Commit
•
b13bb8a
1
Parent(s):
131c977
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 +777 -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: tulu-2-7b-rlced-conifer-dpo
|
| 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 |
+
# tulu-2-7b-rlced-conifer-dpo
|
| 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": 0.9984431759211209,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.21509297628908297,
|
| 5 |
+
"train_runtime": 13791.8998,
|
| 6 |
+
"train_samples": 123309,
|
| 7 |
+
"train_samples_per_second": 8.941,
|
| 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": 0.9984431759211209,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.21509297628908297,
|
| 5 |
+
"train_runtime": 13791.8998,
|
| 6 |
+
"train_samples": 123309,
|
| 7 |
+
"train_samples_per_second": 8.941,
|
| 8 |
+
"train_steps_per_second": 0.035
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,777 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9984431759211209,
|
| 5 |
+
"eval_steps": 1000,
|
| 6 |
+
"global_step": 481,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0020757654385054488,
|
| 13 |
+
"grad_norm": 7.453095565847237,
|
| 14 |
+
"learning_rate": 1.020408163265306e-08,
|
| 15 |
+
"logits/chosen": -2.730942726135254,
|
| 16 |
+
"logits/rejected": -2.654609203338623,
|
| 17 |
+
"logps/chosen": -350.489990234375,
|
| 18 |
+
"logps/rejected": -325.546875,
|
| 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": 0.02075765438505449,
|
| 28 |
+
"grad_norm": 7.86433669667092,
|
| 29 |
+
"learning_rate": 1.0204081632653061e-07,
|
| 30 |
+
"logits/chosen": -2.7328553199768066,
|
| 31 |
+
"logits/rejected": -2.734957695007324,
|
| 32 |
+
"logps/chosen": -366.4744567871094,
|
| 33 |
+
"logps/rejected": -412.2707824707031,
|
| 34 |
+
"loss": 0.693,
|
| 35 |
+
"rewards/accuracies": 0.4583333432674408,
|
| 36 |
+
"rewards/chosen": 0.0002039541577687487,
|
| 37 |
+
"rewards/margins": 0.00048357987543568015,
|
| 38 |
+
"rewards/rejected": -0.00027962555759586394,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.04151530877010898,
|
| 43 |
+
"grad_norm": 7.528284552671321,
|
| 44 |
+
"learning_rate": 2.0408163265306121e-07,
|
| 45 |
+
"logits/chosen": -2.717175006866455,
|
| 46 |
+
"logits/rejected": -2.6938576698303223,
|
| 47 |
+
"logps/chosen": -378.77813720703125,
|
| 48 |
+
"logps/rejected": -404.4546813964844,
|
| 49 |
+
"loss": 0.6892,
|
| 50 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 51 |
+
"rewards/chosen": 0.002144110854715109,
|
| 52 |
+
"rewards/margins": 0.007523530628532171,
|
| 53 |
+
"rewards/rejected": -0.00537942023947835,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.062272963155163466,
|
| 58 |
+
"grad_norm": 8.005217747452576,
|
| 59 |
+
"learning_rate": 3.0612244897959183e-07,
|
| 60 |
+
"logits/chosen": -2.7168049812316895,
|
| 61 |
+
"logits/rejected": -2.7009923458099365,
|
| 62 |
+
"logps/chosen": -363.69952392578125,
|
| 63 |
+
"logps/rejected": -390.5137939453125,
|
| 64 |
+
"loss": 0.6692,
|
| 65 |
+
"rewards/accuracies": 0.8812500238418579,
|
| 66 |
+
"rewards/chosen": 0.024431603029370308,
|
| 67 |
+
"rewards/margins": 0.05050837993621826,
|
| 68 |
+
"rewards/rejected": -0.026076778769493103,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.08303061754021795,
|
| 73 |
+
"grad_norm": 9.363164050327798,
|
| 74 |
+
"learning_rate": 4.0816326530612243e-07,
|
| 75 |
+
"logits/chosen": -2.70872163772583,
|
| 76 |
+
"logits/rejected": -2.6695823669433594,
|
| 77 |
+
"logps/chosen": -347.7822265625,
|
| 78 |
+
"logps/rejected": -376.8358459472656,
|
| 79 |
+
"loss": 0.6045,
|
| 80 |
+
"rewards/accuracies": 0.875,
|
| 81 |
+
"rewards/chosen": 0.05541626736521721,
|
| 82 |
+
"rewards/margins": 0.19534507393836975,
|
| 83 |
+
"rewards/rejected": -0.13992881774902344,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.10378827192527244,
|
| 88 |
+
"grad_norm": 12.348348618840042,
|
| 89 |
+
"learning_rate": 4.999933894080444e-07,
|
| 90 |
+
"logits/chosen": -2.714632511138916,
|
| 91 |
+
"logits/rejected": -2.6948368549346924,
|
| 92 |
+
"logps/chosen": -402.96624755859375,
|
| 93 |
+
"logps/rejected": -494.892578125,
|
| 94 |
+
"loss": 0.4675,
|
| 95 |
+
"rewards/accuracies": 0.8656250238418579,
|
| 96 |
+
"rewards/chosen": -0.3671617805957794,
|
| 97 |
+
"rewards/margins": 0.6783493161201477,
|
| 98 |
+
"rewards/rejected": -1.04551100730896,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.12454592631032693,
|
| 103 |
+
"grad_norm": 17.768413475155672,
|
| 104 |
+
"learning_rate": 4.992005413014143e-07,
|
| 105 |
+
"logits/chosen": -2.7408804893493652,
|
| 106 |
+
"logits/rejected": -2.736759662628174,
|
| 107 |
+
"logps/chosen": -526.3796997070312,
|
| 108 |
+
"logps/rejected": -740.1470947265625,
|
| 109 |
+
"loss": 0.3505,
|
| 110 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 111 |
+
"rewards/chosen": -1.7920459508895874,
|
| 112 |
+
"rewards/margins": 1.7205908298492432,
|
| 113 |
+
"rewards/rejected": -3.512636661529541,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.14530358069538143,
|
| 118 |
+
"grad_norm": 20.91483447209899,
|
| 119 |
+
"learning_rate": 4.970903776169402e-07,
|
| 120 |
+
"logits/chosen": -2.7555935382843018,
|
| 121 |
+
"logits/rejected": -2.731161117553711,
|
| 122 |
+
"logps/chosen": -641.5057373046875,
|
| 123 |
+
"logps/rejected": -879.7139892578125,
|
| 124 |
+
"loss": 0.3023,
|
| 125 |
+
"rewards/accuracies": 0.846875011920929,
|
| 126 |
+
"rewards/chosen": -2.449052333831787,
|
| 127 |
+
"rewards/margins": 2.306373119354248,
|
| 128 |
+
"rewards/rejected": -4.755425453186035,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.1660612350804359,
|
| 133 |
+
"grad_norm": 18.23581038330664,
|
| 134 |
+
"learning_rate": 4.936740530314087e-07,
|
| 135 |
+
"logits/chosen": -2.339261054992676,
|
| 136 |
+
"logits/rejected": -2.113029956817627,
|
| 137 |
+
"logps/chosen": -601.9942626953125,
|
| 138 |
+
"logps/rejected": -914.1705932617188,
|
| 139 |
+
"loss": 0.2471,
|
| 140 |
+
"rewards/accuracies": 0.887499988079071,
|
| 141 |
+
"rewards/chosen": -2.1097259521484375,
|
| 142 |
+
"rewards/margins": 3.008650064468384,
|
| 143 |
+
"rewards/rejected": -5.1183762550354,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.1868188894654904,
|
| 148 |
+
"grad_norm": 22.34513227915026,
|
| 149 |
+
"learning_rate": 4.889696268057348e-07,
|
| 150 |
+
"logits/chosen": -1.7139612436294556,
|
| 151 |
+
"logits/rejected": -0.9158167839050293,
|
| 152 |
+
"logps/chosen": -591.7190551757812,
|
| 153 |
+
"logps/rejected": -955.1848754882812,
|
| 154 |
+
"loss": 0.2266,
|
| 155 |
+
"rewards/accuracies": 0.909375011920929,
|
| 156 |
+
"rewards/chosen": -2.2645881175994873,
|
| 157 |
+
"rewards/margins": 3.4328598976135254,
|
| 158 |
+
"rewards/rejected": -5.697447776794434,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.2075765438505449,
|
| 163 |
+
"grad_norm": 15.36420864345506,
|
| 164 |
+
"learning_rate": 4.830019673206996e-07,
|
| 165 |
+
"logits/chosen": -1.521246075630188,
|
| 166 |
+
"logits/rejected": -0.20611262321472168,
|
| 167 |
+
"logps/chosen": -612.281005859375,
|
| 168 |
+
"logps/rejected": -1068.5550537109375,
|
| 169 |
+
"loss": 0.2001,
|
| 170 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 171 |
+
"rewards/chosen": -2.3677589893341064,
|
| 172 |
+
"rewards/margins": 4.30899715423584,
|
| 173 |
+
"rewards/rejected": -6.676755428314209,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.2283341982355994,
|
| 178 |
+
"grad_norm": 21.617954842362128,
|
| 179 |
+
"learning_rate": 4.7580262061854606e-07,
|
| 180 |
+
"logits/chosen": -1.2374727725982666,
|
| 181 |
+
"logits/rejected": 0.18440499901771545,
|
| 182 |
+
"logps/chosen": -599.9720458984375,
|
| 183 |
+
"logps/rejected": -1089.302001953125,
|
| 184 |
+
"loss": 0.2019,
|
| 185 |
+
"rewards/accuracies": 0.940625011920929,
|
| 186 |
+
"rewards/chosen": -2.2848434448242188,
|
| 187 |
+
"rewards/margins": 4.657064914703369,
|
| 188 |
+
"rewards/rejected": -6.9419074058532715,
|
| 189 |
+
"step": 110
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.24909185262065386,
|
| 193 |
+
"grad_norm": 14.742905110037716,
|
| 194 |
+
"learning_rate": 4.674096436453447e-07,
|
| 195 |
+
"logits/chosen": -0.7879268527030945,
|
| 196 |
+
"logits/rejected": 0.3564757704734802,
|
| 197 |
+
"logps/chosen": -665.6008911132812,
|
| 198 |
+
"logps/rejected": -1112.988037109375,
|
| 199 |
+
"loss": 0.1942,
|
| 200 |
+
"rewards/accuracies": 0.909375011920929,
|
| 201 |
+
"rewards/chosen": -2.7063541412353516,
|
| 202 |
+
"rewards/margins": 4.290424346923828,
|
| 203 |
+
"rewards/rejected": -6.9967780113220215,
|
| 204 |
+
"step": 120
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.26984950700570837,
|
| 208 |
+
"grad_norm": 11.973289583871454,
|
| 209 |
+
"learning_rate": 4.578674030756363e-07,
|
| 210 |
+
"logits/chosen": -1.418628454208374,
|
| 211 |
+
"logits/rejected": 0.030070018023252487,
|
| 212 |
+
"logps/chosen": -627.1656494140625,
|
| 213 |
+
"logps/rejected": -1140.0579833984375,
|
| 214 |
+
"loss": 0.1772,
|
| 215 |
+
"rewards/accuracies": 0.890625,
|
| 216 |
+
"rewards/chosen": -2.4757330417633057,
|
| 217 |
+
"rewards/margins": 4.854559898376465,
|
| 218 |
+
"rewards/rejected": -7.330292701721191,
|
| 219 |
+
"step": 130
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.29060716139076287,
|
| 223 |
+
"grad_norm": 15.153812063781148,
|
| 224 |
+
"learning_rate": 4.4722634078279865e-07,
|
| 225 |
+
"logits/chosen": -1.3328105211257935,
|
| 226 |
+
"logits/rejected": 0.4199509620666504,
|
| 227 |
+
"logps/chosen": -610.9807739257812,
|
| 228 |
+
"logps/rejected": -1104.3985595703125,
|
| 229 |
+
"loss": 0.1988,
|
| 230 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 231 |
+
"rewards/chosen": -2.4051623344421387,
|
| 232 |
+
"rewards/margins": 4.705986976623535,
|
| 233 |
+
"rewards/rejected": -7.111149787902832,
|
| 234 |
+
"step": 140
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.3113648157758173,
|
| 238 |
+
"grad_norm": 17.491121999774048,
|
| 239 |
+
"learning_rate": 4.355427071949004e-07,
|
| 240 |
+
"logits/chosen": -1.5610778331756592,
|
| 241 |
+
"logits/rejected": 0.6799826622009277,
|
| 242 |
+
"logps/chosen": -596.3429565429688,
|
| 243 |
+
"logps/rejected": -1119.742431640625,
|
| 244 |
+
"loss": 0.1647,
|
| 245 |
+
"rewards/accuracies": 0.934374988079071,
|
| 246 |
+
"rewards/chosen": -2.3613154888153076,
|
| 247 |
+
"rewards/margins": 5.037424564361572,
|
| 248 |
+
"rewards/rejected": -7.398741245269775,
|
| 249 |
+
"step": 150
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.3321224701608718,
|
| 253 |
+
"grad_norm": 19.441381636430275,
|
| 254 |
+
"learning_rate": 4.228782639455674e-07,
|
| 255 |
+
"logits/chosen": -1.1263091564178467,
|
| 256 |
+
"logits/rejected": 1.3055658340454102,
|
| 257 |
+
"logps/chosen": -651.0065307617188,
|
| 258 |
+
"logps/rejected": -1252.4947509765625,
|
| 259 |
+
"loss": 0.169,
|
| 260 |
+
"rewards/accuracies": 0.9375,
|
| 261 |
+
"rewards/chosen": -2.671025276184082,
|
| 262 |
+
"rewards/margins": 5.759725570678711,
|
| 263 |
+
"rewards/rejected": -8.430750846862793,
|
| 264 |
+
"step": 160
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.3528801245459263,
|
| 268 |
+
"grad_norm": 15.454380512498973,
|
| 269 |
+
"learning_rate": 4.092999573916971e-07,
|
| 270 |
+
"logits/chosen": -0.8963528871536255,
|
| 271 |
+
"logits/rejected": 1.6645536422729492,
|
| 272 |
+
"logps/chosen": -615.8960571289062,
|
| 273 |
+
"logps/rejected": -1188.300048828125,
|
| 274 |
+
"loss": 0.1723,
|
| 275 |
+
"rewards/accuracies": 0.9125000238418579,
|
| 276 |
+
"rewards/chosen": -2.5705718994140625,
|
| 277 |
+
"rewards/margins": 5.532119274139404,
|
| 278 |
+
"rewards/rejected": -8.102691650390625,
|
| 279 |
+
"step": 170
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.3736377789309808,
|
| 283 |
+
"grad_norm": 18.19445786127511,
|
| 284 |
+
"learning_rate": 3.948795647238637e-07,
|
| 285 |
+
"logits/chosen": -0.9939937591552734,
|
| 286 |
+
"logits/rejected": 1.5464706420898438,
|
| 287 |
+
"logps/chosen": -629.0621948242188,
|
| 288 |
+
"logps/rejected": -1243.3023681640625,
|
| 289 |
+
"loss": 0.175,
|
| 290 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 291 |
+
"rewards/chosen": -2.624224901199341,
|
| 292 |
+
"rewards/margins": 5.946156978607178,
|
| 293 |
+
"rewards/rejected": -8.570383071899414,
|
| 294 |
+
"step": 180
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.39439543331603527,
|
| 298 |
+
"grad_norm": 16.368013896038917,
|
| 299 |
+
"learning_rate": 3.796933145401304e-07,
|
| 300 |
+
"logits/chosen": -1.174119472503662,
|
| 301 |
+
"logits/rejected": 1.0995179414749146,
|
| 302 |
+
"logps/chosen": -662.320556640625,
|
| 303 |
+
"logps/rejected": -1293.4354248046875,
|
| 304 |
+
"loss": 0.1671,
|
| 305 |
+
"rewards/accuracies": 0.9375,
|
| 306 |
+
"rewards/chosen": -2.7964344024658203,
|
| 307 |
+
"rewards/margins": 5.946679592132568,
|
| 308 |
+
"rewards/rejected": -8.743115425109863,
|
| 309 |
+
"step": 190
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.4151530877010898,
|
| 313 |
+
"grad_norm": 14.042046955172085,
|
| 314 |
+
"learning_rate": 3.638214838889801e-07,
|
| 315 |
+
"logits/chosen": -0.7083699107170105,
|
| 316 |
+
"logits/rejected": 1.8419784307479858,
|
| 317 |
+
"logps/chosen": -618.2698364257812,
|
| 318 |
+
"logps/rejected": -1210.1761474609375,
|
| 319 |
+
"loss": 0.1636,
|
| 320 |
+
"rewards/accuracies": 0.925000011920929,
|
| 321 |
+
"rewards/chosen": -2.5069775581359863,
|
| 322 |
+
"rewards/margins": 5.630901336669922,
|
| 323 |
+
"rewards/rejected": -8.137879371643066,
|
| 324 |
+
"step": 200
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.4359107420861443,
|
| 328 |
+
"grad_norm": 16.47603854480112,
|
| 329 |
+
"learning_rate": 3.4734797391146383e-07,
|
| 330 |
+
"logits/chosen": -0.4444239139556885,
|
| 331 |
+
"logits/rejected": 2.3249332904815674,
|
| 332 |
+
"logps/chosen": -619.8510131835938,
|
| 333 |
+
"logps/rejected": -1215.9896240234375,
|
| 334 |
+
"loss": 0.1522,
|
| 335 |
+
"rewards/accuracies": 0.934374988079071,
|
| 336 |
+
"rewards/chosen": -2.5075623989105225,
|
| 337 |
+
"rewards/margins": 5.768418312072754,
|
| 338 |
+
"rewards/rejected": -8.275980949401855,
|
| 339 |
+
"step": 210
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.4566683964711988,
|
| 343 |
+
"grad_norm": 29.46403047682205,
|
| 344 |
+
"learning_rate": 3.3035986632579036e-07,
|
| 345 |
+
"logits/chosen": 0.16329942643642426,
|
| 346 |
+
"logits/rejected": 2.8997693061828613,
|
| 347 |
+
"logps/chosen": -659.2125854492188,
|
| 348 |
+
"logps/rejected": -1382.5194091796875,
|
| 349 |
+
"loss": 0.1615,
|
| 350 |
+
"rewards/accuracies": 0.925000011920929,
|
| 351 |
+
"rewards/chosen": -2.939960479736328,
|
| 352 |
+
"rewards/margins": 7.053119659423828,
|
| 353 |
+
"rewards/rejected": -9.993081092834473,
|
| 354 |
+
"step": 220
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.4774260508562532,
|
| 358 |
+
"grad_norm": 21.839461051416258,
|
| 359 |
+
"learning_rate": 3.1294696309885716e-07,
|
| 360 |
+
"logits/chosen": 0.3789651095867157,
|
| 361 |
+
"logits/rejected": 3.142848253250122,
|
| 362 |
+
"logps/chosen": -676.8601684570312,
|
| 363 |
+
"logps/rejected": -1436.1219482421875,
|
| 364 |
+
"loss": 0.1628,
|
| 365 |
+
"rewards/accuracies": 0.9156249761581421,
|
| 366 |
+
"rewards/chosen": -3.1580305099487305,
|
| 367 |
+
"rewards/margins": 7.436465263366699,
|
| 368 |
+
"rewards/rejected": -10.59449577331543,
|
| 369 |
+
"step": 230
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.49818370524130773,
|
| 373 |
+
"grad_norm": 16.48504711480412,
|
| 374 |
+
"learning_rate": 2.952013117380913e-07,
|
| 375 |
+
"logits/chosen": -0.004539084620773792,
|
| 376 |
+
"logits/rejected": 3.229100465774536,
|
| 377 |
+
"logps/chosen": -651.2918701171875,
|
| 378 |
+
"logps/rejected": -1440.683837890625,
|
| 379 |
+
"loss": 0.1534,
|
| 380 |
+
"rewards/accuracies": 0.956250011920929,
|
| 381 |
+
"rewards/chosen": -2.9313418865203857,
|
| 382 |
+
"rewards/margins": 7.655765533447266,
|
| 383 |
+
"rewards/rejected": -10.587106704711914,
|
| 384 |
+
"step": 240
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.5189413596263622,
|
| 388 |
+
"grad_norm": 18.156585865508323,
|
| 389 |
+
"learning_rate": 2.7721671871299114e-07,
|
| 390 |
+
"logits/chosen": -0.13780581951141357,
|
| 391 |
+
"logits/rejected": 2.874418258666992,
|
| 392 |
+
"logps/chosen": -682.3836669921875,
|
| 393 |
+
"logps/rejected": -1423.32861328125,
|
| 394 |
+
"loss": 0.1577,
|
| 395 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 396 |
+
"rewards/chosen": -3.025876522064209,
|
| 397 |
+
"rewards/margins": 7.198524475097656,
|
| 398 |
+
"rewards/rejected": -10.224401473999023,
|
| 399 |
+
"step": 250
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.5396990140114167,
|
| 403 |
+
"grad_norm": 17.237094148806953,
|
| 404 |
+
"learning_rate": 2.5908825357849993e-07,
|
| 405 |
+
"logits/chosen": -0.7950791716575623,
|
| 406 |
+
"logits/rejected": 2.015465497970581,
|
| 407 |
+
"logps/chosen": -645.5744018554688,
|
| 408 |
+
"logps/rejected": -1297.5784912109375,
|
| 409 |
+
"loss": 0.1566,
|
| 410 |
+
"rewards/accuracies": 0.9156249761581421,
|
| 411 |
+
"rewards/chosen": -2.724034309387207,
|
| 412 |
+
"rewards/margins": 6.29718542098999,
|
| 413 |
+
"rewards/rejected": -9.021219253540039,
|
| 414 |
+
"step": 260
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.5604566683964712,
|
| 418 |
+
"grad_norm": 17.46783320885481,
|
| 419 |
+
"learning_rate": 2.409117464215001e-07,
|
| 420 |
+
"logits/chosen": 0.14433838427066803,
|
| 421 |
+
"logits/rejected": 3.1040682792663574,
|
| 422 |
+
"logps/chosen": -684.5079345703125,
|
| 423 |
+
"logps/rejected": -1501.366943359375,
|
| 424 |
+
"loss": 0.1413,
|
| 425 |
+
"rewards/accuracies": 0.9468749761581421,
|
| 426 |
+
"rewards/chosen": -3.3087615966796875,
|
| 427 |
+
"rewards/margins": 7.854517936706543,
|
| 428 |
+
"rewards/rejected": -11.16327953338623,
|
| 429 |
+
"step": 270
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.5812143227815257,
|
| 433 |
+
"grad_norm": 17.63149610037019,
|
| 434 |
+
"learning_rate": 2.227832812870089e-07,
|
| 435 |
+
"logits/chosen": 0.3872329294681549,
|
| 436 |
+
"logits/rejected": 3.5623180866241455,
|
| 437 |
+
"logps/chosen": -775.79736328125,
|
| 438 |
+
"logps/rejected": -1631.552490234375,
|
| 439 |
+
"loss": 0.1516,
|
| 440 |
+
"rewards/accuracies": 0.925000011920929,
|
| 441 |
+
"rewards/chosen": -3.998131275177002,
|
| 442 |
+
"rewards/margins": 8.329412460327148,
|
| 443 |
+
"rewards/rejected": -12.327543258666992,
|
| 444 |
+
"step": 280
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.6019719771665801,
|
| 448 |
+
"grad_norm": 21.86816160236415,
|
| 449 |
+
"learning_rate": 2.0479868826190871e-07,
|
| 450 |
+
"logits/chosen": -0.06945541501045227,
|
| 451 |
+
"logits/rejected": 2.7963638305664062,
|
| 452 |
+
"logps/chosen": -767.2408447265625,
|
| 453 |
+
"logps/rejected": -1517.348876953125,
|
| 454 |
+
"loss": 0.1596,
|
| 455 |
+
"rewards/accuracies": 0.918749988079071,
|
| 456 |
+
"rewards/chosen": -3.9247982501983643,
|
| 457 |
+
"rewards/margins": 7.42672872543335,
|
| 458 |
+
"rewards/rejected": -11.351527214050293,
|
| 459 |
+
"step": 290
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.6227296315516346,
|
| 463 |
+
"grad_norm": 14.515183994248952,
|
| 464 |
+
"learning_rate": 1.8705303690114287e-07,
|
| 465 |
+
"logits/chosen": 0.07333996891975403,
|
| 466 |
+
"logits/rejected": 2.5020909309387207,
|
| 467 |
+
"logps/chosen": -785.2969360351562,
|
| 468 |
+
"logps/rejected": -1529.5716552734375,
|
| 469 |
+
"loss": 0.1421,
|
| 470 |
+
"rewards/accuracies": 0.949999988079071,
|
| 471 |
+
"rewards/chosen": -4.092496395111084,
|
| 472 |
+
"rewards/margins": 7.192312717437744,
|
| 473 |
+
"rewards/rejected": -11.284809112548828,
|
| 474 |
+
"step": 300
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.6434872859366891,
|
| 478 |
+
"grad_norm": 17.979677522182545,
|
| 479 |
+
"learning_rate": 1.6964013367420965e-07,
|
| 480 |
+
"logits/chosen": -0.13572996854782104,
|
| 481 |
+
"logits/rejected": 2.765768527984619,
|
| 482 |
+
"logps/chosen": -797.8541870117188,
|
| 483 |
+
"logps/rejected": -1586.290771484375,
|
| 484 |
+
"loss": 0.1457,
|
| 485 |
+
"rewards/accuracies": 0.90625,
|
| 486 |
+
"rewards/chosen": -4.280360221862793,
|
| 487 |
+
"rewards/margins": 7.809360504150391,
|
| 488 |
+
"rewards/rejected": -12.0897216796875,
|
| 489 |
+
"step": 310
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.6642449403217436,
|
| 493 |
+
"grad_norm": 12.321699055408487,
|
| 494 |
+
"learning_rate": 1.5265202608853628e-07,
|
| 495 |
+
"logits/chosen": -0.27073314785957336,
|
| 496 |
+
"logits/rejected": 2.6812171936035156,
|
| 497 |
+
"logps/chosen": -769.6759643554688,
|
| 498 |
+
"logps/rejected": -1649.497802734375,
|
| 499 |
+
"loss": 0.1453,
|
| 500 |
+
"rewards/accuracies": 0.925000011920929,
|
| 501 |
+
"rewards/chosen": -4.045531272888184,
|
| 502 |
+
"rewards/margins": 8.580347061157227,
|
| 503 |
+
"rewards/rejected": -12.625877380371094,
|
| 504 |
+
"step": 320
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 0.6850025947067981,
|
| 508 |
+
"grad_norm": 11.097171754192187,
|
| 509 |
+
"learning_rate": 1.3617851611101993e-07,
|
| 510 |
+
"logits/chosen": -0.5093133449554443,
|
| 511 |
+
"logits/rejected": 2.036428928375244,
|
| 512 |
+
"logps/chosen": -759.6971435546875,
|
| 513 |
+
"logps/rejected": -1584.39892578125,
|
| 514 |
+
"loss": 0.1446,
|
| 515 |
+
"rewards/accuracies": 0.921875,
|
| 516 |
+
"rewards/chosen": -3.8719277381896973,
|
| 517 |
+
"rewards/margins": 7.951642036437988,
|
| 518 |
+
"rewards/rejected": -11.823568344116211,
|
| 519 |
+
"step": 330
|
| 520 |
+
},
|
| 521 |
+
{
|
| 522 |
+
"epoch": 0.7057602490918526,
|
| 523 |
+
"grad_norm": 14.623856678041504,
|
| 524 |
+
"learning_rate": 1.2030668545986958e-07,
|
| 525 |
+
"logits/chosen": -0.3804703652858734,
|
| 526 |
+
"logits/rejected": 2.350010871887207,
|
| 527 |
+
"logps/chosen": -824.4983520507812,
|
| 528 |
+
"logps/rejected": -1692.0347900390625,
|
| 529 |
+
"loss": 0.137,
|
| 530 |
+
"rewards/accuracies": 0.940625011920929,
|
| 531 |
+
"rewards/chosen": -4.51037073135376,
|
| 532 |
+
"rewards/margins": 8.377652168273926,
|
| 533 |
+
"rewards/rejected": -12.888023376464844,
|
| 534 |
+
"step": 340
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 0.7265179034769071,
|
| 538 |
+
"grad_norm": 17.122700499170215,
|
| 539 |
+
"learning_rate": 1.0512043527613623e-07,
|
| 540 |
+
"logits/chosen": -0.2618183493614197,
|
| 541 |
+
"logits/rejected": 2.3642690181732178,
|
| 542 |
+
"logps/chosen": -845.6104736328125,
|
| 543 |
+
"logps/rejected": -1714.4976806640625,
|
| 544 |
+
"loss": 0.1494,
|
| 545 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 546 |
+
"rewards/chosen": -4.845407009124756,
|
| 547 |
+
"rewards/margins": 8.386329650878906,
|
| 548 |
+
"rewards/rejected": -13.23173713684082,
|
| 549 |
+
"step": 350
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.7472755578619616,
|
| 553 |
+
"grad_norm": 19.444982747452297,
|
| 554 |
+
"learning_rate": 9.070004260830294e-08,
|
| 555 |
+
"logits/chosen": -0.7990102767944336,
|
| 556 |
+
"logits/rejected": 2.3197755813598633,
|
| 557 |
+
"logps/chosen": -802.9656982421875,
|
| 558 |
+
"logps/rejected": -1649.383544921875,
|
| 559 |
+
"loss": 0.1393,
|
| 560 |
+
"rewards/accuracies": 0.934374988079071,
|
| 561 |
+
"rewards/chosen": -4.384912490844727,
|
| 562 |
+
"rewards/margins": 8.375471115112305,
|
| 563 |
+
"rewards/rejected": -12.760382652282715,
|
| 564 |
+
"step": 360
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.768033212247016,
|
| 568 |
+
"grad_norm": 17.93208817126597,
|
| 569 |
+
"learning_rate": 7.712173605443267e-08,
|
| 570 |
+
"logits/chosen": -0.9170758128166199,
|
| 571 |
+
"logits/rejected": 2.2540900707244873,
|
| 572 |
+
"logps/chosen": -794.91015625,
|
| 573 |
+
"logps/rejected": -1676.653564453125,
|
| 574 |
+
"loss": 0.1463,
|
| 575 |
+
"rewards/accuracies": 0.9156249761581421,
|
| 576 |
+
"rewards/chosen": -4.2809953689575195,
|
| 577 |
+
"rewards/margins": 8.688231468200684,
|
| 578 |
+
"rewards/rejected": -12.96922779083252,
|
| 579 |
+
"step": 370
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.7887908666320705,
|
| 583 |
+
"grad_norm": 11.992323551908767,
|
| 584 |
+
"learning_rate": 6.445729280509957e-08,
|
| 585 |
+
"logits/chosen": -0.5615069270133972,
|
| 586 |
+
"logits/rejected": 2.490211009979248,
|
| 587 |
+
"logps/chosen": -826.1986083984375,
|
| 588 |
+
"logps/rejected": -1815.263916015625,
|
| 589 |
+
"loss": 0.1361,
|
| 590 |
+
"rewards/accuracies": 0.9281250238418579,
|
| 591 |
+
"rewards/chosen": -4.737757682800293,
|
| 592 |
+
"rewards/margins": 9.302997589111328,
|
| 593 |
+
"rewards/rejected": -14.040754318237305,
|
| 594 |
+
"step": 380
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 0.809548521017125,
|
| 598 |
+
"grad_norm": 14.867121986591556,
|
| 599 |
+
"learning_rate": 5.2773659217201364e-08,
|
| 600 |
+
"logits/chosen": -0.4665653109550476,
|
| 601 |
+
"logits/rejected": 2.803576946258545,
|
| 602 |
+
"logps/chosen": -860.3473510742188,
|
| 603 |
+
"logps/rejected": -1715.7301025390625,
|
| 604 |
+
"loss": 0.1459,
|
| 605 |
+
"rewards/accuracies": 0.9437500238418579,
|
| 606 |
+
"rewards/chosen": -4.846836566925049,
|
| 607 |
+
"rewards/margins": 8.348875045776367,
|
| 608 |
+
"rewards/rejected": -13.195712089538574,
|
| 609 |
+
"step": 390
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.8303061754021795,
|
| 613 |
+
"grad_norm": 14.137593981426674,
|
| 614 |
+
"learning_rate": 4.213259692436366e-08,
|
| 615 |
+
"logits/chosen": -0.40155744552612305,
|
| 616 |
+
"logits/rejected": 2.635136604309082,
|
| 617 |
+
"logps/chosen": -858.0413818359375,
|
| 618 |
+
"logps/rejected": -1749.977294921875,
|
| 619 |
+
"loss": 0.1388,
|
| 620 |
+
"rewards/accuracies": 0.940625011920929,
|
| 621 |
+
"rewards/chosen": -4.928585529327393,
|
| 622 |
+
"rewards/margins": 8.627421379089355,
|
| 623 |
+
"rewards/rejected": -13.556007385253906,
|
| 624 |
+
"step": 400
|
| 625 |
+
},
|
| 626 |
+
{
|
| 627 |
+
"epoch": 0.851063829787234,
|
| 628 |
+
"grad_norm": 18.46721900012293,
|
| 629 |
+
"learning_rate": 3.259035635465529e-08,
|
| 630 |
+
"logits/chosen": -0.4056881070137024,
|
| 631 |
+
"logits/rejected": 2.437743902206421,
|
| 632 |
+
"logps/chosen": -871.8626098632812,
|
| 633 |
+
"logps/rejected": -1658.709716796875,
|
| 634 |
+
"loss": 0.1416,
|
| 635 |
+
"rewards/accuracies": 0.925000011920929,
|
| 636 |
+
"rewards/chosen": -5.064044952392578,
|
| 637 |
+
"rewards/margins": 7.721222877502441,
|
| 638 |
+
"rewards/rejected": -12.785266876220703,
|
| 639 |
+
"step": 410
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.8718214841722886,
|
| 643 |
+
"grad_norm": 18.74063979970814,
|
| 644 |
+
"learning_rate": 2.4197379381453942e-08,
|
| 645 |
+
"logits/chosen": -0.33163413405418396,
|
| 646 |
+
"logits/rejected": 2.5784339904785156,
|
| 647 |
+
"logps/chosen": -880.70654296875,
|
| 648 |
+
"logps/rejected": -1740.9976806640625,
|
| 649 |
+
"loss": 0.1443,
|
| 650 |
+
"rewards/accuracies": 0.921875,
|
| 651 |
+
"rewards/chosen": -5.066455364227295,
|
| 652 |
+
"rewards/margins": 8.281970024108887,
|
| 653 |
+
"rewards/rejected": -13.348426818847656,
|
| 654 |
+
"step": 420
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.892579138557343,
|
| 658 |
+
"grad_norm": 17.16551956889699,
|
| 659 |
+
"learning_rate": 1.699803267930039e-08,
|
| 660 |
+
"logits/chosen": -0.3569720387458801,
|
| 661 |
+
"logits/rejected": 2.325591564178467,
|
| 662 |
+
"logps/chosen": -865.0040283203125,
|
| 663 |
+
"logps/rejected": -1691.733642578125,
|
| 664 |
+
"loss": 0.1333,
|
| 665 |
+
"rewards/accuracies": 0.921875,
|
| 666 |
+
"rewards/chosen": -4.9201178550720215,
|
| 667 |
+
"rewards/margins": 8.041346549987793,
|
| 668 |
+
"rewards/rejected": -12.961465835571289,
|
| 669 |
+
"step": 430
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.9133367929423976,
|
| 673 |
+
"grad_norm": 15.033949759220667,
|
| 674 |
+
"learning_rate": 1.1030373194265114e-08,
|
| 675 |
+
"logits/chosen": -0.3576432466506958,
|
| 676 |
+
"logits/rejected": 2.751772403717041,
|
| 677 |
+
"logps/chosen": -866.8192138671875,
|
| 678 |
+
"logps/rejected": -1763.2193603515625,
|
| 679 |
+
"loss": 0.1491,
|
| 680 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 681 |
+
"rewards/chosen": -4.9160566329956055,
|
| 682 |
+
"rewards/margins": 8.817602157592773,
|
| 683 |
+
"rewards/rejected": -13.733657836914062,
|
| 684 |
+
"step": 440
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 0.934094447327452,
|
| 688 |
+
"grad_norm": 17.88061807294248,
|
| 689 |
+
"learning_rate": 6.325946968591317e-09,
|
| 690 |
+
"logits/chosen": -0.2799724042415619,
|
| 691 |
+
"logits/rejected": 2.8802428245544434,
|
| 692 |
+
"logps/chosen": -871.8025512695312,
|
| 693 |
+
"logps/rejected": -1729.011962890625,
|
| 694 |
+
"loss": 0.1411,
|
| 695 |
+
"rewards/accuracies": 0.9281250238418579,
|
| 696 |
+
"rewards/chosen": -5.046684265136719,
|
| 697 |
+
"rewards/margins": 8.479043960571289,
|
| 698 |
+
"rewards/rejected": -13.525728225708008,
|
| 699 |
+
"step": 450
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 0.9548521017125065,
|
| 703 |
+
"grad_norm": 16.510600900187168,
|
| 704 |
+
"learning_rate": 2.909622383059834e-09,
|
| 705 |
+
"logits/chosen": -0.2751966118812561,
|
| 706 |
+
"logits/rejected": 2.7767112255096436,
|
| 707 |
+
"logps/chosen": -868.7591552734375,
|
| 708 |
+
"logps/rejected": -1734.588134765625,
|
| 709 |
+
"loss": 0.1428,
|
| 710 |
+
"rewards/accuracies": 0.940625011920929,
|
| 711 |
+
"rewards/chosen": -4.989295959472656,
|
| 712 |
+
"rewards/margins": 8.519960403442383,
|
| 713 |
+
"rewards/rejected": -13.509256362915039,
|
| 714 |
+
"step": 460
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 0.975609756097561,
|
| 718 |
+
"grad_norm": 19.07149083473138,
|
| 719 |
+
"learning_rate": 7.994586985856089e-10,
|
| 720 |
+
"logits/chosen": -0.4134143888950348,
|
| 721 |
+
"logits/rejected": 2.6374351978302,
|
| 722 |
+
"logps/chosen": -877.0185546875,
|
| 723 |
+
"logps/rejected": -1758.679443359375,
|
| 724 |
+
"loss": 0.1319,
|
| 725 |
+
"rewards/accuracies": 0.9312499761581421,
|
| 726 |
+
"rewards/chosen": -5.058596611022949,
|
| 727 |
+
"rewards/margins": 8.577057838439941,
|
| 728 |
+
"rewards/rejected": -13.635653495788574,
|
| 729 |
+
"step": 470
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.9963674104826155,
|
| 733 |
+
"grad_norm": 12.263362857750089,
|
| 734 |
+
"learning_rate": 6.610591955641398e-12,
|
| 735 |
+
"logits/chosen": -0.18534007668495178,
|
| 736 |
+
"logits/rejected": 2.617680311203003,
|
| 737 |
+
"logps/chosen": -867.3248291015625,
|
| 738 |
+
"logps/rejected": -1753.2431640625,
|
| 739 |
+
"loss": 0.1436,
|
| 740 |
+
"rewards/accuracies": 0.9437500238418579,
|
| 741 |
+
"rewards/chosen": -5.15399169921875,
|
| 742 |
+
"rewards/margins": 8.56091594696045,
|
| 743 |
+
"rewards/rejected": -13.7149076461792,
|
| 744 |
+
"step": 480
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 0.9984431759211209,
|
| 748 |
+
"step": 481,
|
| 749 |
+
"total_flos": 0.0,
|
| 750 |
+
"train_loss": 0.21509297628908297,
|
| 751 |
+
"train_runtime": 13791.8998,
|
| 752 |
+
"train_samples_per_second": 8.941,
|
| 753 |
+
"train_steps_per_second": 0.035
|
| 754 |
+
}
|
| 755 |
+
],
|
| 756 |
+
"logging_steps": 10,
|
| 757 |
+
"max_steps": 481,
|
| 758 |
+
"num_input_tokens_seen": 0,
|
| 759 |
+
"num_train_epochs": 1,
|
| 760 |
+
"save_steps": 100,
|
| 761 |
+
"stateful_callbacks": {
|
| 762 |
+
"TrainerControl": {
|
| 763 |
+
"args": {
|
| 764 |
+
"should_epoch_stop": false,
|
| 765 |
+
"should_evaluate": false,
|
| 766 |
+
"should_log": false,
|
| 767 |
+
"should_save": true,
|
| 768 |
+
"should_training_stop": true
|
| 769 |
+
},
|
| 770 |
+
"attributes": {}
|
| 771 |
+
}
|
| 772 |
+
},
|
| 773 |
+
"total_flos": 0.0,
|
| 774 |
+
"train_batch_size": 8,
|
| 775 |
+
"trial_name": null,
|
| 776 |
+
"trial_params": null
|
| 777 |
+
}
|