Model save
Browse files- README.md +62 -0
- adapter_model.safetensors +1 -1
- all_results.json +8 -0
- runs/May17_00-38-25_gpu4-119-5/events.out.tfevents.1715870420.gpu4-119-5.781775.0 +2 -2
- train_results.json +8 -0
- trainer_state.json +912 -0
README.md
ADDED
|
@@ -0,0 +1,62 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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-v36-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-v36-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 |
+
- num_devices: 2
|
| 44 |
+
- gradient_accumulation_steps: 4
|
| 45 |
+
- total_train_batch_size: 32
|
| 46 |
+
- total_eval_batch_size: 8
|
| 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 |
+
- PEFT 0.7.1
|
| 59 |
+
- Transformers 4.36.2
|
| 60 |
+
- Pytorch 2.1.2+cu121
|
| 61 |
+
- Datasets 2.14.6
|
| 62 |
+
- 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 sha256:61c36bae57908807c677a4da395401bc602b50ac81e95f4eb70f1762ab95488e
|
| 3 |
size 167807296
|
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.3311275300979614,
|
| 4 |
+
"train_runtime": 7525.4167,
|
| 5 |
+
"train_samples": 20000,
|
| 6 |
+
"train_samples_per_second": 2.658,
|
| 7 |
+
"train_steps_per_second": 0.083
|
| 8 |
+
}
|
runs/May17_00-38-25_gpu4-119-5/events.out.tfevents.1715870420.gpu4-119-5.781775.0
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:ed8e41187b2d361799b4eec8c3b88ad9c986a9afe2607f0f568fa65cc480062c
|
| 3 |
+
size 44988
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.3311275300979614,
|
| 4 |
+
"train_runtime": 7525.4167,
|
| 5 |
+
"train_samples": 20000,
|
| 6 |
+
"train_samples_per_second": 2.658,
|
| 7 |
+
"train_steps_per_second": 0.083
|
| 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": 0.09552346915006638,
|
| 15 |
+
"logits/rejected": 0.17362232506275177,
|
| 16 |
+
"logps/chosen": -255.44039916992188,
|
| 17 |
+
"logps/rejected": -210.80226135253906,
|
| 18 |
+
"loss": 0.3612,
|
| 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": 0.11144200712442398,
|
| 29 |
+
"logits/rejected": 0.20884405076503754,
|
| 30 |
+
"logps/chosen": -359.81549072265625,
|
| 31 |
+
"logps/rejected": -336.5404968261719,
|
| 32 |
+
"loss": 0.3743,
|
| 33 |
+
"rewards/accuracies": 0.4166666567325592,
|
| 34 |
+
"rewards/chosen": 7.773819379508495e-05,
|
| 35 |
+
"rewards/margins": 3.821194331976585e-05,
|
| 36 |
+
"rewards/rejected": 3.9526246837340295e-05,
|
| 37 |
+
"step": 10
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.03,
|
| 41 |
+
"learning_rate": 1.5873015873015873e-06,
|
| 42 |
+
"logits/chosen": 0.12432358413934708,
|
| 43 |
+
"logits/rejected": 0.2087073028087616,
|
| 44 |
+
"logps/chosen": -346.46246337890625,
|
| 45 |
+
"logps/rejected": -331.729248046875,
|
| 46 |
+
"loss": 0.369,
|
| 47 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 48 |
+
"rewards/chosen": 0.00011548679322004318,
|
| 49 |
+
"rewards/margins": 0.00013375042180996388,
|
| 50 |
+
"rewards/rejected": -1.82636285899207e-05,
|
| 51 |
+
"step": 20
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.05,
|
| 55 |
+
"learning_rate": 2.380952380952381e-06,
|
| 56 |
+
"logits/chosen": 0.1460120528936386,
|
| 57 |
+
"logits/rejected": 0.20255199074745178,
|
| 58 |
+
"logps/chosen": -362.12652587890625,
|
| 59 |
+
"logps/rejected": -310.8555908203125,
|
| 60 |
+
"loss": 0.3676,
|
| 61 |
+
"rewards/accuracies": 0.550000011920929,
|
| 62 |
+
"rewards/chosen": 0.00011238453589612618,
|
| 63 |
+
"rewards/margins": 0.00017934020434040576,
|
| 64 |
+
"rewards/rejected": -6.695566116832197e-05,
|
| 65 |
+
"step": 30
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.06,
|
| 69 |
+
"learning_rate": 3.1746031746031746e-06,
|
| 70 |
+
"logits/chosen": 0.13341203331947327,
|
| 71 |
+
"logits/rejected": 0.2324620932340622,
|
| 72 |
+
"logps/chosen": -350.6195983886719,
|
| 73 |
+
"logps/rejected": -316.5322265625,
|
| 74 |
+
"loss": 0.3806,
|
| 75 |
+
"rewards/accuracies": 0.512499988079071,
|
| 76 |
+
"rewards/chosen": 0.00011245281348237768,
|
| 77 |
+
"rewards/margins": 0.00028168410062789917,
|
| 78 |
+
"rewards/rejected": -0.0001692312944214791,
|
| 79 |
+
"step": 40
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.08,
|
| 83 |
+
"learning_rate": 3.968253968253968e-06,
|
| 84 |
+
"logits/chosen": 0.0949329286813736,
|
| 85 |
+
"logits/rejected": 0.1779319941997528,
|
| 86 |
+
"logps/chosen": -320.3857727050781,
|
| 87 |
+
"logps/rejected": -307.16876220703125,
|
| 88 |
+
"loss": 0.3807,
|
| 89 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 90 |
+
"rewards/chosen": 8.781128417467698e-05,
|
| 91 |
+
"rewards/margins": 0.0011059035314247012,
|
| 92 |
+
"rewards/rejected": -0.001018092269077897,
|
| 93 |
+
"step": 50
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.1,
|
| 97 |
+
"learning_rate": 4.761904761904762e-06,
|
| 98 |
+
"logits/chosen": 0.09137465059757233,
|
| 99 |
+
"logits/rejected": 0.20532509684562683,
|
| 100 |
+
"logps/chosen": -352.49993896484375,
|
| 101 |
+
"logps/rejected": -326.77874755859375,
|
| 102 |
+
"loss": 0.3651,
|
| 103 |
+
"rewards/accuracies": 0.5625,
|
| 104 |
+
"rewards/chosen": 0.00026215435354970396,
|
| 105 |
+
"rewards/margins": 0.001010752865113318,
|
| 106 |
+
"rewards/rejected": -0.0007485984242521226,
|
| 107 |
+
"step": 60
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.11,
|
| 111 |
+
"learning_rate": 4.998086282661188e-06,
|
| 112 |
+
"logits/chosen": 0.045028798282146454,
|
| 113 |
+
"logits/rejected": 0.17268504202365875,
|
| 114 |
+
"logps/chosen": -356.53717041015625,
|
| 115 |
+
"logps/rejected": -328.4739074707031,
|
| 116 |
+
"loss": 0.3691,
|
| 117 |
+
"rewards/accuracies": 0.6000000238418579,
|
| 118 |
+
"rewards/chosen": 0.0009223738452419639,
|
| 119 |
+
"rewards/margins": 0.0022405553609132767,
|
| 120 |
+
"rewards/rejected": -0.0013181815156713128,
|
| 121 |
+
"step": 70
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.13,
|
| 125 |
+
"learning_rate": 4.988720025682995e-06,
|
| 126 |
+
"logits/chosen": 0.1980675309896469,
|
| 127 |
+
"logits/rejected": 0.18946149945259094,
|
| 128 |
+
"logps/chosen": -318.8916015625,
|
| 129 |
+
"logps/rejected": -316.4183044433594,
|
| 130 |
+
"loss": 0.3524,
|
| 131 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 132 |
+
"rewards/chosen": 0.0015520222950726748,
|
| 133 |
+
"rewards/margins": 0.006260824855417013,
|
| 134 |
+
"rewards/rejected": -0.004708803258836269,
|
| 135 |
+
"step": 80
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.14,
|
| 139 |
+
"learning_rate": 4.9715789537359126e-06,
|
| 140 |
+
"logits/chosen": 0.1147293671965599,
|
| 141 |
+
"logits/rejected": 0.2110525667667389,
|
| 142 |
+
"logps/chosen": -335.58905029296875,
|
| 143 |
+
"logps/rejected": -333.1260681152344,
|
| 144 |
+
"loss": 0.3675,
|
| 145 |
+
"rewards/accuracies": 0.543749988079071,
|
| 146 |
+
"rewards/chosen": -0.002089377725496888,
|
| 147 |
+
"rewards/margins": 0.0062202513217926025,
|
| 148 |
+
"rewards/rejected": -0.008309627883136272,
|
| 149 |
+
"step": 90
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.16,
|
| 153 |
+
"learning_rate": 4.946716615897932e-06,
|
| 154 |
+
"logits/chosen": 0.09352072328329086,
|
| 155 |
+
"logits/rejected": 0.131501242518425,
|
| 156 |
+
"logps/chosen": -339.08734130859375,
|
| 157 |
+
"logps/rejected": -325.3585205078125,
|
| 158 |
+
"loss": 0.3558,
|
| 159 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 160 |
+
"rewards/chosen": -0.008228459395468235,
|
| 161 |
+
"rewards/margins": 0.010374611243605614,
|
| 162 |
+
"rewards/rejected": -0.018603071570396423,
|
| 163 |
+
"step": 100
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.18,
|
| 167 |
+
"learning_rate": 4.9142106826480114e-06,
|
| 168 |
+
"logits/chosen": -0.004745665937662125,
|
| 169 |
+
"logits/rejected": 0.03175293654203415,
|
| 170 |
+
"logps/chosen": -348.41973876953125,
|
| 171 |
+
"logps/rejected": -358.1392517089844,
|
| 172 |
+
"loss": 0.3559,
|
| 173 |
+
"rewards/accuracies": 0.518750011920929,
|
| 174 |
+
"rewards/chosen": -0.028692331165075302,
|
| 175 |
+
"rewards/margins": 0.02663610503077507,
|
| 176 |
+
"rewards/rejected": -0.05532843619585037,
|
| 177 |
+
"step": 110
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.19,
|
| 181 |
+
"learning_rate": 4.874162703221823e-06,
|
| 182 |
+
"logits/chosen": 0.09396891295909882,
|
| 183 |
+
"logits/rejected": 0.14528635144233704,
|
| 184 |
+
"logps/chosen": -388.5705261230469,
|
| 185 |
+
"logps/rejected": -387.454833984375,
|
| 186 |
+
"loss": 0.3569,
|
| 187 |
+
"rewards/accuracies": 0.574999988079071,
|
| 188 |
+
"rewards/chosen": -0.04682096466422081,
|
| 189 |
+
"rewards/margins": 0.02794179879128933,
|
| 190 |
+
"rewards/rejected": -0.0747627541422844,
|
| 191 |
+
"step": 120
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.21,
|
| 195 |
+
"learning_rate": 4.826697788369752e-06,
|
| 196 |
+
"logits/chosen": 0.06088032200932503,
|
| 197 |
+
"logits/rejected": 0.09309231489896774,
|
| 198 |
+
"logps/chosen": -444.643798828125,
|
| 199 |
+
"logps/rejected": -433.63409423828125,
|
| 200 |
+
"loss": 0.3579,
|
| 201 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 202 |
+
"rewards/chosen": -0.07657051086425781,
|
| 203 |
+
"rewards/margins": 0.03327987343072891,
|
| 204 |
+
"rewards/rejected": -0.10985038429498672,
|
| 205 |
+
"step": 130
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.22,
|
| 209 |
+
"learning_rate": 4.7719642195082224e-06,
|
| 210 |
+
"logits/chosen": -0.0028548731934279203,
|
| 211 |
+
"logits/rejected": 0.0651206225156784,
|
| 212 |
+
"logps/chosen": -398.1200256347656,
|
| 213 |
+
"logps/rejected": -429.90447998046875,
|
| 214 |
+
"loss": 0.3439,
|
| 215 |
+
"rewards/accuracies": 0.5625,
|
| 216 |
+
"rewards/chosen": -0.07342123985290527,
|
| 217 |
+
"rewards/margins": 0.04399186372756958,
|
| 218 |
+
"rewards/rejected": -0.11741310358047485,
|
| 219 |
+
"step": 140
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.24,
|
| 223 |
+
"learning_rate": 4.710132985485355e-06,
|
| 224 |
+
"logits/chosen": 0.0589798204600811,
|
| 225 |
+
"logits/rejected": 0.0974341481924057,
|
| 226 |
+
"logps/chosen": -432.10809326171875,
|
| 227 |
+
"logps/rejected": -462.78778076171875,
|
| 228 |
+
"loss": 0.3586,
|
| 229 |
+
"rewards/accuracies": 0.5687500238418579,
|
| 230 |
+
"rewards/chosen": -0.09363020956516266,
|
| 231 |
+
"rewards/margins": 0.04327515512704849,
|
| 232 |
+
"rewards/rejected": -0.13690535724163055,
|
| 233 |
+
"step": 150
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.26,
|
| 237 |
+
"learning_rate": 4.641397248408122e-06,
|
| 238 |
+
"logits/chosen": 0.06426632404327393,
|
| 239 |
+
"logits/rejected": 0.1016223207116127,
|
| 240 |
+
"logps/chosen": -401.81256103515625,
|
| 241 |
+
"logps/rejected": -423.08984375,
|
| 242 |
+
"loss": 0.3355,
|
| 243 |
+
"rewards/accuracies": 0.625,
|
| 244 |
+
"rewards/chosen": -0.074070505797863,
|
| 245 |
+
"rewards/margins": 0.038354430347681046,
|
| 246 |
+
"rewards/rejected": -0.11242493242025375,
|
| 247 |
+
"step": 160
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.27,
|
| 251 |
+
"learning_rate": 4.5659717401997655e-06,
|
| 252 |
+
"logits/chosen": 0.06136215850710869,
|
| 253 |
+
"logits/rejected": 0.07516863942146301,
|
| 254 |
+
"logps/chosen": -438.4203186035156,
|
| 255 |
+
"logps/rejected": -483.72845458984375,
|
| 256 |
+
"loss": 0.3374,
|
| 257 |
+
"rewards/accuracies": 0.675000011920929,
|
| 258 |
+
"rewards/chosen": -0.09795868396759033,
|
| 259 |
+
"rewards/margins": 0.06916960328817368,
|
| 260 |
+
"rewards/rejected": -0.1671282798051834,
|
| 261 |
+
"step": 170
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.29,
|
| 265 |
+
"learning_rate": 4.4840920917726425e-06,
|
| 266 |
+
"logits/chosen": 0.07146959006786346,
|
| 267 |
+
"logits/rejected": 0.13864199817180634,
|
| 268 |
+
"logps/chosen": -459.9623107910156,
|
| 269 |
+
"logps/rejected": -529.416015625,
|
| 270 |
+
"loss": 0.321,
|
| 271 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 272 |
+
"rewards/chosen": -0.1282103806734085,
|
| 273 |
+
"rewards/margins": 0.07925314456224442,
|
| 274 |
+
"rewards/rejected": -0.20746353268623352,
|
| 275 |
+
"step": 180
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.3,
|
| 279 |
+
"learning_rate": 4.396014096912182e-06,
|
| 280 |
+
"logits/chosen": 0.041384804993867874,
|
| 281 |
+
"logits/rejected": 0.0659438818693161,
|
| 282 |
+
"logps/chosen": -481.56793212890625,
|
| 283 |
+
"logps/rejected": -541.1165771484375,
|
| 284 |
+
"loss": 0.3344,
|
| 285 |
+
"rewards/accuracies": 0.625,
|
| 286 |
+
"rewards/chosen": -0.14985546469688416,
|
| 287 |
+
"rewards/margins": 0.07496772706508636,
|
| 288 |
+
"rewards/rejected": -0.22482319176197052,
|
| 289 |
+
"step": 190
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.32,
|
| 293 |
+
"learning_rate": 4.302012913171584e-06,
|
| 294 |
+
"logits/chosen": 0.04430466145277023,
|
| 295 |
+
"logits/rejected": 0.11885523796081543,
|
| 296 |
+
"logps/chosen": -511.34857177734375,
|
| 297 |
+
"logps/rejected": -522.5247192382812,
|
| 298 |
+
"loss": 0.3249,
|
| 299 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 300 |
+
"rewards/chosen": -0.11126357316970825,
|
| 301 |
+
"rewards/margins": 0.06654877960681915,
|
| 302 |
+
"rewards/rejected": -0.1778123527765274,
|
| 303 |
+
"step": 200
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.34,
|
| 307 |
+
"learning_rate": 4.202382202273702e-06,
|
| 308 |
+
"logits/chosen": 0.05428556352853775,
|
| 309 |
+
"logits/rejected": 0.15545043349266052,
|
| 310 |
+
"logps/chosen": -450.3092346191406,
|
| 311 |
+
"logps/rejected": -418.51678466796875,
|
| 312 |
+
"loss": 0.3455,
|
| 313 |
+
"rewards/accuracies": 0.581250011920929,
|
| 314 |
+
"rewards/chosen": -0.1011916771531105,
|
| 315 |
+
"rewards/margins": 0.02224273420870304,
|
| 316 |
+
"rewards/rejected": -0.1234344020485878,
|
| 317 |
+
"step": 210
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.35,
|
| 321 |
+
"learning_rate": 4.097433212705492e-06,
|
| 322 |
+
"logits/chosen": 0.10550371557474136,
|
| 323 |
+
"logits/rejected": 0.1396268755197525,
|
| 324 |
+
"logps/chosen": -428.6062927246094,
|
| 325 |
+
"logps/rejected": -454.2294921875,
|
| 326 |
+
"loss": 0.3386,
|
| 327 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 328 |
+
"rewards/chosen": -0.08421845734119415,
|
| 329 |
+
"rewards/margins": 0.04463706165552139,
|
| 330 |
+
"rewards/rejected": -0.12885551154613495,
|
| 331 |
+
"step": 220
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.37,
|
| 335 |
+
"learning_rate": 3.987493807371033e-06,
|
| 336 |
+
"logits/chosen": 0.1371748000383377,
|
| 337 |
+
"logits/rejected": 0.1429559886455536,
|
| 338 |
+
"logps/chosen": -413.91241455078125,
|
| 339 |
+
"logps/rejected": -443.8023376464844,
|
| 340 |
+
"loss": 0.3146,
|
| 341 |
+
"rewards/accuracies": 0.637499988079071,
|
| 342 |
+
"rewards/chosen": -0.08825449645519257,
|
| 343 |
+
"rewards/margins": 0.05310072749853134,
|
| 344 |
+
"rewards/rejected": -0.1413552314043045,
|
| 345 |
+
"step": 230
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.38,
|
| 349 |
+
"learning_rate": 3.872907439340758e-06,
|
| 350 |
+
"logits/chosen": 0.13840351998806,
|
| 351 |
+
"logits/rejected": 0.1320694237947464,
|
| 352 |
+
"logps/chosen": -445.927490234375,
|
| 353 |
+
"logps/rejected": -498.40325927734375,
|
| 354 |
+
"loss": 0.3388,
|
| 355 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 356 |
+
"rewards/chosen": -0.10737401247024536,
|
| 357 |
+
"rewards/margins": 0.05018274113535881,
|
| 358 |
+
"rewards/rejected": -0.15755674242973328,
|
| 359 |
+
"step": 240
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.4,
|
| 363 |
+
"learning_rate": 3.75403207889666e-06,
|
| 364 |
+
"logits/chosen": 0.0929916724562645,
|
| 365 |
+
"logits/rejected": 0.1281127631664276,
|
| 366 |
+
"logps/chosen": -415.629150390625,
|
| 367 |
+
"logps/rejected": -462.8294372558594,
|
| 368 |
+
"loss": 0.3343,
|
| 369 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 370 |
+
"rewards/chosen": -0.10600423812866211,
|
| 371 |
+
"rewards/margins": 0.05615564063191414,
|
| 372 |
+
"rewards/rejected": -0.16215987503528595,
|
| 373 |
+
"step": 250
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 0.42,
|
| 377 |
+
"learning_rate": 3.631239095225417e-06,
|
| 378 |
+
"logits/chosen": 0.07575414329767227,
|
| 379 |
+
"logits/rejected": 0.18414750695228577,
|
| 380 |
+
"logps/chosen": -428.9398498535156,
|
| 381 |
+
"logps/rejected": -469.5541076660156,
|
| 382 |
+
"loss": 0.322,
|
| 383 |
+
"rewards/accuracies": 0.637499988079071,
|
| 384 |
+
"rewards/chosen": -0.11417442560195923,
|
| 385 |
+
"rewards/margins": 0.0532112643122673,
|
| 386 |
+
"rewards/rejected": -0.16738571226596832,
|
| 387 |
+
"step": 260
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.43,
|
| 391 |
+
"learning_rate": 3.5049120962530608e-06,
|
| 392 |
+
"logits/chosen": 0.08079143613576889,
|
| 393 |
+
"logits/rejected": 0.14295849204063416,
|
| 394 |
+
"logps/chosen": -512.96533203125,
|
| 395 |
+
"logps/rejected": -558.752685546875,
|
| 396 |
+
"loss": 0.3222,
|
| 397 |
+
"rewards/accuracies": 0.59375,
|
| 398 |
+
"rewards/chosen": -0.16307619214057922,
|
| 399 |
+
"rewards/margins": 0.07023187726736069,
|
| 400 |
+
"rewards/rejected": -0.2333080768585205,
|
| 401 |
+
"step": 270
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.45,
|
| 405 |
+
"learning_rate": 3.3754457302455464e-06,
|
| 406 |
+
"logits/chosen": 0.07979480177164078,
|
| 407 |
+
"logits/rejected": 0.17251968383789062,
|
| 408 |
+
"logps/chosen": -526.539794921875,
|
| 409 |
+
"logps/rejected": -628.0927124023438,
|
| 410 |
+
"loss": 0.3058,
|
| 411 |
+
"rewards/accuracies": 0.637499988079071,
|
| 412 |
+
"rewards/chosen": -0.1753678023815155,
|
| 413 |
+
"rewards/margins": 0.09301600605249405,
|
| 414 |
+
"rewards/rejected": -0.26838380098342896,
|
| 415 |
+
"step": 280
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.46,
|
| 419 |
+
"learning_rate": 3.2432444529190714e-06,
|
| 420 |
+
"logits/chosen": 0.09937838464975357,
|
| 421 |
+
"logits/rejected": 0.16915322840213776,
|
| 422 |
+
"logps/chosen": -547.1159057617188,
|
| 423 |
+
"logps/rejected": -621.634765625,
|
| 424 |
+
"loss": 0.316,
|
| 425 |
+
"rewards/accuracies": 0.643750011920929,
|
| 426 |
+
"rewards/chosen": -0.17411155998706818,
|
| 427 |
+
"rewards/margins": 0.09796580672264099,
|
| 428 |
+
"rewards/rejected": -0.27207738161087036,
|
| 429 |
+
"step": 290
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 0.48,
|
| 433 |
+
"learning_rate": 3.1087212639117057e-06,
|
| 434 |
+
"logits/chosen": 0.09575396776199341,
|
| 435 |
+
"logits/rejected": 0.11415497213602066,
|
| 436 |
+
"logps/chosen": -475.705078125,
|
| 437 |
+
"logps/rejected": -556.6187744140625,
|
| 438 |
+
"loss": 0.323,
|
| 439 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 440 |
+
"rewards/chosen": -0.15034952759742737,
|
| 441 |
+
"rewards/margins": 0.0816831961274147,
|
| 442 |
+
"rewards/rejected": -0.23203274607658386,
|
| 443 |
+
"step": 300
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.5,
|
| 447 |
+
"learning_rate": 2.9722964165636263e-06,
|
| 448 |
+
"logits/chosen": 0.1228911504149437,
|
| 449 |
+
"logits/rejected": 0.18562354147434235,
|
| 450 |
+
"logps/chosen": -498.4169921875,
|
| 451 |
+
"logps/rejected": -489.9468688964844,
|
| 452 |
+
"loss": 0.34,
|
| 453 |
+
"rewards/accuracies": 0.637499988079071,
|
| 454 |
+
"rewards/chosen": -0.1489831507205963,
|
| 455 |
+
"rewards/margins": 0.03880687803030014,
|
| 456 |
+
"rewards/rejected": -0.18779003620147705,
|
| 457 |
+
"step": 310
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.51,
|
| 461 |
+
"learning_rate": 2.8343961050366275e-06,
|
| 462 |
+
"logits/chosen": 0.09400780498981476,
|
| 463 |
+
"logits/rejected": 0.1238022893667221,
|
| 464 |
+
"logps/chosen": -470.9700622558594,
|
| 465 |
+
"logps/rejected": -553.8331909179688,
|
| 466 |
+
"loss": 0.3093,
|
| 467 |
+
"rewards/accuracies": 0.699999988079071,
|
| 468 |
+
"rewards/chosen": -0.12979908287525177,
|
| 469 |
+
"rewards/margins": 0.09255190938711166,
|
| 470 |
+
"rewards/rejected": -0.22235099971294403,
|
| 471 |
+
"step": 320
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 0.53,
|
| 475 |
+
"learning_rate": 2.695451132874385e-06,
|
| 476 |
+
"logits/chosen": 0.12480980157852173,
|
| 477 |
+
"logits/rejected": 0.17045611143112183,
|
| 478 |
+
"logps/chosen": -504.2953186035156,
|
| 479 |
+
"logps/rejected": -529.3466796875,
|
| 480 |
+
"loss": 0.3287,
|
| 481 |
+
"rewards/accuracies": 0.59375,
|
| 482 |
+
"rewards/chosen": -0.1506374180316925,
|
| 483 |
+
"rewards/margins": 0.05277082324028015,
|
| 484 |
+
"rewards/rejected": -0.20340824127197266,
|
| 485 |
+
"step": 330
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.54,
|
| 489 |
+
"learning_rate": 2.5558955671628964e-06,
|
| 490 |
+
"logits/chosen": 0.0985703319311142,
|
| 491 |
+
"logits/rejected": 0.2024538218975067,
|
| 492 |
+
"logps/chosen": -503.4974670410156,
|
| 493 |
+
"logps/rejected": -588.9210815429688,
|
| 494 |
+
"loss": 0.3256,
|
| 495 |
+
"rewards/accuracies": 0.6875,
|
| 496 |
+
"rewards/chosen": -0.16819895803928375,
|
| 497 |
+
"rewards/margins": 0.09579765796661377,
|
| 498 |
+
"rewards/rejected": -0.2639966309070587,
|
| 499 |
+
"step": 340
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 0.56,
|
| 503 |
+
"learning_rate": 2.4161653824955654e-06,
|
| 504 |
+
"logits/chosen": 0.11853840202093124,
|
| 505 |
+
"logits/rejected": 0.14462777972221375,
|
| 506 |
+
"logps/chosen": -469.1441955566406,
|
| 507 |
+
"logps/rejected": -537.0656127929688,
|
| 508 |
+
"loss": 0.3294,
|
| 509 |
+
"rewards/accuracies": 0.65625,
|
| 510 |
+
"rewards/chosen": -0.15223057568073273,
|
| 511 |
+
"rewards/margins": 0.08282224088907242,
|
| 512 |
+
"rewards/rejected": -0.23505279421806335,
|
| 513 |
+
"step": 350
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.58,
|
| 517 |
+
"learning_rate": 2.2766970989791697e-06,
|
| 518 |
+
"logits/chosen": 0.153191938996315,
|
| 519 |
+
"logits/rejected": 0.22593048214912415,
|
| 520 |
+
"logps/chosen": -507.7642517089844,
|
| 521 |
+
"logps/rejected": -573.9854736328125,
|
| 522 |
+
"loss": 0.2987,
|
| 523 |
+
"rewards/accuracies": 0.675000011920929,
|
| 524 |
+
"rewards/chosen": -0.15160496532917023,
|
| 525 |
+
"rewards/margins": 0.10073844343423843,
|
| 526 |
+
"rewards/rejected": -0.25234344601631165,
|
| 527 |
+
"step": 360
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.59,
|
| 531 |
+
"learning_rate": 2.1379264185356545e-06,
|
| 532 |
+
"logits/chosen": 0.1527024805545807,
|
| 533 |
+
"logits/rejected": 0.127252459526062,
|
| 534 |
+
"logps/chosen": -481.59722900390625,
|
| 535 |
+
"logps/rejected": -552.0794067382812,
|
| 536 |
+
"loss": 0.3347,
|
| 537 |
+
"rewards/accuracies": 0.606249988079071,
|
| 538 |
+
"rewards/chosen": -0.17246612906455994,
|
| 539 |
+
"rewards/margins": 0.06559783220291138,
|
| 540 |
+
"rewards/rejected": -0.23806393146514893,
|
| 541 |
+
"step": 370
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.61,
|
| 545 |
+
"learning_rate": 2.000286863759934e-06,
|
| 546 |
+
"logits/chosen": 0.14587077498435974,
|
| 547 |
+
"logits/rejected": 0.14887812733650208,
|
| 548 |
+
"logps/chosen": -488.64801025390625,
|
| 549 |
+
"logps/rejected": -543.491455078125,
|
| 550 |
+
"loss": 0.3183,
|
| 551 |
+
"rewards/accuracies": 0.699999988079071,
|
| 552 |
+
"rewards/chosen": -0.16818055510520935,
|
| 553 |
+
"rewards/margins": 0.08099476993083954,
|
| 554 |
+
"rewards/rejected": -0.24917533993721008,
|
| 555 |
+
"step": 380
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.62,
|
| 559 |
+
"learning_rate": 1.8642084235859764e-06,
|
| 560 |
+
"logits/chosen": 0.12905414402484894,
|
| 561 |
+
"logits/rejected": 0.17015649378299713,
|
| 562 |
+
"logps/chosen": -498.19451904296875,
|
| 563 |
+
"logps/rejected": -579.4054565429688,
|
| 564 |
+
"loss": 0.366,
|
| 565 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 566 |
+
"rewards/chosen": -0.18893127143383026,
|
| 567 |
+
"rewards/margins": 0.05997669696807861,
|
| 568 |
+
"rewards/rejected": -0.24890796840190887,
|
| 569 |
+
"step": 390
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.64,
|
| 573 |
+
"learning_rate": 1.7301162099921013e-06,
|
| 574 |
+
"logits/chosen": 0.15795116126537323,
|
| 575 |
+
"logits/rejected": 0.23071245849132538,
|
| 576 |
+
"logps/chosen": -514.3444213867188,
|
| 577 |
+
"logps/rejected": -514.5433349609375,
|
| 578 |
+
"loss": 0.3127,
|
| 579 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 580 |
+
"rewards/chosen": -0.12764492630958557,
|
| 581 |
+
"rewards/margins": 0.07431678473949432,
|
| 582 |
+
"rewards/rejected": -0.2019617259502411,
|
| 583 |
+
"step": 400
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 0.66,
|
| 587 |
+
"learning_rate": 1.5984291299420117e-06,
|
| 588 |
+
"logits/chosen": 0.17297211289405823,
|
| 589 |
+
"logits/rejected": 0.19146080315113068,
|
| 590 |
+
"logps/chosen": -451.94317626953125,
|
| 591 |
+
"logps/rejected": -519.0740966796875,
|
| 592 |
+
"loss": 0.3096,
|
| 593 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 594 |
+
"rewards/chosen": -0.12079250812530518,
|
| 595 |
+
"rewards/margins": 0.07710902392864227,
|
| 596 |
+
"rewards/rejected": -0.19790153205394745,
|
| 597 |
+
"step": 410
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.67,
|
| 601 |
+
"learning_rate": 1.4695585767104092e-06,
|
| 602 |
+
"logits/chosen": 0.1951906979084015,
|
| 603 |
+
"logits/rejected": 0.24034972488880157,
|
| 604 |
+
"logps/chosen": -488.6905212402344,
|
| 605 |
+
"logps/rejected": -560.0809936523438,
|
| 606 |
+
"loss": 0.3168,
|
| 607 |
+
"rewards/accuracies": 0.699999988079071,
|
| 608 |
+
"rewards/chosen": -0.1418723165988922,
|
| 609 |
+
"rewards/margins": 0.07965030521154404,
|
| 610 |
+
"rewards/rejected": -0.22152259945869446,
|
| 611 |
+
"step": 420
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 0.69,
|
| 615 |
+
"learning_rate": 1.3439071446815452e-06,
|
| 616 |
+
"logits/chosen": 0.16284561157226562,
|
| 617 |
+
"logits/rejected": 0.18018727004528046,
|
| 618 |
+
"logps/chosen": -507.633056640625,
|
| 619 |
+
"logps/rejected": -534.8673095703125,
|
| 620 |
+
"loss": 0.3152,
|
| 621 |
+
"rewards/accuracies": 0.625,
|
| 622 |
+
"rewards/chosen": -0.1576879769563675,
|
| 623 |
+
"rewards/margins": 0.06594133377075195,
|
| 624 |
+
"rewards/rejected": -0.22362928092479706,
|
| 625 |
+
"step": 430
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.7,
|
| 629 |
+
"learning_rate": 1.2218673716356919e-06,
|
| 630 |
+
"logits/chosen": 0.1776401400566101,
|
| 631 |
+
"logits/rejected": 0.2645563781261444,
|
| 632 |
+
"logps/chosen": -505.73052978515625,
|
| 633 |
+
"logps/rejected": -560.3125,
|
| 634 |
+
"loss": 0.327,
|
| 635 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 636 |
+
"rewards/chosen": -0.16492930054664612,
|
| 637 |
+
"rewards/margins": 0.060159534215927124,
|
| 638 |
+
"rewards/rejected": -0.22508880496025085,
|
| 639 |
+
"step": 440
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.72,
|
| 643 |
+
"learning_rate": 1.103820512452661e-06,
|
| 644 |
+
"logits/chosen": 0.1745072603225708,
|
| 645 |
+
"logits/rejected": 0.22326946258544922,
|
| 646 |
+
"logps/chosen": -491.3525390625,
|
| 647 |
+
"logps/rejected": -580.8945922851562,
|
| 648 |
+
"loss": 0.3292,
|
| 649 |
+
"rewards/accuracies": 0.65625,
|
| 650 |
+
"rewards/chosen": -0.1665906310081482,
|
| 651 |
+
"rewards/margins": 0.07879535853862762,
|
| 652 |
+
"rewards/rejected": -0.24538597464561462,
|
| 653 |
+
"step": 450
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.74,
|
| 657 |
+
"learning_rate": 9.901353480633468e-07,
|
| 658 |
+
"logits/chosen": 0.2245282232761383,
|
| 659 |
+
"logits/rejected": 0.21928434073925018,
|
| 660 |
+
"logps/chosen": -502.446044921875,
|
| 661 |
+
"logps/rejected": -564.1083374023438,
|
| 662 |
+
"loss": 0.3117,
|
| 663 |
+
"rewards/accuracies": 0.637499988079071,
|
| 664 |
+
"rewards/chosen": -0.15784719586372375,
|
| 665 |
+
"rewards/margins": 0.06698840856552124,
|
| 666 |
+
"rewards/rejected": -0.224835604429245,
|
| 667 |
+
"step": 460
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 0.75,
|
| 671 |
+
"learning_rate": 8.811670333701544e-07,
|
| 672 |
+
"logits/chosen": 0.17931757867336273,
|
| 673 |
+
"logits/rejected": 0.19498419761657715,
|
| 674 |
+
"logps/chosen": -488.80633544921875,
|
| 675 |
+
"logps/rejected": -543.345458984375,
|
| 676 |
+
"loss": 0.306,
|
| 677 |
+
"rewards/accuracies": 0.625,
|
| 678 |
+
"rewards/chosen": -0.15909579396247864,
|
| 679 |
+
"rewards/margins": 0.06999148428440094,
|
| 680 |
+
"rewards/rejected": -0.22908727824687958,
|
| 681 |
+
"step": 470
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 0.77,
|
| 685 |
+
"learning_rate": 7.772559877354341e-07,
|
| 686 |
+
"logits/chosen": 0.18104666471481323,
|
| 687 |
+
"logits/rejected": 0.2650128901004791,
|
| 688 |
+
"logps/chosen": -486.8435974121094,
|
| 689 |
+
"logps/rejected": -529.2096557617188,
|
| 690 |
+
"loss": 0.3362,
|
| 691 |
+
"rewards/accuracies": 0.5874999761581421,
|
| 692 |
+
"rewards/chosen": -0.15709495544433594,
|
| 693 |
+
"rewards/margins": 0.06306286156177521,
|
| 694 |
+
"rewards/rejected": -0.22015781700611115,
|
| 695 |
+
"step": 480
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.78,
|
| 699 |
+
"learning_rate": 6.787268315040604e-07,
|
| 700 |
+
"logits/chosen": 0.17338337004184723,
|
| 701 |
+
"logits/rejected": 0.21364276111125946,
|
| 702 |
+
"logps/chosen": -494.8131408691406,
|
| 703 |
+
"logps/rejected": -559.9661865234375,
|
| 704 |
+
"loss": 0.3009,
|
| 705 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 706 |
+
"rewards/chosen": -0.16898290812969208,
|
| 707 |
+
"rewards/margins": 0.07606662809848785,
|
| 708 |
+
"rewards/rejected": -0.24504955112934113,
|
| 709 |
+
"step": 490
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 0.8,
|
| 713 |
+
"learning_rate": 5.858873718824829e-07,
|
| 714 |
+
"logits/chosen": 0.18185105919837952,
|
| 715 |
+
"logits/rejected": 0.24735283851623535,
|
| 716 |
+
"logps/chosen": -555.6572875976562,
|
| 717 |
+
"logps/rejected": -560.2110595703125,
|
| 718 |
+
"loss": 0.3324,
|
| 719 |
+
"rewards/accuracies": 0.612500011920929,
|
| 720 |
+
"rewards/chosen": -0.17251154780387878,
|
| 721 |
+
"rewards/margins": 0.05710861086845398,
|
| 722 |
+
"rewards/rejected": -0.22962014377117157,
|
| 723 |
+
"step": 500
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 0.82,
|
| 727 |
+
"learning_rate": 4.990276413423817e-07,
|
| 728 |
+
"logits/chosen": 0.17972733080387115,
|
| 729 |
+
"logits/rejected": 0.22244539856910706,
|
| 730 |
+
"logps/chosen": -524.2530517578125,
|
| 731 |
+
"logps/rejected": -599.024658203125,
|
| 732 |
+
"loss": 0.3117,
|
| 733 |
+
"rewards/accuracies": 0.668749988079071,
|
| 734 |
+
"rewards/chosen": -0.161757230758667,
|
| 735 |
+
"rewards/margins": 0.09672929346561432,
|
| 736 |
+
"rewards/rejected": -0.2584865093231201,
|
| 737 |
+
"step": 510
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.83,
|
| 741 |
+
"learning_rate": 4.184189915529796e-07,
|
| 742 |
+
"logits/chosen": 0.10895649343729019,
|
| 743 |
+
"logits/rejected": 0.20955803990364075,
|
| 744 |
+
"logps/chosen": -491.8091735839844,
|
| 745 |
+
"logps/rejected": -556.9285278320312,
|
| 746 |
+
"loss": 0.3011,
|
| 747 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 748 |
+
"rewards/chosen": -0.15785539150238037,
|
| 749 |
+
"rewards/margins": 0.08289924263954163,
|
| 750 |
+
"rewards/rejected": -0.240754634141922,
|
| 751 |
+
"step": 520
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 0.85,
|
| 755 |
+
"learning_rate": 3.4431324567258176e-07,
|
| 756 |
+
"logits/chosen": 0.19427216053009033,
|
| 757 |
+
"logits/rejected": 0.2759999632835388,
|
| 758 |
+
"logps/chosen": -519.3431396484375,
|
| 759 |
+
"logps/rejected": -586.2860107421875,
|
| 760 |
+
"loss": 0.2963,
|
| 761 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 762 |
+
"rewards/chosen": -0.15744614601135254,
|
| 763 |
+
"rewards/margins": 0.1040191501379013,
|
| 764 |
+
"rewards/rejected": -0.26146528124809265,
|
| 765 |
+
"step": 530
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.86,
|
| 769 |
+
"learning_rate": 2.769419116476052e-07,
|
| 770 |
+
"logits/chosen": 0.1790754646062851,
|
| 771 |
+
"logits/rejected": 0.24030852317810059,
|
| 772 |
+
"logps/chosen": -473.218017578125,
|
| 773 |
+
"logps/rejected": -557.5300903320312,
|
| 774 |
+
"loss": 0.3077,
|
| 775 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 776 |
+
"rewards/chosen": -0.15667091310024261,
|
| 777 |
+
"rewards/margins": 0.09858374297618866,
|
| 778 |
+
"rewards/rejected": -0.2552546262741089,
|
| 779 |
+
"step": 540
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.88,
|
| 783 |
+
"learning_rate": 2.1651545897676512e-07,
|
| 784 |
+
"logits/chosen": 0.2089765965938568,
|
| 785 |
+
"logits/rejected": 0.25732511281967163,
|
| 786 |
+
"logps/chosen": -567.176025390625,
|
| 787 |
+
"logps/rejected": -613.021484375,
|
| 788 |
+
"loss": 0.3129,
|
| 789 |
+
"rewards/accuracies": 0.71875,
|
| 790 |
+
"rewards/chosen": -0.17653754353523254,
|
| 791 |
+
"rewards/margins": 0.09360690414905548,
|
| 792 |
+
"rewards/rejected": -0.2701444625854492,
|
| 793 |
+
"step": 550
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.9,
|
| 797 |
+
"learning_rate": 1.6322266119983222e-07,
|
| 798 |
+
"logits/chosen": 0.16576240956783295,
|
| 799 |
+
"logits/rejected": 0.24667489528656006,
|
| 800 |
+
"logps/chosen": -502.44158935546875,
|
| 801 |
+
"logps/rejected": -564.0174560546875,
|
| 802 |
+
"loss": 0.3299,
|
| 803 |
+
"rewards/accuracies": 0.668749988079071,
|
| 804 |
+
"rewards/chosen": -0.1819026917219162,
|
| 805 |
+
"rewards/margins": 0.09441694617271423,
|
| 806 |
+
"rewards/rejected": -0.2763196527957916,
|
| 807 |
+
"step": 560
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.91,
|
| 811 |
+
"learning_rate": 1.1723000616502167e-07,
|
| 812 |
+
"logits/chosen": 0.16597183048725128,
|
| 813 |
+
"logits/rejected": 0.13934046030044556,
|
| 814 |
+
"logps/chosen": -527.2802734375,
|
| 815 |
+
"logps/rejected": -672.5037841796875,
|
| 816 |
+
"loss": 0.2996,
|
| 817 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 818 |
+
"rewards/chosen": -0.18256615102291107,
|
| 819 |
+
"rewards/margins": 0.12837597727775574,
|
| 820 |
+
"rewards/rejected": -0.310942143201828,
|
| 821 |
+
"step": 570
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.93,
|
| 825 |
+
"learning_rate": 7.868117591737585e-08,
|
| 826 |
+
"logits/chosen": 0.20055902004241943,
|
| 827 |
+
"logits/rejected": 0.20812377333641052,
|
| 828 |
+
"logps/chosen": -483.41082763671875,
|
| 829 |
+
"logps/rejected": -559.0455932617188,
|
| 830 |
+
"loss": 0.3085,
|
| 831 |
+
"rewards/accuracies": 0.643750011920929,
|
| 832 |
+
"rewards/chosen": -0.16648918390274048,
|
| 833 |
+
"rewards/margins": 0.0913398414850235,
|
| 834 |
+
"rewards/rejected": -0.25782904028892517,
|
| 835 |
+
"step": 580
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.94,
|
| 839 |
+
"learning_rate": 4.769659783295383e-08,
|
| 840 |
+
"logits/chosen": 0.17583322525024414,
|
| 841 |
+
"logits/rejected": 0.1814696043729782,
|
| 842 |
+
"logps/chosen": -517.0927734375,
|
| 843 |
+
"logps/rejected": -562.7242431640625,
|
| 844 |
+
"loss": 0.3134,
|
| 845 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 846 |
+
"rewards/chosen": -0.1781904697418213,
|
| 847 |
+
"rewards/margins": 0.08398625254631042,
|
| 848 |
+
"rewards/rejected": -0.2621766924858093,
|
| 849 |
+
"step": 590
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.96,
|
| 853 |
+
"learning_rate": 2.4373068401120358e-08,
|
| 854 |
+
"logits/chosen": 0.16853031516075134,
|
| 855 |
+
"logits/rejected": 0.19935330748558044,
|
| 856 |
+
"logps/chosen": -528.5987548828125,
|
| 857 |
+
"logps/rejected": -601.7249755859375,
|
| 858 |
+
"loss": 0.323,
|
| 859 |
+
"rewards/accuracies": 0.675000011920929,
|
| 860 |
+
"rewards/chosen": -0.17777393758296967,
|
| 861 |
+
"rewards/margins": 0.08920314162969589,
|
| 862 |
+
"rewards/rejected": -0.26697710156440735,
|
| 863 |
+
"step": 600
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.98,
|
| 867 |
+
"learning_rate": 8.78345083022425e-09,
|
| 868 |
+
"logits/chosen": 0.16420051455497742,
|
| 869 |
+
"logits/rejected": 0.18029369413852692,
|
| 870 |
+
"logps/chosen": -543.4736938476562,
|
| 871 |
+
"logps/rejected": -574.1061401367188,
|
| 872 |
+
"loss": 0.3197,
|
| 873 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 874 |
+
"rewards/chosen": -0.1843399703502655,
|
| 875 |
+
"rewards/margins": 0.060389935970306396,
|
| 876 |
+
"rewards/rejected": -0.2447299212217331,
|
| 877 |
+
"step": 610
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 0.99,
|
| 881 |
+
"learning_rate": 9.764474213677654e-10,
|
| 882 |
+
"logits/chosen": 0.19845640659332275,
|
| 883 |
+
"logits/rejected": 0.22712858021259308,
|
| 884 |
+
"logps/chosen": -513.5285034179688,
|
| 885 |
+
"logps/rejected": -559.9097290039062,
|
| 886 |
+
"loss": 0.2998,
|
| 887 |
+
"rewards/accuracies": 0.637499988079071,
|
| 888 |
+
"rewards/chosen": -0.16102448105812073,
|
| 889 |
+
"rewards/margins": 0.07719925045967102,
|
| 890 |
+
"rewards/rejected": -0.23822371661663055,
|
| 891 |
+
"step": 620
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 1.0,
|
| 895 |
+
"step": 625,
|
| 896 |
+
"total_flos": 0.0,
|
| 897 |
+
"train_loss": 0.3311275300979614,
|
| 898 |
+
"train_runtime": 7525.4167,
|
| 899 |
+
"train_samples_per_second": 2.658,
|
| 900 |
+
"train_steps_per_second": 0.083
|
| 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 |
+
}
|