Model save
Browse files- README.md +60 -0
- adapter_model.safetensors +1 -1
- all_results.json +8 -0
- runs/May15_02-53-00_gpu4-119-5/events.out.tfevents.1715705669.gpu4-119-5.2318867.0 +2 -2
- train_results.json +8 -0
- trainer_state.json +912 -0
README.md
ADDED
|
@@ -0,0 +1,60 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: mit
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- trl
|
| 6 |
+
- dpo
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
base_model: DUAL-GPO/phi-2-gpo-new-i0
|
| 9 |
+
model-index:
|
| 10 |
+
- name: phi-2-gpo-v24-i1
|
| 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 |
+
# phi-2-gpo-v24-i1
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [DUAL-GPO/phi-2-gpo-new-i0](https://huggingface.co/DUAL-GPO/phi-2-gpo-new-i0) on the None 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-06
|
| 39 |
+
- train_batch_size: 4
|
| 40 |
+
- eval_batch_size: 4
|
| 41 |
+
- seed: 42
|
| 42 |
+
- distributed_type: multi-GPU
|
| 43 |
+
- gradient_accumulation_steps: 4
|
| 44 |
+
- total_train_batch_size: 16
|
| 45 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 46 |
+
- lr_scheduler_type: cosine
|
| 47 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 48 |
+
- num_epochs: 1
|
| 49 |
+
|
| 50 |
+
### Training results
|
| 51 |
+
|
| 52 |
+
|
| 53 |
+
|
| 54 |
+
### Framework versions
|
| 55 |
+
|
| 56 |
+
- PEFT 0.7.1
|
| 57 |
+
- Transformers 4.36.2
|
| 58 |
+
- Pytorch 2.1.2+cu121
|
| 59 |
+
- Datasets 2.14.6
|
| 60 |
+
- Tokenizers 0.15.2
|
adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 167807296
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2e9ededd6546bef48c1d75faf9a9db23f35510b0069a2dbde0887ff32ec33f80
|
| 3 |
size 167807296
|
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.31405294857025146,
|
| 4 |
+
"train_runtime": 5821.3149,
|
| 5 |
+
"train_samples": 10000,
|
| 6 |
+
"train_samples_per_second": 1.718,
|
| 7 |
+
"train_steps_per_second": 0.107
|
| 8 |
+
}
|
runs/May15_02-53-00_gpu4-119-5/events.out.tfevents.1715705669.gpu4-119-5.2318867.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fd228d93140b8881733256f6f4282807ffaee83dde6593eeca4446676e75f830
|
| 3 |
+
size 44980
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.31405294857025146,
|
| 4 |
+
"train_runtime": 5821.3149,
|
| 5 |
+
"train_samples": 10000,
|
| 6 |
+
"train_samples_per_second": 1.718,
|
| 7 |
+
"train_steps_per_second": 0.107
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,912 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 625,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0,
|
| 13 |
+
"learning_rate": 7.936507936507937e-08,
|
| 14 |
+
"logits/chosen": 0.16854308545589447,
|
| 15 |
+
"logits/rejected": 0.17400208115577698,
|
| 16 |
+
"logps/chosen": -327.90460205078125,
|
| 17 |
+
"logps/rejected": -253.07960510253906,
|
| 18 |
+
"loss": 0.4067,
|
| 19 |
+
"rewards/accuracies": 0.0,
|
| 20 |
+
"rewards/chosen": 0.0,
|
| 21 |
+
"rewards/margins": 0.0,
|
| 22 |
+
"rewards/rejected": 0.0,
|
| 23 |
+
"step": 1
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.02,
|
| 27 |
+
"learning_rate": 7.936507936507937e-07,
|
| 28 |
+
"logits/chosen": 0.08604434132575989,
|
| 29 |
+
"logits/rejected": 0.1809283345937729,
|
| 30 |
+
"logps/chosen": -327.1376647949219,
|
| 31 |
+
"logps/rejected": -306.0588684082031,
|
| 32 |
+
"loss": 0.3864,
|
| 33 |
+
"rewards/accuracies": 0.375,
|
| 34 |
+
"rewards/chosen": -5.549057823373005e-05,
|
| 35 |
+
"rewards/margins": -0.0001069721911335364,
|
| 36 |
+
"rewards/rejected": 5.148161653778516e-05,
|
| 37 |
+
"step": 10
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.03,
|
| 41 |
+
"learning_rate": 1.5873015873015873e-06,
|
| 42 |
+
"logits/chosen": 0.08089748024940491,
|
| 43 |
+
"logits/rejected": 0.2491210699081421,
|
| 44 |
+
"logps/chosen": -362.65435791015625,
|
| 45 |
+
"logps/rejected": -352.3981018066406,
|
| 46 |
+
"loss": 0.3729,
|
| 47 |
+
"rewards/accuracies": 0.48124998807907104,
|
| 48 |
+
"rewards/chosen": 2.957274773507379e-05,
|
| 49 |
+
"rewards/margins": 5.038862582296133e-05,
|
| 50 |
+
"rewards/rejected": -2.0815870811929926e-05,
|
| 51 |
+
"step": 20
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.05,
|
| 55 |
+
"learning_rate": 2.380952380952381e-06,
|
| 56 |
+
"logits/chosen": 0.18664827942848206,
|
| 57 |
+
"logits/rejected": 0.14790232479572296,
|
| 58 |
+
"logps/chosen": -312.6676940917969,
|
| 59 |
+
"logps/rejected": -324.85516357421875,
|
| 60 |
+
"loss": 0.3789,
|
| 61 |
+
"rewards/accuracies": 0.5,
|
| 62 |
+
"rewards/chosen": 8.431552851106972e-05,
|
| 63 |
+
"rewards/margins": 0.00019130071450490505,
|
| 64 |
+
"rewards/rejected": -0.00010698519326979294,
|
| 65 |
+
"step": 30
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.06,
|
| 69 |
+
"learning_rate": 3.1746031746031746e-06,
|
| 70 |
+
"logits/chosen": 0.08645667880773544,
|
| 71 |
+
"logits/rejected": 0.2045074999332428,
|
| 72 |
+
"logps/chosen": -347.3286437988281,
|
| 73 |
+
"logps/rejected": -375.4449768066406,
|
| 74 |
+
"loss": 0.3765,
|
| 75 |
+
"rewards/accuracies": 0.512499988079071,
|
| 76 |
+
"rewards/chosen": -0.0001862197823356837,
|
| 77 |
+
"rewards/margins": 6.710310117341578e-05,
|
| 78 |
+
"rewards/rejected": -0.000253322854405269,
|
| 79 |
+
"step": 40
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.08,
|
| 83 |
+
"learning_rate": 3.968253968253968e-06,
|
| 84 |
+
"logits/chosen": 0.11011010408401489,
|
| 85 |
+
"logits/rejected": 0.1821037232875824,
|
| 86 |
+
"logps/chosen": -385.66693115234375,
|
| 87 |
+
"logps/rejected": -349.8896789550781,
|
| 88 |
+
"loss": 0.3738,
|
| 89 |
+
"rewards/accuracies": 0.53125,
|
| 90 |
+
"rewards/chosen": 4.727130362880416e-05,
|
| 91 |
+
"rewards/margins": 0.00021036244288552552,
|
| 92 |
+
"rewards/rejected": -0.00016309114289470017,
|
| 93 |
+
"step": 50
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.1,
|
| 97 |
+
"learning_rate": 4.761904761904762e-06,
|
| 98 |
+
"logits/chosen": 0.07650097459554672,
|
| 99 |
+
"logits/rejected": 0.177745059132576,
|
| 100 |
+
"logps/chosen": -343.5960998535156,
|
| 101 |
+
"logps/rejected": -331.03912353515625,
|
| 102 |
+
"loss": 0.3796,
|
| 103 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 104 |
+
"rewards/chosen": 0.0009819746483117342,
|
| 105 |
+
"rewards/margins": 0.0015515821287408471,
|
| 106 |
+
"rewards/rejected": -0.0005696073058061302,
|
| 107 |
+
"step": 60
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.11,
|
| 111 |
+
"learning_rate": 4.998086282661188e-06,
|
| 112 |
+
"logits/chosen": 0.11013742536306381,
|
| 113 |
+
"logits/rejected": 0.16680167615413666,
|
| 114 |
+
"logps/chosen": -361.571044921875,
|
| 115 |
+
"logps/rejected": -318.3884582519531,
|
| 116 |
+
"loss": 0.3614,
|
| 117 |
+
"rewards/accuracies": 0.668749988079071,
|
| 118 |
+
"rewards/chosen": 0.0014994305092841387,
|
| 119 |
+
"rewards/margins": 0.0018448818009346724,
|
| 120 |
+
"rewards/rejected": -0.00034545123344287276,
|
| 121 |
+
"step": 70
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.13,
|
| 125 |
+
"learning_rate": 4.988720025682995e-06,
|
| 126 |
+
"logits/chosen": 0.1778658628463745,
|
| 127 |
+
"logits/rejected": 0.19966575503349304,
|
| 128 |
+
"logps/chosen": -292.3515625,
|
| 129 |
+
"logps/rejected": -321.511962890625,
|
| 130 |
+
"loss": 0.3511,
|
| 131 |
+
"rewards/accuracies": 0.625,
|
| 132 |
+
"rewards/chosen": 0.0031275060027837753,
|
| 133 |
+
"rewards/margins": 0.0046959021128714085,
|
| 134 |
+
"rewards/rejected": -0.0015683959936723113,
|
| 135 |
+
"step": 80
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.14,
|
| 139 |
+
"learning_rate": 4.9715789537359126e-06,
|
| 140 |
+
"logits/chosen": 0.14186400175094604,
|
| 141 |
+
"logits/rejected": 0.18624775111675262,
|
| 142 |
+
"logps/chosen": -360.6056823730469,
|
| 143 |
+
"logps/rejected": -317.9211730957031,
|
| 144 |
+
"loss": 0.3683,
|
| 145 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 146 |
+
"rewards/chosen": 0.004151093773543835,
|
| 147 |
+
"rewards/margins": 0.006570989731699228,
|
| 148 |
+
"rewards/rejected": -0.002419896889477968,
|
| 149 |
+
"step": 90
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.16,
|
| 153 |
+
"learning_rate": 4.946716615897932e-06,
|
| 154 |
+
"logits/chosen": 0.11935708671808243,
|
| 155 |
+
"logits/rejected": 0.18251103162765503,
|
| 156 |
+
"logps/chosen": -354.6888732910156,
|
| 157 |
+
"logps/rejected": -341.3345642089844,
|
| 158 |
+
"loss": 0.3703,
|
| 159 |
+
"rewards/accuracies": 0.637499988079071,
|
| 160 |
+
"rewards/chosen": -0.0003122333437204361,
|
| 161 |
+
"rewards/margins": 0.010582786984741688,
|
| 162 |
+
"rewards/rejected": -0.010895021259784698,
|
| 163 |
+
"step": 100
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.18,
|
| 167 |
+
"learning_rate": 4.9142106826480114e-06,
|
| 168 |
+
"logits/chosen": 0.06461165845394135,
|
| 169 |
+
"logits/rejected": 0.18685249984264374,
|
| 170 |
+
"logps/chosen": -362.90142822265625,
|
| 171 |
+
"logps/rejected": -353.4388122558594,
|
| 172 |
+
"loss": 0.3406,
|
| 173 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 174 |
+
"rewards/chosen": -0.007556666620075703,
|
| 175 |
+
"rewards/margins": 0.0255865640938282,
|
| 176 |
+
"rewards/rejected": -0.03314322978258133,
|
| 177 |
+
"step": 110
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.19,
|
| 181 |
+
"learning_rate": 4.874162703221823e-06,
|
| 182 |
+
"logits/chosen": 0.07998766005039215,
|
| 183 |
+
"logits/rejected": 0.17009273171424866,
|
| 184 |
+
"logps/chosen": -389.65362548828125,
|
| 185 |
+
"logps/rejected": -363.02374267578125,
|
| 186 |
+
"loss": 0.3605,
|
| 187 |
+
"rewards/accuracies": 0.606249988079071,
|
| 188 |
+
"rewards/chosen": -0.036845412105321884,
|
| 189 |
+
"rewards/margins": 0.020548993721604347,
|
| 190 |
+
"rewards/rejected": -0.05739440768957138,
|
| 191 |
+
"step": 120
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.21,
|
| 195 |
+
"learning_rate": 4.826697788369752e-06,
|
| 196 |
+
"logits/chosen": 0.08132342249155045,
|
| 197 |
+
"logits/rejected": 0.14768512547016144,
|
| 198 |
+
"logps/chosen": -431.18109130859375,
|
| 199 |
+
"logps/rejected": -458.8611755371094,
|
| 200 |
+
"loss": 0.3284,
|
| 201 |
+
"rewards/accuracies": 0.71875,
|
| 202 |
+
"rewards/chosen": -0.06240760162472725,
|
| 203 |
+
"rewards/margins": 0.06149229407310486,
|
| 204 |
+
"rewards/rejected": -0.12389989197254181,
|
| 205 |
+
"step": 130
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.22,
|
| 209 |
+
"learning_rate": 4.7719642195082224e-06,
|
| 210 |
+
"logits/chosen": 0.07962881028652191,
|
| 211 |
+
"logits/rejected": 0.07184718549251556,
|
| 212 |
+
"logps/chosen": -449.84393310546875,
|
| 213 |
+
"logps/rejected": -531.15087890625,
|
| 214 |
+
"loss": 0.2981,
|
| 215 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 216 |
+
"rewards/chosen": -0.13113334774971008,
|
| 217 |
+
"rewards/margins": 0.08747557550668716,
|
| 218 |
+
"rewards/rejected": -0.21860890090465546,
|
| 219 |
+
"step": 140
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.24,
|
| 223 |
+
"learning_rate": 4.710132985485355e-06,
|
| 224 |
+
"logits/chosen": 0.048229388892650604,
|
| 225 |
+
"logits/rejected": 0.13614359498023987,
|
| 226 |
+
"logps/chosen": -528.4464721679688,
|
| 227 |
+
"logps/rejected": -578.7511596679688,
|
| 228 |
+
"loss": 0.3243,
|
| 229 |
+
"rewards/accuracies": 0.59375,
|
| 230 |
+
"rewards/chosen": -0.18633897602558136,
|
| 231 |
+
"rewards/margins": 0.07957583665847778,
|
| 232 |
+
"rewards/rejected": -0.26591476798057556,
|
| 233 |
+
"step": 150
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.26,
|
| 237 |
+
"learning_rate": 4.641397248408122e-06,
|
| 238 |
+
"logits/chosen": 0.11920128762722015,
|
| 239 |
+
"logits/rejected": 0.15292052924633026,
|
| 240 |
+
"logps/chosen": -552.9255981445312,
|
| 241 |
+
"logps/rejected": -630.448486328125,
|
| 242 |
+
"loss": 0.3063,
|
| 243 |
+
"rewards/accuracies": 0.706250011920929,
|
| 244 |
+
"rewards/chosen": -0.1733545958995819,
|
| 245 |
+
"rewards/margins": 0.08863510191440582,
|
| 246 |
+
"rewards/rejected": -0.2619897127151489,
|
| 247 |
+
"step": 160
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.27,
|
| 251 |
+
"learning_rate": 4.5659717401997655e-06,
|
| 252 |
+
"logits/chosen": 0.06991209089756012,
|
| 253 |
+
"logits/rejected": 0.12328378856182098,
|
| 254 |
+
"logps/chosen": -470.7796325683594,
|
| 255 |
+
"logps/rejected": -537.2589721679688,
|
| 256 |
+
"loss": 0.3181,
|
| 257 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 258 |
+
"rewards/chosen": -0.14576277136802673,
|
| 259 |
+
"rewards/margins": 0.07790832221508026,
|
| 260 |
+
"rewards/rejected": -0.2236711084842682,
|
| 261 |
+
"step": 170
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.29,
|
| 265 |
+
"learning_rate": 4.4840920917726425e-06,
|
| 266 |
+
"logits/chosen": 0.128515362739563,
|
| 267 |
+
"logits/rejected": 0.18245835602283478,
|
| 268 |
+
"logps/chosen": -488.5677185058594,
|
| 269 |
+
"logps/rejected": -617.6085205078125,
|
| 270 |
+
"loss": 0.3053,
|
| 271 |
+
"rewards/accuracies": 0.65625,
|
| 272 |
+
"rewards/chosen": -0.15714891254901886,
|
| 273 |
+
"rewards/margins": 0.1121206283569336,
|
| 274 |
+
"rewards/rejected": -0.26926952600479126,
|
| 275 |
+
"step": 180
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.3,
|
| 279 |
+
"learning_rate": 4.396014096912182e-06,
|
| 280 |
+
"logits/chosen": 0.06968175619840622,
|
| 281 |
+
"logits/rejected": 0.1405131071805954,
|
| 282 |
+
"logps/chosen": -457.48687744140625,
|
| 283 |
+
"logps/rejected": -507.72491455078125,
|
| 284 |
+
"loss": 0.3217,
|
| 285 |
+
"rewards/accuracies": 0.65625,
|
| 286 |
+
"rewards/chosen": -0.144833505153656,
|
| 287 |
+
"rewards/margins": 0.07206179946660995,
|
| 288 |
+
"rewards/rejected": -0.21689531207084656,
|
| 289 |
+
"step": 190
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.32,
|
| 293 |
+
"learning_rate": 4.302012913171584e-06,
|
| 294 |
+
"logits/chosen": 0.11664710193872452,
|
| 295 |
+
"logits/rejected": 0.1742321252822876,
|
| 296 |
+
"logps/chosen": -498.94708251953125,
|
| 297 |
+
"logps/rejected": -543.8130493164062,
|
| 298 |
+
"loss": 0.3178,
|
| 299 |
+
"rewards/accuracies": 0.606249988079071,
|
| 300 |
+
"rewards/chosen": -0.1338798701763153,
|
| 301 |
+
"rewards/margins": 0.07743647694587708,
|
| 302 |
+
"rewards/rejected": -0.21131634712219238,
|
| 303 |
+
"step": 200
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.34,
|
| 307 |
+
"learning_rate": 4.202382202273702e-06,
|
| 308 |
+
"logits/chosen": 0.09494765847921371,
|
| 309 |
+
"logits/rejected": 0.1851937472820282,
|
| 310 |
+
"logps/chosen": -520.176025390625,
|
| 311 |
+
"logps/rejected": -567.4844970703125,
|
| 312 |
+
"loss": 0.3061,
|
| 313 |
+
"rewards/accuracies": 0.668749988079071,
|
| 314 |
+
"rewards/chosen": -0.14064568281173706,
|
| 315 |
+
"rewards/margins": 0.08944287896156311,
|
| 316 |
+
"rewards/rejected": -0.23008854687213898,
|
| 317 |
+
"step": 210
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.35,
|
| 321 |
+
"learning_rate": 4.097433212705492e-06,
|
| 322 |
+
"logits/chosen": 0.12784181535243988,
|
| 323 |
+
"logits/rejected": 0.12376650422811508,
|
| 324 |
+
"logps/chosen": -478.936767578125,
|
| 325 |
+
"logps/rejected": -583.5801391601562,
|
| 326 |
+
"loss": 0.3215,
|
| 327 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 328 |
+
"rewards/chosen": -0.16416385769844055,
|
| 329 |
+
"rewards/margins": 0.08470286428928375,
|
| 330 |
+
"rewards/rejected": -0.2488667219877243,
|
| 331 |
+
"step": 220
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.37,
|
| 335 |
+
"learning_rate": 3.987493807371033e-06,
|
| 336 |
+
"logits/chosen": 0.14440543949604034,
|
| 337 |
+
"logits/rejected": 0.19996806979179382,
|
| 338 |
+
"logps/chosen": -516.4754028320312,
|
| 339 |
+
"logps/rejected": -626.3532104492188,
|
| 340 |
+
"loss": 0.2979,
|
| 341 |
+
"rewards/accuracies": 0.71875,
|
| 342 |
+
"rewards/chosen": -0.1449437290430069,
|
| 343 |
+
"rewards/margins": 0.12377123534679413,
|
| 344 |
+
"rewards/rejected": -0.268714964389801,
|
| 345 |
+
"step": 230
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.38,
|
| 349 |
+
"learning_rate": 3.872907439340758e-06,
|
| 350 |
+
"logits/chosen": 0.17143338918685913,
|
| 351 |
+
"logits/rejected": 0.19446048140525818,
|
| 352 |
+
"logps/chosen": -530.1805419921875,
|
| 353 |
+
"logps/rejected": -642.0897827148438,
|
| 354 |
+
"loss": 0.3007,
|
| 355 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 356 |
+
"rewards/chosen": -0.1711062490940094,
|
| 357 |
+
"rewards/margins": 0.11968924105167389,
|
| 358 |
+
"rewards/rejected": -0.2907955050468445,
|
| 359 |
+
"step": 240
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.4,
|
| 363 |
+
"learning_rate": 3.75403207889666e-06,
|
| 364 |
+
"logits/chosen": 0.13282717764377594,
|
| 365 |
+
"logits/rejected": 0.22809620201587677,
|
| 366 |
+
"logps/chosen": -522.2652587890625,
|
| 367 |
+
"logps/rejected": -587.6581420898438,
|
| 368 |
+
"loss": 0.3026,
|
| 369 |
+
"rewards/accuracies": 0.668749988079071,
|
| 370 |
+
"rewards/chosen": -0.16960139572620392,
|
| 371 |
+
"rewards/margins": 0.09930792450904846,
|
| 372 |
+
"rewards/rejected": -0.2689093053340912,
|
| 373 |
+
"step": 250
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.42,
|
| 377 |
+
"learning_rate": 3.631239095225417e-06,
|
| 378 |
+
"logits/chosen": 0.17684991657733917,
|
| 379 |
+
"logits/rejected": 0.17547938227653503,
|
| 380 |
+
"logps/chosen": -496.5772399902344,
|
| 381 |
+
"logps/rejected": -577.0806884765625,
|
| 382 |
+
"loss": 0.3156,
|
| 383 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 384 |
+
"rewards/chosen": -0.1634175181388855,
|
| 385 |
+
"rewards/margins": 0.09483225643634796,
|
| 386 |
+
"rewards/rejected": -0.25824978947639465,
|
| 387 |
+
"step": 260
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.43,
|
| 391 |
+
"learning_rate": 3.5049120962530608e-06,
|
| 392 |
+
"logits/chosen": 0.09195151180028915,
|
| 393 |
+
"logits/rejected": 0.21280470490455627,
|
| 394 |
+
"logps/chosen": -544.9461669921875,
|
| 395 |
+
"logps/rejected": -664.9347534179688,
|
| 396 |
+
"loss": 0.2795,
|
| 397 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 398 |
+
"rewards/chosen": -0.18792234361171722,
|
| 399 |
+
"rewards/margins": 0.1312183439731598,
|
| 400 |
+
"rewards/rejected": -0.3191407024860382,
|
| 401 |
+
"step": 270
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.45,
|
| 405 |
+
"learning_rate": 3.3754457302455464e-06,
|
| 406 |
+
"logits/chosen": 0.14985263347625732,
|
| 407 |
+
"logits/rejected": 0.22761961817741394,
|
| 408 |
+
"logps/chosen": -607.9620361328125,
|
| 409 |
+
"logps/rejected": -663.138427734375,
|
| 410 |
+
"loss": 0.3365,
|
| 411 |
+
"rewards/accuracies": 0.668749988079071,
|
| 412 |
+
"rewards/chosen": -0.24214263260364532,
|
| 413 |
+
"rewards/margins": 0.07931643724441528,
|
| 414 |
+
"rewards/rejected": -0.3214590847492218,
|
| 415 |
+
"step": 280
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.46,
|
| 419 |
+
"learning_rate": 3.2432444529190714e-06,
|
| 420 |
+
"logits/chosen": 0.15639743208885193,
|
| 421 |
+
"logits/rejected": 0.23564644157886505,
|
| 422 |
+
"logps/chosen": -586.5012817382812,
|
| 423 |
+
"logps/rejected": -683.345947265625,
|
| 424 |
+
"loss": 0.2713,
|
| 425 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 426 |
+
"rewards/chosen": -0.21934108436107635,
|
| 427 |
+
"rewards/margins": 0.14433029294013977,
|
| 428 |
+
"rewards/rejected": -0.36367136240005493,
|
| 429 |
+
"step": 290
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.48,
|
| 433 |
+
"learning_rate": 3.1087212639117057e-06,
|
| 434 |
+
"logits/chosen": 0.1709739863872528,
|
| 435 |
+
"logits/rejected": 0.2245495617389679,
|
| 436 |
+
"logps/chosen": -541.6049194335938,
|
| 437 |
+
"logps/rejected": -656.5075073242188,
|
| 438 |
+
"loss": 0.3202,
|
| 439 |
+
"rewards/accuracies": 0.65625,
|
| 440 |
+
"rewards/chosen": -0.21079179644584656,
|
| 441 |
+
"rewards/margins": 0.11757495254278183,
|
| 442 |
+
"rewards/rejected": -0.3283667266368866,
|
| 443 |
+
"step": 300
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.5,
|
| 447 |
+
"learning_rate": 2.9722964165636263e-06,
|
| 448 |
+
"logits/chosen": 0.14368459582328796,
|
| 449 |
+
"logits/rejected": 0.2023853063583374,
|
| 450 |
+
"logps/chosen": -536.142822265625,
|
| 451 |
+
"logps/rejected": -626.310546875,
|
| 452 |
+
"loss": 0.3197,
|
| 453 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 454 |
+
"rewards/chosen": -0.1972382515668869,
|
| 455 |
+
"rewards/margins": 0.10271339118480682,
|
| 456 |
+
"rewards/rejected": -0.2999516427516937,
|
| 457 |
+
"step": 310
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.51,
|
| 461 |
+
"learning_rate": 2.8343961050366275e-06,
|
| 462 |
+
"logits/chosen": 0.10999581962823868,
|
| 463 |
+
"logits/rejected": 0.19454975426197052,
|
| 464 |
+
"logps/chosen": -585.6015625,
|
| 465 |
+
"logps/rejected": -633.619140625,
|
| 466 |
+
"loss": 0.3227,
|
| 467 |
+
"rewards/accuracies": 0.637499988079071,
|
| 468 |
+
"rewards/chosen": -0.18746989965438843,
|
| 469 |
+
"rewards/margins": 0.09407122433185577,
|
| 470 |
+
"rewards/rejected": -0.2815411686897278,
|
| 471 |
+
"step": 320
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.53,
|
| 475 |
+
"learning_rate": 2.695451132874385e-06,
|
| 476 |
+
"logits/chosen": 0.12884345650672913,
|
| 477 |
+
"logits/rejected": 0.21605506539344788,
|
| 478 |
+
"logps/chosen": -508.71331787109375,
|
| 479 |
+
"logps/rejected": -659.6421508789062,
|
| 480 |
+
"loss": 0.2864,
|
| 481 |
+
"rewards/accuracies": 0.65625,
|
| 482 |
+
"rewards/chosen": -0.17948082089424133,
|
| 483 |
+
"rewards/margins": 0.12334544956684113,
|
| 484 |
+
"rewards/rejected": -0.30282628536224365,
|
| 485 |
+
"step": 330
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.54,
|
| 489 |
+
"learning_rate": 2.5558955671628964e-06,
|
| 490 |
+
"logits/chosen": 0.1270858496427536,
|
| 491 |
+
"logits/rejected": 0.17638283967971802,
|
| 492 |
+
"logps/chosen": -505.85235595703125,
|
| 493 |
+
"logps/rejected": -656.5067138671875,
|
| 494 |
+
"loss": 0.2753,
|
| 495 |
+
"rewards/accuracies": 0.71875,
|
| 496 |
+
"rewards/chosen": -0.1641542613506317,
|
| 497 |
+
"rewards/margins": 0.12927937507629395,
|
| 498 |
+
"rewards/rejected": -0.29343363642692566,
|
| 499 |
+
"step": 340
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.56,
|
| 503 |
+
"learning_rate": 2.4161653824955654e-06,
|
| 504 |
+
"logits/chosen": 0.15247094631195068,
|
| 505 |
+
"logits/rejected": 0.14038711786270142,
|
| 506 |
+
"logps/chosen": -518.3817138671875,
|
| 507 |
+
"logps/rejected": -598.987548828125,
|
| 508 |
+
"loss": 0.2873,
|
| 509 |
+
"rewards/accuracies": 0.675000011920929,
|
| 510 |
+
"rewards/chosen": -0.181033656001091,
|
| 511 |
+
"rewards/margins": 0.09822895377874374,
|
| 512 |
+
"rewards/rejected": -0.27926260232925415,
|
| 513 |
+
"step": 350
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.58,
|
| 517 |
+
"learning_rate": 2.2766970989791697e-06,
|
| 518 |
+
"logits/chosen": 0.14109697937965393,
|
| 519 |
+
"logits/rejected": 0.25879988074302673,
|
| 520 |
+
"logps/chosen": -527.6212158203125,
|
| 521 |
+
"logps/rejected": -657.6229248046875,
|
| 522 |
+
"loss": 0.2737,
|
| 523 |
+
"rewards/accuracies": 0.731249988079071,
|
| 524 |
+
"rewards/chosen": -0.17991268634796143,
|
| 525 |
+
"rewards/margins": 0.14120043814182281,
|
| 526 |
+
"rewards/rejected": -0.32111310958862305,
|
| 527 |
+
"step": 360
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.59,
|
| 531 |
+
"learning_rate": 2.1379264185356545e-06,
|
| 532 |
+
"logits/chosen": 0.18675974011421204,
|
| 533 |
+
"logits/rejected": 0.2627638578414917,
|
| 534 |
+
"logps/chosen": -579.1731567382812,
|
| 535 |
+
"logps/rejected": -669.7673950195312,
|
| 536 |
+
"loss": 0.2839,
|
| 537 |
+
"rewards/accuracies": 0.71875,
|
| 538 |
+
"rewards/chosen": -0.20380540192127228,
|
| 539 |
+
"rewards/margins": 0.12797793745994568,
|
| 540 |
+
"rewards/rejected": -0.33178335428237915,
|
| 541 |
+
"step": 370
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.61,
|
| 545 |
+
"learning_rate": 2.000286863759934e-06,
|
| 546 |
+
"logits/chosen": 0.1522998809814453,
|
| 547 |
+
"logits/rejected": 0.20338353514671326,
|
| 548 |
+
"logps/chosen": -511.1458435058594,
|
| 549 |
+
"logps/rejected": -621.6883544921875,
|
| 550 |
+
"loss": 0.3078,
|
| 551 |
+
"rewards/accuracies": 0.675000011920929,
|
| 552 |
+
"rewards/chosen": -0.1741541475057602,
|
| 553 |
+
"rewards/margins": 0.13219627737998962,
|
| 554 |
+
"rewards/rejected": -0.306350439786911,
|
| 555 |
+
"step": 380
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.62,
|
| 559 |
+
"learning_rate": 1.8642084235859764e-06,
|
| 560 |
+
"logits/chosen": 0.16782529652118683,
|
| 561 |
+
"logits/rejected": 0.25166839361190796,
|
| 562 |
+
"logps/chosen": -559.2919921875,
|
| 563 |
+
"logps/rejected": -642.5906372070312,
|
| 564 |
+
"loss": 0.3013,
|
| 565 |
+
"rewards/accuracies": 0.675000011920929,
|
| 566 |
+
"rewards/chosen": -0.17873121798038483,
|
| 567 |
+
"rewards/margins": 0.09633437544107437,
|
| 568 |
+
"rewards/rejected": -0.2750656008720398,
|
| 569 |
+
"step": 390
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.64,
|
| 573 |
+
"learning_rate": 1.7301162099921013e-06,
|
| 574 |
+
"logits/chosen": 0.18475747108459473,
|
| 575 |
+
"logits/rejected": 0.1979983150959015,
|
| 576 |
+
"logps/chosen": -512.885009765625,
|
| 577 |
+
"logps/rejected": -597.8982543945312,
|
| 578 |
+
"loss": 0.3458,
|
| 579 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 580 |
+
"rewards/chosen": -0.20149023830890656,
|
| 581 |
+
"rewards/margins": 0.0695062130689621,
|
| 582 |
+
"rewards/rejected": -0.2709963917732239,
|
| 583 |
+
"step": 400
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.66,
|
| 587 |
+
"learning_rate": 1.5984291299420117e-06,
|
| 588 |
+
"logits/chosen": 0.13157054781913757,
|
| 589 |
+
"logits/rejected": 0.21159140765666962,
|
| 590 |
+
"logps/chosen": -526.1318359375,
|
| 591 |
+
"logps/rejected": -639.5899658203125,
|
| 592 |
+
"loss": 0.2671,
|
| 593 |
+
"rewards/accuracies": 0.6875,
|
| 594 |
+
"rewards/chosen": -0.1575133055448532,
|
| 595 |
+
"rewards/margins": 0.13058245182037354,
|
| 596 |
+
"rewards/rejected": -0.28809577226638794,
|
| 597 |
+
"step": 410
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.67,
|
| 601 |
+
"learning_rate": 1.4695585767104092e-06,
|
| 602 |
+
"logits/chosen": 0.14387401938438416,
|
| 603 |
+
"logits/rejected": 0.23105745017528534,
|
| 604 |
+
"logps/chosen": -597.7404174804688,
|
| 605 |
+
"logps/rejected": -684.18359375,
|
| 606 |
+
"loss": 0.2713,
|
| 607 |
+
"rewards/accuracies": 0.65625,
|
| 608 |
+
"rewards/chosen": -0.18249225616455078,
|
| 609 |
+
"rewards/margins": 0.12386848032474518,
|
| 610 |
+
"rewards/rejected": -0.30636072158813477,
|
| 611 |
+
"step": 420
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.69,
|
| 615 |
+
"learning_rate": 1.3439071446815452e-06,
|
| 616 |
+
"logits/chosen": 0.11251509189605713,
|
| 617 |
+
"logits/rejected": 0.17909184098243713,
|
| 618 |
+
"logps/chosen": -518.1268920898438,
|
| 619 |
+
"logps/rejected": -573.087890625,
|
| 620 |
+
"loss": 0.3273,
|
| 621 |
+
"rewards/accuracies": 0.59375,
|
| 622 |
+
"rewards/chosen": -0.1990637481212616,
|
| 623 |
+
"rewards/margins": 0.07952968031167984,
|
| 624 |
+
"rewards/rejected": -0.2785934507846832,
|
| 625 |
+
"step": 430
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.7,
|
| 629 |
+
"learning_rate": 1.2218673716356919e-06,
|
| 630 |
+
"logits/chosen": 0.11003521829843521,
|
| 631 |
+
"logits/rejected": 0.21477892994880676,
|
| 632 |
+
"logps/chosen": -500.889404296875,
|
| 633 |
+
"logps/rejected": -645.4175415039062,
|
| 634 |
+
"loss": 0.285,
|
| 635 |
+
"rewards/accuracies": 0.71875,
|
| 636 |
+
"rewards/chosen": -0.1877116858959198,
|
| 637 |
+
"rewards/margins": 0.13334953784942627,
|
| 638 |
+
"rewards/rejected": -0.3210611939430237,
|
| 639 |
+
"step": 440
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.72,
|
| 643 |
+
"learning_rate": 1.103820512452661e-06,
|
| 644 |
+
"logits/chosen": 0.1255572885274887,
|
| 645 |
+
"logits/rejected": 0.22474932670593262,
|
| 646 |
+
"logps/chosen": -600.7024536132812,
|
| 647 |
+
"logps/rejected": -698.1460571289062,
|
| 648 |
+
"loss": 0.2723,
|
| 649 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 650 |
+
"rewards/chosen": -0.19767552614212036,
|
| 651 |
+
"rewards/margins": 0.14586390554904938,
|
| 652 |
+
"rewards/rejected": -0.34353941679000854,
|
| 653 |
+
"step": 450
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.74,
|
| 657 |
+
"learning_rate": 9.901353480633468e-07,
|
| 658 |
+
"logits/chosen": 0.13007649779319763,
|
| 659 |
+
"logits/rejected": 0.18805541098117828,
|
| 660 |
+
"logps/chosen": -527.2013549804688,
|
| 661 |
+
"logps/rejected": -628.3349609375,
|
| 662 |
+
"loss": 0.3286,
|
| 663 |
+
"rewards/accuracies": 0.606249988079071,
|
| 664 |
+
"rewards/chosen": -0.22881360352039337,
|
| 665 |
+
"rewards/margins": 0.09889401495456696,
|
| 666 |
+
"rewards/rejected": -0.32770758867263794,
|
| 667 |
+
"step": 460
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.75,
|
| 671 |
+
"learning_rate": 8.811670333701544e-07,
|
| 672 |
+
"logits/chosen": 0.10513252019882202,
|
| 673 |
+
"logits/rejected": 0.1565053015947342,
|
| 674 |
+
"logps/chosen": -563.048828125,
|
| 675 |
+
"logps/rejected": -633.2500610351562,
|
| 676 |
+
"loss": 0.2992,
|
| 677 |
+
"rewards/accuracies": 0.637499988079071,
|
| 678 |
+
"rewards/chosen": -0.20568466186523438,
|
| 679 |
+
"rewards/margins": 0.10016629844903946,
|
| 680 |
+
"rewards/rejected": -0.3058509826660156,
|
| 681 |
+
"step": 470
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.77,
|
| 685 |
+
"learning_rate": 7.772559877354341e-07,
|
| 686 |
+
"logits/chosen": 0.1835373491048813,
|
| 687 |
+
"logits/rejected": 0.19298210740089417,
|
| 688 |
+
"logps/chosen": -548.7342529296875,
|
| 689 |
+
"logps/rejected": -617.98681640625,
|
| 690 |
+
"loss": 0.3108,
|
| 691 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 692 |
+
"rewards/chosen": -0.20563273131847382,
|
| 693 |
+
"rewards/margins": 0.08818547427654266,
|
| 694 |
+
"rewards/rejected": -0.2938181757926941,
|
| 695 |
+
"step": 480
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.78,
|
| 699 |
+
"learning_rate": 6.787268315040604e-07,
|
| 700 |
+
"logits/chosen": 0.11183663457632065,
|
| 701 |
+
"logits/rejected": 0.1670197993516922,
|
| 702 |
+
"logps/chosen": -581.4060668945312,
|
| 703 |
+
"logps/rejected": -655.2175903320312,
|
| 704 |
+
"loss": 0.323,
|
| 705 |
+
"rewards/accuracies": 0.71875,
|
| 706 |
+
"rewards/chosen": -0.21741382777690887,
|
| 707 |
+
"rewards/margins": 0.0863969624042511,
|
| 708 |
+
"rewards/rejected": -0.3038107752799988,
|
| 709 |
+
"step": 490
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.8,
|
| 713 |
+
"learning_rate": 5.858873718824829e-07,
|
| 714 |
+
"logits/chosen": 0.15304869413375854,
|
| 715 |
+
"logits/rejected": 0.18997374176979065,
|
| 716 |
+
"logps/chosen": -525.7789306640625,
|
| 717 |
+
"logps/rejected": -628.69287109375,
|
| 718 |
+
"loss": 0.2961,
|
| 719 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 720 |
+
"rewards/chosen": -0.188993901014328,
|
| 721 |
+
"rewards/margins": 0.11667201668024063,
|
| 722 |
+
"rewards/rejected": -0.30566591024398804,
|
| 723 |
+
"step": 500
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.82,
|
| 727 |
+
"learning_rate": 4.990276413423817e-07,
|
| 728 |
+
"logits/chosen": 0.1436108499765396,
|
| 729 |
+
"logits/rejected": 0.22793538868427277,
|
| 730 |
+
"logps/chosen": -528.6312866210938,
|
| 731 |
+
"logps/rejected": -641.5194091796875,
|
| 732 |
+
"loss": 0.292,
|
| 733 |
+
"rewards/accuracies": 0.699999988079071,
|
| 734 |
+
"rewards/chosen": -0.18891510367393494,
|
| 735 |
+
"rewards/margins": 0.11953170597553253,
|
| 736 |
+
"rewards/rejected": -0.30844682455062866,
|
| 737 |
+
"step": 510
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.83,
|
| 741 |
+
"learning_rate": 4.184189915529796e-07,
|
| 742 |
+
"logits/chosen": 0.08341806381940842,
|
| 743 |
+
"logits/rejected": 0.24343731999397278,
|
| 744 |
+
"logps/chosen": -505.2992248535156,
|
| 745 |
+
"logps/rejected": -625.7940673828125,
|
| 746 |
+
"loss": 0.3127,
|
| 747 |
+
"rewards/accuracies": 0.675000011920929,
|
| 748 |
+
"rewards/chosen": -0.18587087094783783,
|
| 749 |
+
"rewards/margins": 0.10922978073358536,
|
| 750 |
+
"rewards/rejected": -0.2951006591320038,
|
| 751 |
+
"step": 520
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.85,
|
| 755 |
+
"learning_rate": 3.4431324567258176e-07,
|
| 756 |
+
"logits/chosen": 0.1076665073633194,
|
| 757 |
+
"logits/rejected": 0.16546323895454407,
|
| 758 |
+
"logps/chosen": -564.4067993164062,
|
| 759 |
+
"logps/rejected": -617.0671997070312,
|
| 760 |
+
"loss": 0.3086,
|
| 761 |
+
"rewards/accuracies": 0.643750011920929,
|
| 762 |
+
"rewards/chosen": -0.19417104125022888,
|
| 763 |
+
"rewards/margins": 0.11286197602748871,
|
| 764 |
+
"rewards/rejected": -0.3070330023765564,
|
| 765 |
+
"step": 530
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.86,
|
| 769 |
+
"learning_rate": 2.769419116476052e-07,
|
| 770 |
+
"logits/chosen": 0.14298070967197418,
|
| 771 |
+
"logits/rejected": 0.17989978194236755,
|
| 772 |
+
"logps/chosen": -521.3731689453125,
|
| 773 |
+
"logps/rejected": -603.6255493164062,
|
| 774 |
+
"loss": 0.315,
|
| 775 |
+
"rewards/accuracies": 0.612500011920929,
|
| 776 |
+
"rewards/chosen": -0.19311046600341797,
|
| 777 |
+
"rewards/margins": 0.09155095368623734,
|
| 778 |
+
"rewards/rejected": -0.2846614718437195,
|
| 779 |
+
"step": 540
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.88,
|
| 783 |
+
"learning_rate": 2.1651545897676512e-07,
|
| 784 |
+
"logits/chosen": 0.14457222819328308,
|
| 785 |
+
"logits/rejected": 0.1904611885547638,
|
| 786 |
+
"logps/chosen": -503.63385009765625,
|
| 787 |
+
"logps/rejected": -589.3573608398438,
|
| 788 |
+
"loss": 0.3095,
|
| 789 |
+
"rewards/accuracies": 0.637499988079071,
|
| 790 |
+
"rewards/chosen": -0.1882583498954773,
|
| 791 |
+
"rewards/margins": 0.0975576862692833,
|
| 792 |
+
"rewards/rejected": -0.2858160138130188,
|
| 793 |
+
"step": 550
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.9,
|
| 797 |
+
"learning_rate": 1.6322266119983222e-07,
|
| 798 |
+
"logits/chosen": 0.1182747334241867,
|
| 799 |
+
"logits/rejected": 0.21196098625659943,
|
| 800 |
+
"logps/chosen": -558.52099609375,
|
| 801 |
+
"logps/rejected": -705.6689453125,
|
| 802 |
+
"loss": 0.2741,
|
| 803 |
+
"rewards/accuracies": 0.731249988079071,
|
| 804 |
+
"rewards/chosen": -0.2015405148267746,
|
| 805 |
+
"rewards/margins": 0.12743064761161804,
|
| 806 |
+
"rewards/rejected": -0.32897114753723145,
|
| 807 |
+
"step": 560
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.91,
|
| 811 |
+
"learning_rate": 1.1723000616502167e-07,
|
| 812 |
+
"logits/chosen": 0.13327908515930176,
|
| 813 |
+
"logits/rejected": 0.2085684984922409,
|
| 814 |
+
"logps/chosen": -520.0775146484375,
|
| 815 |
+
"logps/rejected": -619.95703125,
|
| 816 |
+
"loss": 0.2912,
|
| 817 |
+
"rewards/accuracies": 0.668749988079071,
|
| 818 |
+
"rewards/chosen": -0.18114188313484192,
|
| 819 |
+
"rewards/margins": 0.1153794676065445,
|
| 820 |
+
"rewards/rejected": -0.29652130603790283,
|
| 821 |
+
"step": 570
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.93,
|
| 825 |
+
"learning_rate": 7.868117591737585e-08,
|
| 826 |
+
"logits/chosen": 0.11308556795120239,
|
| 827 |
+
"logits/rejected": 0.16310641169548035,
|
| 828 |
+
"logps/chosen": -502.0087890625,
|
| 829 |
+
"logps/rejected": -638.6658935546875,
|
| 830 |
+
"loss": 0.2833,
|
| 831 |
+
"rewards/accuracies": 0.706250011920929,
|
| 832 |
+
"rewards/chosen": -0.1852196305990219,
|
| 833 |
+
"rewards/margins": 0.13148216903209686,
|
| 834 |
+
"rewards/rejected": -0.3167017996311188,
|
| 835 |
+
"step": 580
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.94,
|
| 839 |
+
"learning_rate": 4.769659783295383e-08,
|
| 840 |
+
"logits/chosen": 0.1447245329618454,
|
| 841 |
+
"logits/rejected": 0.17959550023078918,
|
| 842 |
+
"logps/chosen": -533.7676391601562,
|
| 843 |
+
"logps/rejected": -665.7962036132812,
|
| 844 |
+
"loss": 0.275,
|
| 845 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 846 |
+
"rewards/chosen": -0.18203890323638916,
|
| 847 |
+
"rewards/margins": 0.13183028995990753,
|
| 848 |
+
"rewards/rejected": -0.3138691782951355,
|
| 849 |
+
"step": 590
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.96,
|
| 853 |
+
"learning_rate": 2.4373068401120358e-08,
|
| 854 |
+
"logits/chosen": 0.1246277466416359,
|
| 855 |
+
"logits/rejected": 0.19313006103038788,
|
| 856 |
+
"logps/chosen": -532.855224609375,
|
| 857 |
+
"logps/rejected": -646.3125,
|
| 858 |
+
"loss": 0.3061,
|
| 859 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 860 |
+
"rewards/chosen": -0.18818099796772003,
|
| 861 |
+
"rewards/margins": 0.12483028322458267,
|
| 862 |
+
"rewards/rejected": -0.3130112886428833,
|
| 863 |
+
"step": 600
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.98,
|
| 867 |
+
"learning_rate": 8.78345083022425e-09,
|
| 868 |
+
"logits/chosen": 0.07625824213027954,
|
| 869 |
+
"logits/rejected": 0.19986259937286377,
|
| 870 |
+
"logps/chosen": -546.0481567382812,
|
| 871 |
+
"logps/rejected": -647.2523193359375,
|
| 872 |
+
"loss": 0.2853,
|
| 873 |
+
"rewards/accuracies": 0.6875,
|
| 874 |
+
"rewards/chosen": -0.207177072763443,
|
| 875 |
+
"rewards/margins": 0.1102672815322876,
|
| 876 |
+
"rewards/rejected": -0.3174443542957306,
|
| 877 |
+
"step": 610
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 0.99,
|
| 881 |
+
"learning_rate": 9.764474213677654e-10,
|
| 882 |
+
"logits/chosen": 0.08471451699733734,
|
| 883 |
+
"logits/rejected": 0.1842333823442459,
|
| 884 |
+
"logps/chosen": -539.9765014648438,
|
| 885 |
+
"logps/rejected": -655.6905517578125,
|
| 886 |
+
"loss": 0.2766,
|
| 887 |
+
"rewards/accuracies": 0.75,
|
| 888 |
+
"rewards/chosen": -0.18608103692531586,
|
| 889 |
+
"rewards/margins": 0.13810980319976807,
|
| 890 |
+
"rewards/rejected": -0.3241908550262451,
|
| 891 |
+
"step": 620
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 1.0,
|
| 895 |
+
"step": 625,
|
| 896 |
+
"total_flos": 0.0,
|
| 897 |
+
"train_loss": 0.31405294857025146,
|
| 898 |
+
"train_runtime": 5821.3149,
|
| 899 |
+
"train_samples_per_second": 1.718,
|
| 900 |
+
"train_steps_per_second": 0.107
|
| 901 |
+
}
|
| 902 |
+
],
|
| 903 |
+
"logging_steps": 10,
|
| 904 |
+
"max_steps": 625,
|
| 905 |
+
"num_input_tokens_seen": 0,
|
| 906 |
+
"num_train_epochs": 1,
|
| 907 |
+
"save_steps": 100,
|
| 908 |
+
"total_flos": 0.0,
|
| 909 |
+
"train_batch_size": 4,
|
| 910 |
+
"trial_name": null,
|
| 911 |
+
"trial_params": null
|
| 912 |
+
}
|