate": 4.8031097869072225e-08,
|
| 900 |
+
"logits/chosen": 1.2332836389541626,
|
| 901 |
+
"logits/rejected": 3.650308609008789,
|
| 902 |
+
"logps/chosen": -693.335205078125,
|
| 903 |
+
"logps/rejected": -1153.408447265625,
|
| 904 |
+
"loss": 0.2768,
|
| 905 |
+
"rewards/accuracies": 0.859375,
|
| 906 |
+
"rewards/chosen": -3.520477294921875,
|
| 907 |
+
"rewards/margins": 4.526196479797363,
|
| 908 |
+
"rewards/rejected": -8.046672821044922,
|
| 909 |
+
"step": 590
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 0.8327550312283136,
|
| 913 |
+
"grad_norm": 17.151234880694616,
|
| 914 |
+
"learning_rate": 4.112804714676593e-08,
|
| 915 |
+
"logits/chosen": 0.998208224773407,
|
| 916 |
+
"logits/rejected": 3.692138195037842,
|
| 917 |
+
"logps/chosen": -708.9652099609375,
|
| 918 |
+
"logps/rejected": -1158.7760009765625,
|
| 919 |
+
"loss": 0.26,
|
| 920 |
+
"rewards/accuracies": 0.8531249761581421,
|
| 921 |
+
"rewards/chosen": -3.4704201221466064,
|
| 922 |
+
"rewards/margins": 4.559679985046387,
|
| 923 |
+
"rewards/rejected": -8.030099868774414,
|
| 924 |
+
"step": 600
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 0.8466342817487855,
|
| 928 |
+
"grad_norm": 15.410888817871795,
|
| 929 |
+
"learning_rate": 3.4715841901871545e-08,
|
| 930 |
+
"logits/chosen": 1.2765249013900757,
|
| 931 |
+
"logits/rejected": 3.9393792152404785,
|
| 932 |
+
"logps/chosen": -706.2520141601562,
|
| 933 |
+
"logps/rejected": -1225.5408935546875,
|
| 934 |
+
"loss": 0.2575,
|
| 935 |
+
"rewards/accuracies": 0.8656250238418579,
|
| 936 |
+
"rewards/chosen": -3.693631410598755,
|
| 937 |
+
"rewards/margins": 5.080514430999756,
|
| 938 |
+
"rewards/rejected": -8.774145126342773,
|
| 939 |
+
"step": 610
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 0.8605135322692574,
|
| 943 |
+
"grad_norm": 20.44808883705845,
|
| 944 |
+
"learning_rate": 2.8809550705835546e-08,
|
| 945 |
+
"logits/chosen": 1.7312864065170288,
|
| 946 |
+
"logits/rejected": 4.368432521820068,
|
| 947 |
+
"logps/chosen": -755.4667358398438,
|
| 948 |
+
"logps/rejected": -1226.697021484375,
|
| 949 |
+
"loss": 0.2755,
|
| 950 |
+
"rewards/accuracies": 0.856249988079071,
|
| 951 |
+
"rewards/chosen": -4.141783714294434,
|
| 952 |
+
"rewards/margins": 4.687798500061035,
|
| 953 |
+
"rewards/rejected": -8.829582214355469,
|
| 954 |
+
"step": 620
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 0.8743927827897293,
|
| 958 |
+
"grad_norm": 16.644054207216232,
|
| 959 |
+
"learning_rate": 2.3423053240837514e-08,
|
| 960 |
+
"logits/chosen": 1.7737137079238892,
|
| 961 |
+
"logits/rejected": 4.126453399658203,
|
| 962 |
+
"logps/chosen": -743.0369873046875,
|
| 963 |
+
"logps/rejected": -1216.58203125,
|
| 964 |
+
"loss": 0.2624,
|
| 965 |
+
"rewards/accuracies": 0.8843749761581421,
|
| 966 |
+
"rewards/chosen": -3.9477715492248535,
|
| 967 |
+
"rewards/margins": 4.61209774017334,
|
| 968 |
+
"rewards/rejected": -8.559869766235352,
|
| 969 |
+
"step": 630
|
| 970 |
+
},
|
| 971 |
+
{
|
| 972 |
+
"epoch": 0.8882720333102012,
|
| 973 |
+
"grad_norm": 16.30128360187393,
|
| 974 |
+
"learning_rate": 1.8569007682777415e-08,
|
| 975 |
+
"logits/chosen": 1.7834386825561523,
|
| 976 |
+
"logits/rejected": 4.101445198059082,
|
| 977 |
+
"logps/chosen": -724.764892578125,
|
| 978 |
+
"logps/rejected": -1201.727783203125,
|
| 979 |
+
"loss": 0.2596,
|
| 980 |
+
"rewards/accuracies": 0.840624988079071,
|
| 981 |
+
"rewards/chosen": -4.080270290374756,
|
| 982 |
+
"rewards/margins": 4.567948341369629,
|
| 983 |
+
"rewards/rejected": -8.648219108581543,
|
| 984 |
+
"step": 640
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 0.9021512838306731,
|
| 988 |
+
"grad_norm": 17.506230515738004,
|
| 989 |
+
"learning_rate": 1.4258820954781037e-08,
|
| 990 |
+
"logits/chosen": 2.0160067081451416,
|
| 991 |
+
"logits/rejected": 4.396459102630615,
|
| 992 |
+
"logps/chosen": -750.851318359375,
|
| 993 |
+
"logps/rejected": -1249.35546875,
|
| 994 |
+
"loss": 0.2658,
|
| 995 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 996 |
+
"rewards/chosen": -4.158956050872803,
|
| 997 |
+
"rewards/margins": 4.926741600036621,
|
| 998 |
+
"rewards/rejected": -9.085699081420898,
|
| 999 |
+
"step": 650
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 0.916030534351145,
|
| 1003 |
+
"grad_norm": 23.02776428486556,
|
| 1004 |
+
"learning_rate": 1.0502621921127774e-08,
|
| 1005 |
+
"logits/chosen": 1.575297474861145,
|
| 1006 |
+
"logits/rejected": 4.593780994415283,
|
| 1007 |
+
"logps/chosen": -728.802978515625,
|
| 1008 |
+
"logps/rejected": -1211.679931640625,
|
| 1009 |
+
"loss": 0.2587,
|
| 1010 |
+
"rewards/accuracies": 0.878125011920929,
|
| 1011 |
+
"rewards/chosen": -4.0324225425720215,
|
| 1012 |
+
"rewards/margins": 4.782987117767334,
|
| 1013 |
+
"rewards/rejected": -8.815410614013672,
|
| 1014 |
+
"step": 660
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 0.9299097848716169,
|
| 1018 |
+
"grad_norm": 16.504585016350322,
|
| 1019 |
+
"learning_rate": 7.309237584595007e-09,
|
| 1020 |
+
"logits/chosen": 1.5119308233261108,
|
| 1021 |
+
"logits/rejected": 4.147110939025879,
|
| 1022 |
+
"logps/chosen": -738.7492065429688,
|
| 1023 |
+
"logps/rejected": -1241.984619140625,
|
| 1024 |
+
"loss": 0.2632,
|
| 1025 |
+
"rewards/accuracies": 0.890625,
|
| 1026 |
+
"rewards/chosen": -3.941183090209961,
|
| 1027 |
+
"rewards/margins": 4.9969072341918945,
|
| 1028 |
+
"rewards/rejected": -8.938089370727539,
|
| 1029 |
+
"step": 670
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 0.9437890353920888,
|
| 1033 |
+
"grad_norm": 29.62214997576546,
|
| 1034 |
+
"learning_rate": 4.6861723431538265e-09,
|
| 1035 |
+
"logits/chosen": 1.4143844842910767,
|
| 1036 |
+
"logits/rejected": 4.09596586227417,
|
| 1037 |
+
"logps/chosen": -763.4151611328125,
|
| 1038 |
+
"logps/rejected": -1327.39599609375,
|
| 1039 |
+
"loss": 0.2657,
|
| 1040 |
+
"rewards/accuracies": 0.878125011920929,
|
| 1041 |
+
"rewards/chosen": -4.114924430847168,
|
| 1042 |
+
"rewards/margins": 5.448840141296387,
|
| 1043 |
+
"rewards/rejected": -9.563764572143555,
|
| 1044 |
+
"step": 680
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 0.9576682859125607,
|
| 1048 |
+
"grad_norm": 18.191320437540885,
|
| 1049 |
+
"learning_rate": 2.639590354763882e-09,
|
| 1050 |
+
"logits/chosen": 1.5057471990585327,
|
| 1051 |
+
"logits/rejected": 3.9795989990234375,
|
| 1052 |
+
"logps/chosen": -759.0143432617188,
|
| 1053 |
+
"logps/rejected": -1212.9630126953125,
|
| 1054 |
+
"loss": 0.2696,
|
| 1055 |
+
"rewards/accuracies": 0.809374988079071,
|
| 1056 |
+
"rewards/chosen": -4.070717811584473,
|
| 1057 |
+
"rewards/margins": 4.46467399597168,
|
| 1058 |
+
"rewards/rejected": -8.535390853881836,
|
| 1059 |
+
"step": 690
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 0.9715475364330326,
|
| 1063 |
+
"grad_norm": 19.757723334110935,
|
| 1064 |
+
"learning_rate": 1.1743010517085427e-09,
|
| 1065 |
+
"logits/chosen": 1.534629225730896,
|
| 1066 |
+
"logits/rejected": 4.418221473693848,
|
| 1067 |
+
"logps/chosen": -719.3196411132812,
|
| 1068 |
+
"logps/rejected": -1227.99609375,
|
| 1069 |
+
"loss": 0.2589,
|
| 1070 |
+
"rewards/accuracies": 0.8656250238418579,
|
| 1071 |
+
"rewards/chosen": -3.9217464923858643,
|
| 1072 |
+
"rewards/margins": 5.034605979919434,
|
| 1073 |
+
"rewards/rejected": -8.956352233886719,
|
| 1074 |
+
"step": 700
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 0.9854267869535045,
|
| 1078 |
+
"grad_norm": 18.633072832178254,
|
| 1079 |
+
"learning_rate": 2.9374783851240923e-10,
|
| 1080 |
+
"logits/chosen": 1.5875966548919678,
|
| 1081 |
+
"logits/rejected": 3.8790760040283203,
|
| 1082 |
+
"logps/chosen": -764.1870727539062,
|
| 1083 |
+
"logps/rejected": -1278.9044189453125,
|
| 1084 |
+
"loss": 0.2748,
|
| 1085 |
+
"rewards/accuracies": 0.878125011920929,
|
| 1086 |
+
"rewards/chosen": -4.169381618499756,
|
| 1087 |
+
"rewards/margins": 4.928309440612793,
|
| 1088 |
+
"rewards/rejected": -9.097692489624023,
|
| 1089 |
+
"step": 710
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 0.9993060374739764,
|
| 1093 |
+
"grad_norm": 17.747866961702037,
|
| 1094 |
+
"learning_rate": 0.0,
|
| 1095 |
+
"logits/chosen": 1.5639232397079468,
|
| 1096 |
+
"logits/rejected": 3.9590256214141846,
|
| 1097 |
+
"logps/chosen": -738.0526123046875,
|
| 1098 |
+
"logps/rejected": -1246.5972900390625,
|
| 1099 |
+
"loss": 0.2629,
|
| 1100 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 1101 |
+
"rewards/chosen": -3.9611778259277344,
|
| 1102 |
+
"rewards/margins": 4.964480400085449,
|
| 1103 |
+
"rewards/rejected": -8.925657272338867,
|
| 1104 |
+
"step": 720
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 0.9993060374739764,
|
| 1108 |
+
"step": 720,
|
| 1109 |
+
"total_flos": 0.0,
|
| 1110 |
+
"train_loss": 0.32248947852187687,
|
| 1111 |
+
"train_runtime": 20768.7862,
|
| 1112 |
+
"train_samples_per_second": 8.881,
|
| 1113 |
+
"train_steps_per_second": 0.035
|
| 1114 |
+
}
|
| 1115 |
+
],
|
| 1116 |
+
"logging_steps": 10,
|
| 1117 |
+
"max_steps": 720,
|
| 1118 |
+
"num_input_tokens_seen": 0,
|
| 1119 |
+
"num_train_epochs": 1,
|
| 1120 |
+
"save_steps": 100,
|
| 1121 |
+
"stateful_callbacks": {
|
| 1122 |
+
"TrainerControl": {
|
| 1123 |
+
"args": {
|
| 1124 |
+
"should_epoch_stop": false,
|
| 1125 |
+
"should_evaluate": false,
|
| 1126 |
+
"should_log": false,
|
| 1127 |
+
"should_save": true,
|
| 1128 |
+
"should_training_stop": true
|
| 1129 |
+
},
|
| 1130 |
+
"attributes": {}
|
| 1131 |
+
}
|
| 1132 |
+
},
|
| 1133 |
+
"total_flos": 0.0,
|
| 1134 |
+
"train_batch_size": 8,
|
| 1135 |
+
"trial_name": null,
|
| 1136 |
+
"trial_params": null
|
| 1137 |
+
}
|