Model save
Browse files- README.md +77 -0
- adapter_model.safetensors +1 -1
- all_results.json +8 -0
- train_results.json +8 -0
- trainer_state.json +1723 -0
README.md
ADDED
|
@@ -0,0 +1,77 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: peft
|
| 3 |
+
tags:
|
| 4 |
+
- trl
|
| 5 |
+
- dpo
|
| 6 |
+
- generated_from_trainer
|
| 7 |
+
base_model: allenai/tulu-2-13b
|
| 8 |
+
model-index:
|
| 9 |
+
- name: tulu2-13b-cost-UF-5e-7
|
| 10 |
+
results: []
|
| 11 |
+
---
|
| 12 |
+
|
| 13 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 14 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 15 |
+
|
| 16 |
+
# tulu2-13b-cost-UF-5e-7
|
| 17 |
+
|
| 18 |
+
This model is a fine-tuned version of [allenai/tulu-2-13b](https://huggingface.co/allenai/tulu-2-13b) on the None dataset.
|
| 19 |
+
It achieves the following results on the evaluation set:
|
| 20 |
+
- Loss: 0.6933
|
| 21 |
+
- Rewards/chosen: 0.0268
|
| 22 |
+
- Rewards/rejected: 0.0265
|
| 23 |
+
- Rewards/accuracies: 0.5360
|
| 24 |
+
- Rewards/margins: 0.0002
|
| 25 |
+
- Rewards/margins Max: 0.0601
|
| 26 |
+
- Rewards/margins Min: -0.0619
|
| 27 |
+
- Rewards/margins Std: 0.0406
|
| 28 |
+
- Logps/rejected: -327.5517
|
| 29 |
+
- Logps/chosen: -331.2336
|
| 30 |
+
- Logits/rejected: -0.8962
|
| 31 |
+
- Logits/chosen: -1.0222
|
| 32 |
+
|
| 33 |
+
## Model description
|
| 34 |
+
|
| 35 |
+
More information needed
|
| 36 |
+
|
| 37 |
+
## Intended uses & limitations
|
| 38 |
+
|
| 39 |
+
More information needed
|
| 40 |
+
|
| 41 |
+
## Training and evaluation data
|
| 42 |
+
|
| 43 |
+
More information needed
|
| 44 |
+
|
| 45 |
+
## Training procedure
|
| 46 |
+
|
| 47 |
+
### Training hyperparameters
|
| 48 |
+
|
| 49 |
+
The following hyperparameters were used during training:
|
| 50 |
+
- learning_rate: 5e-07
|
| 51 |
+
- train_batch_size: 2
|
| 52 |
+
- eval_batch_size: 8
|
| 53 |
+
- seed: 42
|
| 54 |
+
- distributed_type: multi-GPU
|
| 55 |
+
- num_devices: 4
|
| 56 |
+
- gradient_accumulation_steps: 2
|
| 57 |
+
- total_train_batch_size: 16
|
| 58 |
+
- total_eval_batch_size: 32
|
| 59 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 60 |
+
- lr_scheduler_type: cosine
|
| 61 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 62 |
+
- num_epochs: 1
|
| 63 |
+
|
| 64 |
+
### Training results
|
| 65 |
+
|
| 66 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Rewards/margins Max | Rewards/margins Min | Rewards/margins Std | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 67 |
+
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:-------------------:|:-------------------:|:-------------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 68 |
+
| 0.6665 | 1.0 | 927 | 0.6933 | 0.0268 | 0.0265 | 0.5360 | 0.0002 | 0.0601 | -0.0619 | 0.0406 | -327.5517 | -331.2336 | -0.8962 | -1.0222 |
|
| 69 |
+
|
| 70 |
+
|
| 71 |
+
### Framework versions
|
| 72 |
+
|
| 73 |
+
- PEFT 0.7.1
|
| 74 |
+
- Transformers 4.39.0.dev0
|
| 75 |
+
- Pytorch 2.1.2+cu121
|
| 76 |
+
- Datasets 2.14.6
|
| 77 |
+
- 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 1001466944
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:76eecdc522c84a35e4e686020cb6afb523d2d22fce65fcef5530fc0b6afa4ccf
|
| 3 |
size 1001466944
|
all_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.6720605961327414,
|
| 4 |
+
"train_runtime": 8514.5547,
|
| 5 |
+
"train_samples": 14828,
|
| 6 |
+
"train_samples_per_second": 1.741,
|
| 7 |
+
"train_steps_per_second": 0.109
|
| 8 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"train_loss": 0.6720605961327414,
|
| 4 |
+
"train_runtime": 8514.5547,
|
| 5 |
+
"train_samples": 14828,
|
| 6 |
+
"train_samples_per_second": 1.741,
|
| 7 |
+
"train_steps_per_second": 0.109
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1723 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 927,
|
| 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 |
+
"grad_norm": 0.333984375,
|
| 14 |
+
"learning_rate": 5.3763440860215056e-09,
|
| 15 |
+
"logits/chosen": -1.7726776599884033,
|
| 16 |
+
"logits/rejected": -1.019553542137146,
|
| 17 |
+
"logps/chosen": -227.8472900390625,
|
| 18 |
+
"logps/rejected": -244.70220947265625,
|
| 19 |
+
"loss": 0.6931,
|
| 20 |
+
"rewards/accuracies": 0.0,
|
| 21 |
+
"rewards/chosen": 0.0,
|
| 22 |
+
"rewards/margins": 0.0,
|
| 23 |
+
"rewards/margins_max": 0.0,
|
| 24 |
+
"rewards/margins_min": 0.0,
|
| 25 |
+
"rewards/margins_std": 0.0,
|
| 26 |
+
"rewards/rejected": 0.0,
|
| 27 |
+
"step": 1
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.01,
|
| 31 |
+
"grad_norm": 0.33203125,
|
| 32 |
+
"learning_rate": 5.3763440860215054e-08,
|
| 33 |
+
"logits/chosen": -1.2758857011795044,
|
| 34 |
+
"logits/rejected": -0.7481470108032227,
|
| 35 |
+
"logps/chosen": -294.4023132324219,
|
| 36 |
+
"logps/rejected": -209.6774139404297,
|
| 37 |
+
"loss": 0.6931,
|
| 38 |
+
"rewards/accuracies": 0.4722222089767456,
|
| 39 |
+
"rewards/chosen": 0.00042969180503860116,
|
| 40 |
+
"rewards/margins": 0.00026647368213161826,
|
| 41 |
+
"rewards/margins_max": 0.002202093368396163,
|
| 42 |
+
"rewards/margins_min": -0.0016691461205482483,
|
| 43 |
+
"rewards/margins_std": 0.002737379400059581,
|
| 44 |
+
"rewards/rejected": 0.00016321813745889813,
|
| 45 |
+
"step": 10
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.02,
|
| 49 |
+
"grad_norm": 0.3515625,
|
| 50 |
+
"learning_rate": 1.0752688172043011e-07,
|
| 51 |
+
"logits/chosen": -1.4951783418655396,
|
| 52 |
+
"logits/rejected": -1.0168498754501343,
|
| 53 |
+
"logps/chosen": -280.9791259765625,
|
| 54 |
+
"logps/rejected": -274.51055908203125,
|
| 55 |
+
"loss": 0.6928,
|
| 56 |
+
"rewards/accuracies": 0.574999988079071,
|
| 57 |
+
"rewards/chosen": 0.0005079669645056129,
|
| 58 |
+
"rewards/margins": 0.0006722012185491621,
|
| 59 |
+
"rewards/margins_max": 0.003316085785627365,
|
| 60 |
+
"rewards/margins_min": -0.001971683232113719,
|
| 61 |
+
"rewards/margins_std": 0.003739017527550459,
|
| 62 |
+
"rewards/rejected": -0.00016423416673205793,
|
| 63 |
+
"step": 20
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.03,
|
| 67 |
+
"grad_norm": 0.2734375,
|
| 68 |
+
"learning_rate": 1.6129032258064515e-07,
|
| 69 |
+
"logits/chosen": -1.1558444499969482,
|
| 70 |
+
"logits/rejected": -0.7869107723236084,
|
| 71 |
+
"logps/chosen": -235.8018035888672,
|
| 72 |
+
"logps/rejected": -239.435791015625,
|
| 73 |
+
"loss": 0.6926,
|
| 74 |
+
"rewards/accuracies": 0.625,
|
| 75 |
+
"rewards/chosen": 0.0022618253715336323,
|
| 76 |
+
"rewards/margins": 0.001511804643087089,
|
| 77 |
+
"rewards/margins_max": 0.004373847972601652,
|
| 78 |
+
"rewards/margins_min": -0.001350238686427474,
|
| 79 |
+
"rewards/margins_std": 0.004047540482133627,
|
| 80 |
+
"rewards/rejected": 0.000750020903069526,
|
| 81 |
+
"step": 30
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.04,
|
| 85 |
+
"grad_norm": 0.2421875,
|
| 86 |
+
"learning_rate": 2.1505376344086022e-07,
|
| 87 |
+
"logits/chosen": -1.4029566049575806,
|
| 88 |
+
"logits/rejected": -0.8758159875869751,
|
| 89 |
+
"logps/chosen": -256.91668701171875,
|
| 90 |
+
"logps/rejected": -245.5706024169922,
|
| 91 |
+
"loss": 0.6927,
|
| 92 |
+
"rewards/accuracies": 0.5249999761581421,
|
| 93 |
+
"rewards/chosen": 0.0011851616436615586,
|
| 94 |
+
"rewards/margins": 0.0004023304209113121,
|
| 95 |
+
"rewards/margins_max": 0.002305095549672842,
|
| 96 |
+
"rewards/margins_min": -0.0015004349406808615,
|
| 97 |
+
"rewards/margins_std": 0.0026909164153039455,
|
| 98 |
+
"rewards/rejected": 0.0007828312227502465,
|
| 99 |
+
"step": 40
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.05,
|
| 103 |
+
"grad_norm": 0.25390625,
|
| 104 |
+
"learning_rate": 2.6881720430107523e-07,
|
| 105 |
+
"logits/chosen": -1.5180784463882446,
|
| 106 |
+
"logits/rejected": -1.0010168552398682,
|
| 107 |
+
"logps/chosen": -202.52719116210938,
|
| 108 |
+
"logps/rejected": -197.68511962890625,
|
| 109 |
+
"loss": 0.6926,
|
| 110 |
+
"rewards/accuracies": 0.574999988079071,
|
| 111 |
+
"rewards/chosen": 0.0013484725495800376,
|
| 112 |
+
"rewards/margins": 0.0014005316188558936,
|
| 113 |
+
"rewards/margins_max": 0.0035043410025537014,
|
| 114 |
+
"rewards/margins_min": -0.0007032775320112705,
|
| 115 |
+
"rewards/margins_std": 0.002975235693156719,
|
| 116 |
+
"rewards/rejected": -5.2059163863305e-05,
|
| 117 |
+
"step": 50
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.06,
|
| 121 |
+
"grad_norm": 0.287109375,
|
| 122 |
+
"learning_rate": 3.225806451612903e-07,
|
| 123 |
+
"logits/chosen": -1.5808765888214111,
|
| 124 |
+
"logits/rejected": -0.893613338470459,
|
| 125 |
+
"logps/chosen": -291.6551208496094,
|
| 126 |
+
"logps/rejected": -255.85360717773438,
|
| 127 |
+
"loss": 0.6922,
|
| 128 |
+
"rewards/accuracies": 0.574999988079071,
|
| 129 |
+
"rewards/chosen": 0.0011142367729917169,
|
| 130 |
+
"rewards/margins": 0.0015284843975678086,
|
| 131 |
+
"rewards/margins_max": 0.0042257350869476795,
|
| 132 |
+
"rewards/margins_min": -0.0011687660589814186,
|
| 133 |
+
"rewards/margins_std": 0.003814487950876355,
|
| 134 |
+
"rewards/rejected": -0.0004142475372646004,
|
| 135 |
+
"step": 60
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.08,
|
| 139 |
+
"grad_norm": 0.296875,
|
| 140 |
+
"learning_rate": 3.7634408602150537e-07,
|
| 141 |
+
"logits/chosen": -1.6458534002304077,
|
| 142 |
+
"logits/rejected": -0.9159662127494812,
|
| 143 |
+
"logps/chosen": -353.4042053222656,
|
| 144 |
+
"logps/rejected": -295.0526428222656,
|
| 145 |
+
"loss": 0.6916,
|
| 146 |
+
"rewards/accuracies": 0.75,
|
| 147 |
+
"rewards/chosen": 0.003586581675335765,
|
| 148 |
+
"rewards/margins": 0.0029636994004249573,
|
| 149 |
+
"rewards/margins_max": 0.0056008645333349705,
|
| 150 |
+
"rewards/margins_min": 0.00032653429661877453,
|
| 151 |
+
"rewards/margins_std": 0.0037295143119990826,
|
| 152 |
+
"rewards/rejected": 0.0006228827987797558,
|
| 153 |
+
"step": 70
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 0.09,
|
| 157 |
+
"grad_norm": 0.283203125,
|
| 158 |
+
"learning_rate": 4.3010752688172043e-07,
|
| 159 |
+
"logits/chosen": -1.3944181203842163,
|
| 160 |
+
"logits/rejected": -0.9768520593643188,
|
| 161 |
+
"logps/chosen": -247.8656463623047,
|
| 162 |
+
"logps/rejected": -222.736083984375,
|
| 163 |
+
"loss": 0.691,
|
| 164 |
+
"rewards/accuracies": 0.75,
|
| 165 |
+
"rewards/chosen": 0.004138199612498283,
|
| 166 |
+
"rewards/margins": 0.0038313076365739107,
|
| 167 |
+
"rewards/margins_max": 0.007481383625417948,
|
| 168 |
+
"rewards/margins_min": 0.00018123061454389244,
|
| 169 |
+
"rewards/margins_std": 0.005161988083273172,
|
| 170 |
+
"rewards/rejected": 0.00030689238337799907,
|
| 171 |
+
"step": 80
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.1,
|
| 175 |
+
"grad_norm": 0.263671875,
|
| 176 |
+
"learning_rate": 4.838709677419355e-07,
|
| 177 |
+
"logits/chosen": -1.4816954135894775,
|
| 178 |
+
"logits/rejected": -1.0011855363845825,
|
| 179 |
+
"logps/chosen": -307.66729736328125,
|
| 180 |
+
"logps/rejected": -218.35110473632812,
|
| 181 |
+
"loss": 0.6901,
|
| 182 |
+
"rewards/accuracies": 0.925000011920929,
|
| 183 |
+
"rewards/chosen": 0.006525175180286169,
|
| 184 |
+
"rewards/margins": 0.006856041494756937,
|
| 185 |
+
"rewards/margins_max": 0.012032730504870415,
|
| 186 |
+
"rewards/margins_min": 0.0016793517861515284,
|
| 187 |
+
"rewards/margins_std": 0.007320943288505077,
|
| 188 |
+
"rewards/rejected": -0.0003308658779133111,
|
| 189 |
+
"step": 90
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 0.11,
|
| 193 |
+
"grad_norm": 0.275390625,
|
| 194 |
+
"learning_rate": 4.999130942376231e-07,
|
| 195 |
+
"logits/chosen": -1.4475181102752686,
|
| 196 |
+
"logits/rejected": -0.8290830850601196,
|
| 197 |
+
"logps/chosen": -247.40322875976562,
|
| 198 |
+
"logps/rejected": -220.42355346679688,
|
| 199 |
+
"loss": 0.6897,
|
| 200 |
+
"rewards/accuracies": 0.875,
|
| 201 |
+
"rewards/chosen": 0.007414130959659815,
|
| 202 |
+
"rewards/margins": 0.0059156701900064945,
|
| 203 |
+
"rewards/margins_max": 0.00915153045207262,
|
| 204 |
+
"rewards/margins_min": 0.0026798094622790813,
|
| 205 |
+
"rewards/margins_std": 0.004576197825372219,
|
| 206 |
+
"rewards/rejected": 0.0014984606532379985,
|
| 207 |
+
"step": 100
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.12,
|
| 211 |
+
"grad_norm": 0.328125,
|
| 212 |
+
"learning_rate": 4.994875788073206e-07,
|
| 213 |
+
"logits/chosen": -1.3422911167144775,
|
| 214 |
+
"logits/rejected": -0.9277170300483704,
|
| 215 |
+
"logps/chosen": -265.04791259765625,
|
| 216 |
+
"logps/rejected": -291.83612060546875,
|
| 217 |
+
"loss": 0.6882,
|
| 218 |
+
"rewards/accuracies": 0.925000011920929,
|
| 219 |
+
"rewards/chosen": 0.009832861833274364,
|
| 220 |
+
"rewards/margins": 0.010079814121127129,
|
| 221 |
+
"rewards/margins_max": 0.014872267842292786,
|
| 222 |
+
"rewards/margins_min": 0.005287360865622759,
|
| 223 |
+
"rewards/margins_std": 0.006777553353458643,
|
| 224 |
+
"rewards/rejected": -0.00024695199681445956,
|
| 225 |
+
"step": 110
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.13,
|
| 229 |
+
"grad_norm": 0.3515625,
|
| 230 |
+
"learning_rate": 4.987080943856886e-07,
|
| 231 |
+
"logits/chosen": -1.4355990886688232,
|
| 232 |
+
"logits/rejected": -0.9039069414138794,
|
| 233 |
+
"logps/chosen": -241.0117950439453,
|
| 234 |
+
"logps/rejected": -261.8793640136719,
|
| 235 |
+
"loss": 0.6878,
|
| 236 |
+
"rewards/accuracies": 0.875,
|
| 237 |
+
"rewards/chosen": 0.010962730273604393,
|
| 238 |
+
"rewards/margins": 0.011187642812728882,
|
| 239 |
+
"rewards/margins_max": 0.01832672953605652,
|
| 240 |
+
"rewards/margins_min": 0.0040485551580786705,
|
| 241 |
+
"rewards/margins_std": 0.010096193291246891,
|
| 242 |
+
"rewards/rejected": -0.0002249126264359802,
|
| 243 |
+
"step": 120
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.14,
|
| 247 |
+
"grad_norm": 0.318359375,
|
| 248 |
+
"learning_rate": 4.975757468927726e-07,
|
| 249 |
+
"logits/chosen": -1.5994830131530762,
|
| 250 |
+
"logits/rejected": -0.8840494155883789,
|
| 251 |
+
"logps/chosen": -262.6464538574219,
|
| 252 |
+
"logps/rejected": -225.1462860107422,
|
| 253 |
+
"loss": 0.6878,
|
| 254 |
+
"rewards/accuracies": 0.875,
|
| 255 |
+
"rewards/chosen": 0.01504091639071703,
|
| 256 |
+
"rewards/margins": 0.014881642535328865,
|
| 257 |
+
"rewards/margins_max": 0.021485231816768646,
|
| 258 |
+
"rewards/margins_min": 0.008278051391243935,
|
| 259 |
+
"rewards/margins_std": 0.00933888740837574,
|
| 260 |
+
"rewards/rejected": 0.00015927411732263863,
|
| 261 |
+
"step": 130
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.15,
|
| 265 |
+
"grad_norm": 0.28125,
|
| 266 |
+
"learning_rate": 4.960921428851066e-07,
|
| 267 |
+
"logits/chosen": -1.3022377490997314,
|
| 268 |
+
"logits/rejected": -0.9370013475418091,
|
| 269 |
+
"logps/chosen": -238.2805938720703,
|
| 270 |
+
"logps/rejected": -267.19854736328125,
|
| 271 |
+
"loss": 0.6864,
|
| 272 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 273 |
+
"rewards/chosen": 0.015973640605807304,
|
| 274 |
+
"rewards/margins": 0.014827728271484375,
|
| 275 |
+
"rewards/margins_max": 0.02404005080461502,
|
| 276 |
+
"rewards/margins_min": 0.005615406669676304,
|
| 277 |
+
"rewards/margins_std": 0.013028192333877087,
|
| 278 |
+
"rewards/rejected": 0.001145911985076964,
|
| 279 |
+
"step": 140
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.16,
|
| 283 |
+
"grad_norm": 0.37890625,
|
| 284 |
+
"learning_rate": 4.942593872763566e-07,
|
| 285 |
+
"logits/chosen": -1.4023020267486572,
|
| 286 |
+
"logits/rejected": -0.8060510754585266,
|
| 287 |
+
"logps/chosen": -228.21420288085938,
|
| 288 |
+
"logps/rejected": -223.0884246826172,
|
| 289 |
+
"loss": 0.6856,
|
| 290 |
+
"rewards/accuracies": 0.875,
|
| 291 |
+
"rewards/chosen": 0.01750522293150425,
|
| 292 |
+
"rewards/margins": 0.016751740127801895,
|
| 293 |
+
"rewards/margins_max": 0.026506105437874794,
|
| 294 |
+
"rewards/margins_min": 0.006997367832809687,
|
| 295 |
+
"rewards/margins_std": 0.013794762082397938,
|
| 296 |
+
"rewards/rejected": 0.0007534866454079747,
|
| 297 |
+
"step": 150
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.17,
|
| 301 |
+
"grad_norm": 0.345703125,
|
| 302 |
+
"learning_rate": 4.920800803509025e-07,
|
| 303 |
+
"logits/chosen": -1.6911264657974243,
|
| 304 |
+
"logits/rejected": -0.8840080499649048,
|
| 305 |
+
"logps/chosen": -303.69427490234375,
|
| 306 |
+
"logps/rejected": -275.5614318847656,
|
| 307 |
+
"loss": 0.6842,
|
| 308 |
+
"rewards/accuracies": 0.949999988079071,
|
| 309 |
+
"rewards/chosen": 0.022266970947384834,
|
| 310 |
+
"rewards/margins": 0.02082091197371483,
|
| 311 |
+
"rewards/margins_max": 0.02934589982032776,
|
| 312 |
+
"rewards/margins_min": 0.012295925989747047,
|
| 313 |
+
"rewards/margins_std": 0.012056154198944569,
|
| 314 |
+
"rewards/rejected": 0.0014460586244240403,
|
| 315 |
+
"step": 160
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.18,
|
| 319 |
+
"grad_norm": 0.3046875,
|
| 320 |
+
"learning_rate": 4.895573140745967e-07,
|
| 321 |
+
"logits/chosen": -1.5039719343185425,
|
| 322 |
+
"logits/rejected": -1.0316977500915527,
|
| 323 |
+
"logps/chosen": -345.7312316894531,
|
| 324 |
+
"logps/rejected": -288.9708251953125,
|
| 325 |
+
"loss": 0.6836,
|
| 326 |
+
"rewards/accuracies": 0.949999988079071,
|
| 327 |
+
"rewards/chosen": 0.02365526184439659,
|
| 328 |
+
"rewards/margins": 0.02123275026679039,
|
| 329 |
+
"rewards/margins_max": 0.03283926099538803,
|
| 330 |
+
"rewards/margins_min": 0.009626244194805622,
|
| 331 |
+
"rewards/margins_std": 0.016414081677794456,
|
| 332 |
+
"rewards/rejected": 0.002422512974590063,
|
| 333 |
+
"step": 170
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.19,
|
| 337 |
+
"grad_norm": 0.265625,
|
| 338 |
+
"learning_rate": 4.866946677079314e-07,
|
| 339 |
+
"logits/chosen": -1.5601823329925537,
|
| 340 |
+
"logits/rejected": -1.0506742000579834,
|
| 341 |
+
"logps/chosen": -231.99703979492188,
|
| 342 |
+
"logps/rejected": -237.2670135498047,
|
| 343 |
+
"loss": 0.6828,
|
| 344 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 345 |
+
"rewards/chosen": 0.022809389978647232,
|
| 346 |
+
"rewards/margins": 0.020736858248710632,
|
| 347 |
+
"rewards/margins_max": 0.030726289376616478,
|
| 348 |
+
"rewards/margins_min": 0.010747427120804787,
|
| 349 |
+
"rewards/margins_std": 0.014127190224826336,
|
| 350 |
+
"rewards/rejected": 0.002072530798614025,
|
| 351 |
+
"step": 180
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.2,
|
| 355 |
+
"grad_norm": 0.365234375,
|
| 356 |
+
"learning_rate": 4.834962027278417e-07,
|
| 357 |
+
"logits/chosen": -1.4942362308502197,
|
| 358 |
+
"logits/rejected": -0.8470790982246399,
|
| 359 |
+
"logps/chosen": -291.2981872558594,
|
| 360 |
+
"logps/rejected": -240.0912628173828,
|
| 361 |
+
"loss": 0.682,
|
| 362 |
+
"rewards/accuracies": 0.925000011920929,
|
| 363 |
+
"rewards/chosen": 0.017627181485295296,
|
| 364 |
+
"rewards/margins": 0.022342149168252945,
|
| 365 |
+
"rewards/margins_max": 0.033914387226104736,
|
| 366 |
+
"rewards/margins_min": 0.01076990831643343,
|
| 367 |
+
"rewards/margins_std": 0.016365615651011467,
|
| 368 |
+
"rewards/rejected": -0.004714967682957649,
|
| 369 |
+
"step": 190
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.22,
|
| 373 |
+
"grad_norm": 0.345703125,
|
| 374 |
+
"learning_rate": 4.799664570653473e-07,
|
| 375 |
+
"logits/chosen": -1.4884960651397705,
|
| 376 |
+
"logits/rejected": -0.7421751022338867,
|
| 377 |
+
"logps/chosen": -295.2840270996094,
|
| 378 |
+
"logps/rejected": -230.6145477294922,
|
| 379 |
+
"loss": 0.6808,
|
| 380 |
+
"rewards/accuracies": 1.0,
|
| 381 |
+
"rewards/chosen": 0.02664310857653618,
|
| 382 |
+
"rewards/margins": 0.031296949833631516,
|
| 383 |
+
"rewards/margins_max": 0.04525149241089821,
|
| 384 |
+
"rewards/margins_min": 0.01734241284430027,
|
| 385 |
+
"rewards/margins_std": 0.0197346992790699,
|
| 386 |
+
"rewards/rejected": -0.00465384079143405,
|
| 387 |
+
"step": 200
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.23,
|
| 391 |
+
"grad_norm": 0.2578125,
|
| 392 |
+
"learning_rate": 4.7611043866720737e-07,
|
| 393 |
+
"logits/chosen": -1.49364173412323,
|
| 394 |
+
"logits/rejected": -1.031049132347107,
|
| 395 |
+
"logps/chosen": -258.94512939453125,
|
| 396 |
+
"logps/rejected": -287.6114196777344,
|
| 397 |
+
"loss": 0.6808,
|
| 398 |
+
"rewards/accuracies": 0.925000011920929,
|
| 399 |
+
"rewards/chosen": 0.021609986200928688,
|
| 400 |
+
"rewards/margins": 0.020240817219018936,
|
| 401 |
+
"rewards/margins_max": 0.033506136387586594,
|
| 402 |
+
"rewards/margins_min": 0.006975496653467417,
|
| 403 |
+
"rewards/margins_std": 0.018759997561573982,
|
| 404 |
+
"rewards/rejected": 0.0013691672356799245,
|
| 405 |
+
"step": 210
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.24,
|
| 409 |
+
"grad_norm": 0.291015625,
|
| 410 |
+
"learning_rate": 4.719336183907265e-07,
|
| 411 |
+
"logits/chosen": -1.3044389486312866,
|
| 412 |
+
"logits/rejected": -0.9332467317581177,
|
| 413 |
+
"logps/chosen": -223.7823944091797,
|
| 414 |
+
"logps/rejected": -217.8594207763672,
|
| 415 |
+
"loss": 0.6799,
|
| 416 |
+
"rewards/accuracies": 0.925000011920929,
|
| 417 |
+
"rewards/chosen": 0.02070688083767891,
|
| 418 |
+
"rewards/margins": 0.020120607689023018,
|
| 419 |
+
"rewards/margins_max": 0.03160488232970238,
|
| 420 |
+
"rewards/margins_min": 0.008636328391730785,
|
| 421 |
+
"rewards/margins_std": 0.016241220757365227,
|
| 422 |
+
"rewards/rejected": 0.0005862751277163625,
|
| 423 |
+
"step": 220
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 0.25,
|
| 427 |
+
"grad_norm": 0.3125,
|
| 428 |
+
"learning_rate": 4.6744192224178984e-07,
|
| 429 |
+
"logits/chosen": -1.3842111825942993,
|
| 430 |
+
"logits/rejected": -0.9888660311698914,
|
| 431 |
+
"logps/chosen": -236.2473907470703,
|
| 432 |
+
"logps/rejected": -271.76641845703125,
|
| 433 |
+
"loss": 0.6802,
|
| 434 |
+
"rewards/accuracies": 0.925000011920929,
|
| 435 |
+
"rewards/chosen": 0.022707760334014893,
|
| 436 |
+
"rewards/margins": 0.02468196675181389,
|
| 437 |
+
"rewards/margins_max": 0.03745966777205467,
|
| 438 |
+
"rewards/margins_min": 0.011904269456863403,
|
| 439 |
+
"rewards/margins_std": 0.018070396035909653,
|
| 440 |
+
"rewards/rejected": -0.0019742068834602833,
|
| 441 |
+
"step": 230
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 0.26,
|
| 445 |
+
"grad_norm": 0.30078125,
|
| 446 |
+
"learning_rate": 4.6264172296714e-07,
|
| 447 |
+
"logits/chosen": -1.468925952911377,
|
| 448 |
+
"logits/rejected": -0.8928337097167969,
|
| 449 |
+
"logps/chosen": -218.7871551513672,
|
| 450 |
+
"logps/rejected": -232.1916961669922,
|
| 451 |
+
"loss": 0.6782,
|
| 452 |
+
"rewards/accuracies": 0.949999988079071,
|
| 453 |
+
"rewards/chosen": 0.029307205229997635,
|
| 454 |
+
"rewards/margins": 0.03086397610604763,
|
| 455 |
+
"rewards/margins_max": 0.0454208180308342,
|
| 456 |
+
"rewards/margins_min": 0.016307134181261063,
|
| 457 |
+
"rewards/margins_std": 0.02058648318052292,
|
| 458 |
+
"rewards/rejected": -0.0015567743685096502,
|
| 459 |
+
"step": 240
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.27,
|
| 463 |
+
"grad_norm": 0.263671875,
|
| 464 |
+
"learning_rate": 4.575398310128262e-07,
|
| 465 |
+
"logits/chosen": -1.52159583568573,
|
| 466 |
+
"logits/rejected": -1.062409520149231,
|
| 467 |
+
"logps/chosen": -205.5279083251953,
|
| 468 |
+
"logps/rejected": -209.3750457763672,
|
| 469 |
+
"loss": 0.6784,
|
| 470 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 471 |
+
"rewards/chosen": 0.0256162341684103,
|
| 472 |
+
"rewards/margins": 0.029211264103651047,
|
| 473 |
+
"rewards/margins_max": 0.04594603180885315,
|
| 474 |
+
"rewards/margins_min": 0.012476496398448944,
|
| 475 |
+
"rewards/margins_std": 0.023666534572839737,
|
| 476 |
+
"rewards/rejected": -0.0035950313322246075,
|
| 477 |
+
"step": 250
|
| 478 |
+
},
|
| 479 |
+
{
|
| 480 |
+
"epoch": 0.28,
|
| 481 |
+
"grad_norm": 0.28515625,
|
| 482 |
+
"learning_rate": 4.5214348486165227e-07,
|
| 483 |
+
"logits/chosen": -1.4600447416305542,
|
| 484 |
+
"logits/rejected": -1.0638010501861572,
|
| 485 |
+
"logps/chosen": -263.8650817871094,
|
| 486 |
+
"logps/rejected": -257.2422790527344,
|
| 487 |
+
"loss": 0.6772,
|
| 488 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 489 |
+
"rewards/chosen": 0.031198721379041672,
|
| 490 |
+
"rewards/margins": 0.032171688973903656,
|
| 491 |
+
"rewards/margins_max": 0.04676546901464462,
|
| 492 |
+
"rewards/margins_min": 0.01757790893316269,
|
| 493 |
+
"rewards/margins_std": 0.02063872292637825,
|
| 494 |
+
"rewards/rejected": -0.0009729691664688289,
|
| 495 |
+
"step": 260
|
| 496 |
+
},
|
| 497 |
+
{
|
| 498 |
+
"epoch": 0.29,
|
| 499 |
+
"grad_norm": 0.333984375,
|
| 500 |
+
"learning_rate": 4.4646034076333254e-07,
|
| 501 |
+
"logits/chosen": -1.4118484258651733,
|
| 502 |
+
"logits/rejected": -1.0390139818191528,
|
| 503 |
+
"logps/chosen": -254.69589233398438,
|
| 504 |
+
"logps/rejected": -277.36029052734375,
|
| 505 |
+
"loss": 0.6747,
|
| 506 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 507 |
+
"rewards/chosen": 0.03478916734457016,
|
| 508 |
+
"rewards/margins": 0.03131258860230446,
|
| 509 |
+
"rewards/margins_max": 0.0460633859038353,
|
| 510 |
+
"rewards/margins_min": 0.016561787575483322,
|
| 511 |
+
"rewards/margins_std": 0.020860780030488968,
|
| 512 |
+
"rewards/rejected": 0.0034765794407576323,
|
| 513 |
+
"step": 270
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 0.3,
|
| 517 |
+
"grad_norm": 0.337890625,
|
| 518 |
+
"learning_rate": 4.404984618719274e-07,
|
| 519 |
+
"logits/chosen": -1.5356388092041016,
|
| 520 |
+
"logits/rejected": -0.8826289176940918,
|
| 521 |
+
"logps/chosen": -213.24365234375,
|
| 522 |
+
"logps/rejected": -205.28201293945312,
|
| 523 |
+
"loss": 0.674,
|
| 524 |
+
"rewards/accuracies": 0.949999988079071,
|
| 525 |
+
"rewards/chosen": 0.03563285619020462,
|
| 526 |
+
"rewards/margins": 0.044356830418109894,
|
| 527 |
+
"rewards/margins_max": 0.05965030938386917,
|
| 528 |
+
"rewards/margins_min": 0.029063349589705467,
|
| 529 |
+
"rewards/margins_std": 0.0216282457113266,
|
| 530 |
+
"rewards/rejected": -0.008723974227905273,
|
| 531 |
+
"step": 280
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 0.31,
|
| 535 |
+
"grad_norm": 0.322265625,
|
| 536 |
+
"learning_rate": 4.342663068059689e-07,
|
| 537 |
+
"logits/chosen": -1.5109997987747192,
|
| 538 |
+
"logits/rejected": -1.0301908254623413,
|
| 539 |
+
"logps/chosen": -227.60043334960938,
|
| 540 |
+
"logps/rejected": -225.4455108642578,
|
| 541 |
+
"loss": 0.6737,
|
| 542 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 543 |
+
"rewards/chosen": 0.033510081470012665,
|
| 544 |
+
"rewards/margins": 0.03692306950688362,
|
| 545 |
+
"rewards/margins_max": 0.05596238374710083,
|
| 546 |
+
"rewards/margins_min": 0.017883744090795517,
|
| 547 |
+
"rewards/margins_std": 0.026925668120384216,
|
| 548 |
+
"rewards/rejected": -0.0034129873383790255,
|
| 549 |
+
"step": 290
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.32,
|
| 553 |
+
"grad_norm": 0.361328125,
|
| 554 |
+
"learning_rate": 4.27772717647508e-07,
|
| 555 |
+
"logits/chosen": -1.4661680459976196,
|
| 556 |
+
"logits/rejected": -1.0104472637176514,
|
| 557 |
+
"logps/chosen": -242.0974578857422,
|
| 558 |
+
"logps/rejected": -232.9008026123047,
|
| 559 |
+
"loss": 0.6744,
|
| 560 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 561 |
+
"rewards/chosen": 0.030629118904471397,
|
| 562 |
+
"rewards/margins": 0.03372279554605484,
|
| 563 |
+
"rewards/margins_max": 0.050439924001693726,
|
| 564 |
+
"rewards/margins_min": 0.017005670815706253,
|
| 565 |
+
"rewards/margins_std": 0.023641586303710938,
|
| 566 |
+
"rewards/rejected": -0.00309367710724473,
|
| 567 |
+
"step": 300
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.33,
|
| 571 |
+
"grad_norm": 0.294921875,
|
| 572 |
+
"learning_rate": 4.2102690739710975e-07,
|
| 573 |
+
"logits/chosen": -1.35811448097229,
|
| 574 |
+
"logits/rejected": -0.8667371869087219,
|
| 575 |
+
"logps/chosen": -207.32943725585938,
|
| 576 |
+
"logps/rejected": -239.2294464111328,
|
| 577 |
+
"loss": 0.67,
|
| 578 |
+
"rewards/accuracies": 0.925000011920929,
|
| 579 |
+
"rewards/chosen": 0.04066639393568039,
|
| 580 |
+
"rewards/margins": 0.040916211903095245,
|
| 581 |
+
"rewards/margins_max": 0.05701497197151184,
|
| 582 |
+
"rewards/margins_min": 0.02481745555996895,
|
| 583 |
+
"rewards/margins_std": 0.0227670781314373,
|
| 584 |
+
"rewards/rejected": -0.0002498172107152641,
|
| 585 |
+
"step": 310
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.35,
|
| 589 |
+
"grad_norm": 0.302734375,
|
| 590 |
+
"learning_rate": 4.140384469025954e-07,
|
| 591 |
+
"logits/chosen": -1.4642293453216553,
|
| 592 |
+
"logits/rejected": -0.8783510327339172,
|
| 593 |
+
"logps/chosen": -269.0649719238281,
|
| 594 |
+
"logps/rejected": -253.2615966796875,
|
| 595 |
+
"loss": 0.6722,
|
| 596 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 597 |
+
"rewards/chosen": 0.042783185839653015,
|
| 598 |
+
"rewards/margins": 0.04893990606069565,
|
| 599 |
+
"rewards/margins_max": 0.07330337911844254,
|
| 600 |
+
"rewards/margins_min": 0.024576421827077866,
|
| 601 |
+
"rewards/margins_std": 0.03445516526699066,
|
| 602 |
+
"rewards/rejected": -0.0061567178927361965,
|
| 603 |
+
"step": 320
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.36,
|
| 607 |
+
"grad_norm": 0.326171875,
|
| 608 |
+
"learning_rate": 4.068172512800759e-07,
|
| 609 |
+
"logits/chosen": -1.4688003063201904,
|
| 610 |
+
"logits/rejected": -0.9481694102287292,
|
| 611 |
+
"logps/chosen": -263.0448303222656,
|
| 612 |
+
"logps/rejected": -261.3179931640625,
|
| 613 |
+
"loss": 0.6708,
|
| 614 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 615 |
+
"rewards/chosen": 0.03902563825249672,
|
| 616 |
+
"rewards/margins": 0.04815050959587097,
|
| 617 |
+
"rewards/margins_max": 0.06747350096702576,
|
| 618 |
+
"rewards/margins_min": 0.02882750704884529,
|
| 619 |
+
"rewards/margins_std": 0.02732684649527073,
|
| 620 |
+
"rewards/rejected": -0.009124869480729103,
|
| 621 |
+
"step": 330
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 0.37,
|
| 625 |
+
"grad_norm": 0.271484375,
|
| 626 |
+
"learning_rate": 3.993735658465446e-07,
|
| 627 |
+
"logits/chosen": -1.3654890060424805,
|
| 628 |
+
"logits/rejected": -1.0353434085845947,
|
| 629 |
+
"logps/chosen": -227.32803344726562,
|
| 630 |
+
"logps/rejected": -255.8052520751953,
|
| 631 |
+
"loss": 0.6708,
|
| 632 |
+
"rewards/accuracies": 0.949999988079071,
|
| 633 |
+
"rewards/chosen": 0.035697564482688904,
|
| 634 |
+
"rewards/margins": 0.03952573984861374,
|
| 635 |
+
"rewards/margins_max": 0.06120690703392029,
|
| 636 |
+
"rewards/margins_min": 0.01784456893801689,
|
| 637 |
+
"rewards/margins_std": 0.030661800876259804,
|
| 638 |
+
"rewards/rejected": -0.0038281746674329042,
|
| 639 |
+
"step": 340
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.38,
|
| 643 |
+
"grad_norm": 0.2734375,
|
| 644 |
+
"learning_rate": 3.917179515839839e-07,
|
| 645 |
+
"logits/chosen": -1.522472620010376,
|
| 646 |
+
"logits/rejected": -0.8194535970687866,
|
| 647 |
+
"logps/chosen": -272.1158752441406,
|
| 648 |
+
"logps/rejected": -224.29052734375,
|
| 649 |
+
"loss": 0.6722,
|
| 650 |
+
"rewards/accuracies": 0.949999988079071,
|
| 651 |
+
"rewards/chosen": 0.04334510862827301,
|
| 652 |
+
"rewards/margins": 0.045738715678453445,
|
| 653 |
+
"rewards/margins_max": 0.07282572239637375,
|
| 654 |
+
"rewards/margins_min": 0.01865171454846859,
|
| 655 |
+
"rewards/margins_std": 0.03830680996179581,
|
| 656 |
+
"rewards/rejected": -0.002393609844148159,
|
| 657 |
+
"step": 350
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.39,
|
| 661 |
+
"grad_norm": 0.328125,
|
| 662 |
+
"learning_rate": 3.8386127015561377e-07,
|
| 663 |
+
"logits/chosen": -1.4874508380889893,
|
| 664 |
+
"logits/rejected": -0.9508639574050903,
|
| 665 |
+
"logps/chosen": -264.84063720703125,
|
| 666 |
+
"logps/rejected": -285.2606506347656,
|
| 667 |
+
"loss": 0.6679,
|
| 668 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 669 |
+
"rewards/chosen": 0.038348861038684845,
|
| 670 |
+
"rewards/margins": 0.05200430005788803,
|
| 671 |
+
"rewards/margins_max": 0.07226666063070297,
|
| 672 |
+
"rewards/margins_min": 0.03174193948507309,
|
| 673 |
+
"rewards/margins_std": 0.028655309230089188,
|
| 674 |
+
"rewards/rejected": -0.013655440881848335,
|
| 675 |
+
"step": 360
|
| 676 |
+
},
|
| 677 |
+
{
|
| 678 |
+
"epoch": 0.4,
|
| 679 |
+
"grad_norm": 0.310546875,
|
| 680 |
+
"learning_rate": 3.758146684955368e-07,
|
| 681 |
+
"logits/chosen": -1.4181379079818726,
|
| 682 |
+
"logits/rejected": -0.9576012492179871,
|
| 683 |
+
"logps/chosen": -246.5336151123047,
|
| 684 |
+
"logps/rejected": -277.94232177734375,
|
| 685 |
+
"loss": 0.6709,
|
| 686 |
+
"rewards/accuracies": 0.925000011920929,
|
| 687 |
+
"rewards/chosen": 0.03755969554185867,
|
| 688 |
+
"rewards/margins": 0.04255872964859009,
|
| 689 |
+
"rewards/margins_max": 0.06353293359279633,
|
| 690 |
+
"rewards/margins_min": 0.021584514528512955,
|
| 691 |
+
"rewards/margins_std": 0.029662013053894043,
|
| 692 |
+
"rewards/rejected": -0.00499903317540884,
|
| 693 |
+
"step": 370
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 0.41,
|
| 697 |
+
"grad_norm": 0.349609375,
|
| 698 |
+
"learning_rate": 3.6758956299364643e-07,
|
| 699 |
+
"logits/chosen": -1.6038116216659546,
|
| 700 |
+
"logits/rejected": -1.0424953699111938,
|
| 701 |
+
"logps/chosen": -232.23916625976562,
|
| 702 |
+
"logps/rejected": -258.908203125,
|
| 703 |
+
"loss": 0.6694,
|
| 704 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 705 |
+
"rewards/chosen": 0.044283345341682434,
|
| 706 |
+
"rewards/margins": 0.04850899800658226,
|
| 707 |
+
"rewards/margins_max": 0.07648013532161713,
|
| 708 |
+
"rewards/margins_min": 0.020537864416837692,
|
| 709 |
+
"rewards/margins_std": 0.039557162672281265,
|
| 710 |
+
"rewards/rejected": -0.004225648939609528,
|
| 711 |
+
"step": 380
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 0.42,
|
| 715 |
+
"grad_norm": 0.318359375,
|
| 716 |
+
"learning_rate": 3.591976232982355e-07,
|
| 717 |
+
"logits/chosen": -1.4831396341323853,
|
| 718 |
+
"logits/rejected": -0.7363861203193665,
|
| 719 |
+
"logps/chosen": -274.75311279296875,
|
| 720 |
+
"logps/rejected": -217.2397918701172,
|
| 721 |
+
"loss": 0.6673,
|
| 722 |
+
"rewards/accuracies": 0.949999988079071,
|
| 723 |
+
"rewards/chosen": 0.052159328013658524,
|
| 724 |
+
"rewards/margins": 0.05904749035835266,
|
| 725 |
+
"rewards/margins_max": 0.08474520593881607,
|
| 726 |
+
"rewards/margins_min": 0.03334975615143776,
|
| 727 |
+
"rewards/margins_std": 0.0363420732319355,
|
| 728 |
+
"rewards/rejected": -0.006888158619403839,
|
| 729 |
+
"step": 390
|
| 730 |
+
},
|
| 731 |
+
{
|
| 732 |
+
"epoch": 0.43,
|
| 733 |
+
"grad_norm": 0.3046875,
|
| 734 |
+
"learning_rate": 3.506507557592853e-07,
|
| 735 |
+
"logits/chosen": -1.4179537296295166,
|
| 736 |
+
"logits/rejected": -0.8838945627212524,
|
| 737 |
+
"logps/chosen": -329.03863525390625,
|
| 738 |
+
"logps/rejected": -292.02008056640625,
|
| 739 |
+
"loss": 0.6655,
|
| 740 |
+
"rewards/accuracies": 0.949999988079071,
|
| 741 |
+
"rewards/chosen": 0.044211916625499725,
|
| 742 |
+
"rewards/margins": 0.057223010808229446,
|
| 743 |
+
"rewards/margins_max": 0.08557866513729095,
|
| 744 |
+
"rewards/margins_min": 0.028867345303297043,
|
| 745 |
+
"rewards/margins_std": 0.04010096564888954,
|
| 746 |
+
"rewards/rejected": -0.013011088594794273,
|
| 747 |
+
"step": 400
|
| 748 |
+
},
|
| 749 |
+
{
|
| 750 |
+
"epoch": 0.44,
|
| 751 |
+
"grad_norm": 0.4375,
|
| 752 |
+
"learning_rate": 3.419610865359266e-07,
|
| 753 |
+
"logits/chosen": -1.4828202724456787,
|
| 754 |
+
"logits/rejected": -0.8764745593070984,
|
| 755 |
+
"logps/chosen": -280.54998779296875,
|
| 756 |
+
"logps/rejected": -276.4558410644531,
|
| 757 |
+
"loss": 0.6682,
|
| 758 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 759 |
+
"rewards/chosen": 0.05212799459695816,
|
| 760 |
+
"rewards/margins": 0.05141522362828255,
|
| 761 |
+
"rewards/margins_max": 0.08034422993659973,
|
| 762 |
+
"rewards/margins_min": 0.022486215457320213,
|
| 763 |
+
"rewards/margins_std": 0.04091179370880127,
|
| 764 |
+
"rewards/rejected": 0.0007127688149921596,
|
| 765 |
+
"step": 410
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.45,
|
| 769 |
+
"grad_norm": 0.33984375,
|
| 770 |
+
"learning_rate": 3.33140944392039e-07,
|
| 771 |
+
"logits/chosen": -1.2407147884368896,
|
| 772 |
+
"logits/rejected": -0.8896020650863647,
|
| 773 |
+
"logps/chosen": -235.588134765625,
|
| 774 |
+
"logps/rejected": -242.38858032226562,
|
| 775 |
+
"loss": 0.6687,
|
| 776 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 777 |
+
"rewards/chosen": 0.03107466734945774,
|
| 778 |
+
"rewards/margins": 0.04102586954832077,
|
| 779 |
+
"rewards/margins_max": 0.0646006390452385,
|
| 780 |
+
"rewards/margins_min": 0.017451094463467598,
|
| 781 |
+
"rewards/margins_std": 0.033339761197566986,
|
| 782 |
+
"rewards/rejected": -0.00995120219886303,
|
| 783 |
+
"step": 420
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 0.46,
|
| 787 |
+
"grad_norm": 0.3203125,
|
| 788 |
+
"learning_rate": 3.2420284320439736e-07,
|
| 789 |
+
"logits/chosen": -1.5982177257537842,
|
| 790 |
+
"logits/rejected": -0.8901177644729614,
|
| 791 |
+
"logps/chosen": -235.1398468017578,
|
| 792 |
+
"logps/rejected": -225.75119018554688,
|
| 793 |
+
"loss": 0.667,
|
| 794 |
+
"rewards/accuracies": 0.949999988079071,
|
| 795 |
+
"rewards/chosen": 0.05144185945391655,
|
| 796 |
+
"rewards/margins": 0.0693436712026596,
|
| 797 |
+
"rewards/margins_max": 0.09611108899116516,
|
| 798 |
+
"rewards/margins_min": 0.04257623478770256,
|
| 799 |
+
"rewards/margins_std": 0.03785485774278641,
|
| 800 |
+
"rewards/rejected": -0.01790180616080761,
|
| 801 |
+
"step": 430
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 0.47,
|
| 805 |
+
"grad_norm": 0.291015625,
|
| 806 |
+
"learning_rate": 3.151594642081834e-07,
|
| 807 |
+
"logits/chosen": -1.5106613636016846,
|
| 808 |
+
"logits/rejected": -0.9530634880065918,
|
| 809 |
+
"logps/chosen": -259.29364013671875,
|
| 810 |
+
"logps/rejected": -263.5410461425781,
|
| 811 |
+
"loss": 0.6681,
|
| 812 |
+
"rewards/accuracies": 0.925000011920929,
|
| 813 |
+
"rewards/chosen": 0.054123084992170334,
|
| 814 |
+
"rewards/margins": 0.0667182207107544,
|
| 815 |
+
"rewards/margins_max": 0.09307887405157089,
|
| 816 |
+
"rewards/margins_min": 0.0403575673699379,
|
| 817 |
+
"rewards/margins_std": 0.03727959841489792,
|
| 818 |
+
"rewards/rejected": -0.012595141306519508,
|
| 819 |
+
"step": 440
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 0.49,
|
| 823 |
+
"grad_norm": 0.341796875,
|
| 824 |
+
"learning_rate": 3.060236380050519e-07,
|
| 825 |
+
"logits/chosen": -1.5215215682983398,
|
| 826 |
+
"logits/rejected": -0.915096640586853,
|
| 827 |
+
"logps/chosen": -241.9713897705078,
|
| 828 |
+
"logps/rejected": -212.6835174560547,
|
| 829 |
+
"loss": 0.6648,
|
| 830 |
+
"rewards/accuracies": 0.925000011920929,
|
| 831 |
+
"rewards/chosen": 0.041197773069143295,
|
| 832 |
+
"rewards/margins": 0.06050584465265274,
|
| 833 |
+
"rewards/margins_max": 0.09923966228961945,
|
| 834 |
+
"rewards/margins_min": 0.021772030740976334,
|
| 835 |
+
"rewards/margins_std": 0.05477788299322128,
|
| 836 |
+
"rewards/rejected": -0.019308075308799744,
|
| 837 |
+
"step": 450
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 0.5,
|
| 841 |
+
"grad_norm": 0.369140625,
|
| 842 |
+
"learning_rate": 2.968083263592782e-07,
|
| 843 |
+
"logits/chosen": -1.429099202156067,
|
| 844 |
+
"logits/rejected": -0.9603363275527954,
|
| 845 |
+
"logps/chosen": -226.94985961914062,
|
| 846 |
+
"logps/rejected": -231.18212890625,
|
| 847 |
+
"loss": 0.6669,
|
| 848 |
+
"rewards/accuracies": 0.949999988079071,
|
| 849 |
+
"rewards/chosen": 0.039319347590208054,
|
| 850 |
+
"rewards/margins": 0.0540069118142128,
|
| 851 |
+
"rewards/margins_max": 0.08096911013126373,
|
| 852 |
+
"rewards/margins_min": 0.027044707909226418,
|
| 853 |
+
"rewards/margins_std": 0.03813030570745468,
|
| 854 |
+
"rewards/rejected": -0.014687557704746723,
|
| 855 |
+
"step": 460
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 0.51,
|
| 859 |
+
"grad_norm": 0.3125,
|
| 860 |
+
"learning_rate": 2.875266038078136e-07,
|
| 861 |
+
"logits/chosen": -1.467827320098877,
|
| 862 |
+
"logits/rejected": -0.8274309039115906,
|
| 863 |
+
"logps/chosen": -262.02984619140625,
|
| 864 |
+
"logps/rejected": -258.05291748046875,
|
| 865 |
+
"loss": 0.6663,
|
| 866 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 867 |
+
"rewards/chosen": 0.047955792397260666,
|
| 868 |
+
"rewards/margins": 0.05926694720983505,
|
| 869 |
+
"rewards/margins_max": 0.08313200622797012,
|
| 870 |
+
"rewards/margins_min": 0.035401880741119385,
|
| 871 |
+
"rewards/margins_std": 0.033750299364328384,
|
| 872 |
+
"rewards/rejected": -0.011311152949929237,
|
| 873 |
+
"step": 470
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 0.52,
|
| 877 |
+
"grad_norm": 0.330078125,
|
| 878 |
+
"learning_rate": 2.781916391103417e-07,
|
| 879 |
+
"logits/chosen": -1.4126121997833252,
|
| 880 |
+
"logits/rejected": -1.062652826309204,
|
| 881 |
+
"logps/chosen": -312.4531555175781,
|
| 882 |
+
"logps/rejected": -325.8970642089844,
|
| 883 |
+
"loss": 0.6669,
|
| 884 |
+
"rewards/accuracies": 0.8999999761581421,
|
| 885 |
+
"rewards/chosen": 0.046898700296878815,
|
| 886 |
+
"rewards/margins": 0.05282552167773247,
|
| 887 |
+
"rewards/margins_max": 0.07998780906200409,
|
| 888 |
+
"rewards/margins_min": 0.025663232430815697,
|
| 889 |
+
"rewards/margins_std": 0.03841327875852585,
|
| 890 |
+
"rewards/rejected": -0.005926821380853653,
|
| 891 |
+
"step": 480
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.53,
|
| 895 |
+
"grad_norm": 0.3359375,
|
| 896 |
+
"learning_rate": 2.6881667656565226e-07,
|
| 897 |
+
"logits/chosen": -1.4687728881835938,
|
| 898 |
+
"logits/rejected": -0.979697048664093,
|
| 899 |
+
"logps/chosen": -241.0842742919922,
|
| 900 |
+
"logps/rejected": -232.9574432373047,
|
| 901 |
+
"loss": 0.6659,
|
| 902 |
+
"rewards/accuracies": 0.949999988079071,
|
| 903 |
+
"rewards/chosen": 0.054931215941905975,
|
| 904 |
+
"rewards/margins": 0.0482785627245903,
|
| 905 |
+
"rewards/margins_max": 0.0750807598233223,
|
| 906 |
+
"rewards/margins_min": 0.02147636190056801,
|
| 907 |
+
"rewards/margins_std": 0.037904031574726105,
|
| 908 |
+
"rewards/rejected": 0.006652662996202707,
|
| 909 |
+
"step": 490
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 0.54,
|
| 913 |
+
"grad_norm": 0.28515625,
|
| 914 |
+
"learning_rate": 2.594150172208416e-07,
|
| 915 |
+
"logits/chosen": -1.5185790061950684,
|
| 916 |
+
"logits/rejected": -0.9422761797904968,
|
| 917 |
+
"logps/chosen": -234.38119506835938,
|
| 918 |
+
"logps/rejected": -252.7394561767578,
|
| 919 |
+
"loss": 0.669,
|
| 920 |
+
"rewards/accuracies": 0.949999988079071,
|
| 921 |
+
"rewards/chosen": 0.04691288247704506,
|
| 922 |
+
"rewards/margins": 0.05338384583592415,
|
| 923 |
+
"rewards/margins_max": 0.08102253079414368,
|
| 924 |
+
"rewards/margins_min": 0.025745173916220665,
|
| 925 |
+
"rewards/margins_std": 0.03908699378371239,
|
| 926 |
+
"rewards/rejected": -0.006470963358879089,
|
| 927 |
+
"step": 500
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 0.55,
|
| 931 |
+
"grad_norm": 0.302734375,
|
| 932 |
+
"learning_rate": 2.5e-07,
|
| 933 |
+
"logits/chosen": -1.4481227397918701,
|
| 934 |
+
"logits/rejected": -1.0346171855926514,
|
| 935 |
+
"logps/chosen": -220.05325317382812,
|
| 936 |
+
"logps/rejected": -236.8842315673828,
|
| 937 |
+
"loss": 0.6634,
|
| 938 |
+
"rewards/accuracies": 0.949999988079071,
|
| 939 |
+
"rewards/chosen": 0.044769201427698135,
|
| 940 |
+
"rewards/margins": 0.05242365598678589,
|
| 941 |
+
"rewards/margins_max": 0.08209005743265152,
|
| 942 |
+
"rewards/margins_min": 0.022757260128855705,
|
| 943 |
+
"rewards/margins_std": 0.04195461794734001,
|
| 944 |
+
"rewards/rejected": -0.007654457353055477,
|
| 945 |
+
"step": 510
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 0.56,
|
| 949 |
+
"grad_norm": 0.31640625,
|
| 950 |
+
"learning_rate": 2.405849827791583e-07,
|
| 951 |
+
"logits/chosen": -1.3833669424057007,
|
| 952 |
+
"logits/rejected": -0.881574273109436,
|
| 953 |
+
"logps/chosen": -241.27481079101562,
|
| 954 |
+
"logps/rejected": -263.91351318359375,
|
| 955 |
+
"loss": 0.6641,
|
| 956 |
+
"rewards/accuracies": 1.0,
|
| 957 |
+
"rewards/chosen": 0.05363880842924118,
|
| 958 |
+
"rewards/margins": 0.065273717045784,
|
| 959 |
+
"rewards/margins_max": 0.09218905121088028,
|
| 960 |
+
"rewards/margins_min": 0.03835836052894592,
|
| 961 |
+
"rewards/margins_std": 0.03806404396891594,
|
| 962 |
+
"rewards/rejected": -0.011634895578026772,
|
| 963 |
+
"step": 520
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 0.57,
|
| 967 |
+
"grad_norm": 0.3671875,
|
| 968 |
+
"learning_rate": 2.3118332343434777e-07,
|
| 969 |
+
"logits/chosen": -1.5236294269561768,
|
| 970 |
+
"logits/rejected": -0.975500762462616,
|
| 971 |
+
"logps/chosen": -249.5443572998047,
|
| 972 |
+
"logps/rejected": -252.99618530273438,
|
| 973 |
+
"loss": 0.6649,
|
| 974 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 975 |
+
"rewards/chosen": 0.057201676070690155,
|
| 976 |
+
"rewards/margins": 0.06602860987186432,
|
| 977 |
+
"rewards/margins_max": 0.08838556706905365,
|
| 978 |
+
"rewards/margins_min": 0.04367166385054588,
|
| 979 |
+
"rewards/margins_std": 0.03161751106381416,
|
| 980 |
+
"rewards/rejected": -0.00882694311439991,
|
| 981 |
+
"step": 530
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 0.58,
|
| 985 |
+
"grad_norm": 0.2734375,
|
| 986 |
+
"learning_rate": 2.218083608896583e-07,
|
| 987 |
+
"logits/chosen": -1.4163635969161987,
|
| 988 |
+
"logits/rejected": -1.021444320678711,
|
| 989 |
+
"logps/chosen": -238.6166534423828,
|
| 990 |
+
"logps/rejected": -231.279296875,
|
| 991 |
+
"loss": 0.6678,
|
| 992 |
+
"rewards/accuracies": 0.8500000238418579,
|
| 993 |
+
"rewards/chosen": 0.04542135074734688,
|
| 994 |
+
"rewards/margins": 0.04465199261903763,
|
| 995 |
+
"rewards/margins_max": 0.07032604515552521,
|
| 996 |
+
"rewards/margins_min": 0.018977930769324303,
|
| 997 |
+
"rewards/margins_std": 0.03630860149860382,
|
| 998 |
+
"rewards/rejected": 0.000769357371609658,
|
| 999 |
+
"step": 540
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 0.59,
|
| 1003 |
+
"grad_norm": 0.298828125,
|
| 1004 |
+
"learning_rate": 2.1247339619218638e-07,
|
| 1005 |
+
"logits/chosen": -1.5570924282073975,
|
| 1006 |
+
"logits/rejected": -0.9521455764770508,
|
| 1007 |
+
"logps/chosen": -244.45877075195312,
|
| 1008 |
+
"logps/rejected": -218.03067016601562,
|
| 1009 |
+
"loss": 0.6633,
|
| 1010 |
+
"rewards/accuracies": 0.875,
|
| 1011 |
+
"rewards/chosen": 0.04918726533651352,
|
| 1012 |
+
"rewards/margins": 0.05797078087925911,
|
| 1013 |
+
"rewards/margins_max": 0.07724090665578842,
|
| 1014 |
+
"rewards/margins_min": 0.0387006476521492,
|
| 1015 |
+
"rewards/margins_std": 0.027252081781625748,
|
| 1016 |
+
"rewards/rejected": -0.008783518336713314,
|
| 1017 |
+
"step": 550
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.6,
|
| 1021 |
+
"grad_norm": 0.369140625,
|
| 1022 |
+
"learning_rate": 2.031916736407218e-07,
|
| 1023 |
+
"logits/chosen": -1.246797800064087,
|
| 1024 |
+
"logits/rejected": -0.8754084706306458,
|
| 1025 |
+
"logps/chosen": -255.4636688232422,
|
| 1026 |
+
"logps/rejected": -206.0786590576172,
|
| 1027 |
+
"loss": 0.6682,
|
| 1028 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1029 |
+
"rewards/chosen": 0.04116806760430336,
|
| 1030 |
+
"rewards/margins": 0.05079611390829086,
|
| 1031 |
+
"rewards/margins_max": 0.07820285856723785,
|
| 1032 |
+
"rewards/margins_min": 0.023389369249343872,
|
| 1033 |
+
"rewards/margins_std": 0.03875899314880371,
|
| 1034 |
+
"rewards/rejected": -0.009628048166632652,
|
| 1035 |
+
"step": 560
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 0.61,
|
| 1039 |
+
"grad_norm": 0.3125,
|
| 1040 |
+
"learning_rate": 1.9397636199494806e-07,
|
| 1041 |
+
"logits/chosen": -1.3417741060256958,
|
| 1042 |
+
"logits/rejected": -0.9931136965751648,
|
| 1043 |
+
"logps/chosen": -245.11184692382812,
|
| 1044 |
+
"logps/rejected": -274.5023498535156,
|
| 1045 |
+
"loss": 0.667,
|
| 1046 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1047 |
+
"rewards/chosen": 0.04023710638284683,
|
| 1048 |
+
"rewards/margins": 0.060276590287685394,
|
| 1049 |
+
"rewards/margins_max": 0.08814635127782822,
|
| 1050 |
+
"rewards/margins_min": 0.03240684047341347,
|
| 1051 |
+
"rewards/margins_std": 0.03941378742456436,
|
| 1052 |
+
"rewards/rejected": -0.02003948949277401,
|
| 1053 |
+
"step": 570
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 0.63,
|
| 1057 |
+
"grad_norm": 0.310546875,
|
| 1058 |
+
"learning_rate": 1.8484053579181658e-07,
|
| 1059 |
+
"logits/chosen": -1.4128963947296143,
|
| 1060 |
+
"logits/rejected": -0.9094281196594238,
|
| 1061 |
+
"logps/chosen": -241.9584503173828,
|
| 1062 |
+
"logps/rejected": -255.27676391601562,
|
| 1063 |
+
"loss": 0.6669,
|
| 1064 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1065 |
+
"rewards/chosen": 0.04914793372154236,
|
| 1066 |
+
"rewards/margins": 0.06206917762756348,
|
| 1067 |
+
"rewards/margins_max": 0.09373507648706436,
|
| 1068 |
+
"rewards/margins_min": 0.030403289943933487,
|
| 1069 |
+
"rewards/margins_std": 0.044782333076000214,
|
| 1070 |
+
"rewards/rejected": -0.012921245768666267,
|
| 1071 |
+
"step": 580
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 0.64,
|
| 1075 |
+
"grad_norm": 0.3828125,
|
| 1076 |
+
"learning_rate": 1.757971567956027e-07,
|
| 1077 |
+
"logits/chosen": -1.7144603729248047,
|
| 1078 |
+
"logits/rejected": -0.9307141304016113,
|
| 1079 |
+
"logps/chosen": -272.17547607421875,
|
| 1080 |
+
"logps/rejected": -241.65670776367188,
|
| 1081 |
+
"loss": 0.6637,
|
| 1082 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1083 |
+
"rewards/chosen": 0.05295403674244881,
|
| 1084 |
+
"rewards/margins": 0.05789683386683464,
|
| 1085 |
+
"rewards/margins_max": 0.08770729601383209,
|
| 1086 |
+
"rewards/margins_min": 0.028086364269256592,
|
| 1087 |
+
"rewards/margins_std": 0.04215836524963379,
|
| 1088 |
+
"rewards/rejected": -0.00494279433041811,
|
| 1089 |
+
"step": 590
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 0.65,
|
| 1093 |
+
"grad_norm": 0.2890625,
|
| 1094 |
+
"learning_rate": 1.6685905560796098e-07,
|
| 1095 |
+
"logits/chosen": -1.3933067321777344,
|
| 1096 |
+
"logits/rejected": -0.8825523257255554,
|
| 1097 |
+
"logps/chosen": -218.8330078125,
|
| 1098 |
+
"logps/rejected": -243.945556640625,
|
| 1099 |
+
"loss": 0.6669,
|
| 1100 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1101 |
+
"rewards/chosen": 0.044868241995573044,
|
| 1102 |
+
"rewards/margins": 0.04994089528918266,
|
| 1103 |
+
"rewards/margins_max": 0.07978564500808716,
|
| 1104 |
+
"rewards/margins_min": 0.020096149295568466,
|
| 1105 |
+
"rewards/margins_std": 0.04220684990286827,
|
| 1106 |
+
"rewards/rejected": -0.005072650499641895,
|
| 1107 |
+
"step": 600
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 0.66,
|
| 1111 |
+
"grad_norm": 0.384765625,
|
| 1112 |
+
"learning_rate": 1.580389134640734e-07,
|
| 1113 |
+
"logits/chosen": -1.4454293251037598,
|
| 1114 |
+
"logits/rejected": -1.0624114274978638,
|
| 1115 |
+
"logps/chosen": -232.04562377929688,
|
| 1116 |
+
"logps/rejected": -241.92013549804688,
|
| 1117 |
+
"loss": 0.6626,
|
| 1118 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1119 |
+
"rewards/chosen": 0.048992957919836044,
|
| 1120 |
+
"rewards/margins": 0.06253460794687271,
|
| 1121 |
+
"rewards/margins_max": 0.09021260589361191,
|
| 1122 |
+
"rewards/margins_min": 0.034856610000133514,
|
| 1123 |
+
"rewards/margins_std": 0.03914260491728783,
|
| 1124 |
+
"rewards/rejected": -0.013541650958359241,
|
| 1125 |
+
"step": 610
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 0.67,
|
| 1129 |
+
"grad_norm": 0.41796875,
|
| 1130 |
+
"learning_rate": 1.4934924424071475e-07,
|
| 1131 |
+
"logits/chosen": -1.5101556777954102,
|
| 1132 |
+
"logits/rejected": -0.8701263666152954,
|
| 1133 |
+
"logps/chosen": -268.46563720703125,
|
| 1134 |
+
"logps/rejected": -246.90725708007812,
|
| 1135 |
+
"loss": 0.6673,
|
| 1136 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1137 |
+
"rewards/chosen": 0.047949280589818954,
|
| 1138 |
+
"rewards/margins": 0.05510631948709488,
|
| 1139 |
+
"rewards/margins_max": 0.08622908592224121,
|
| 1140 |
+
"rewards/margins_min": 0.023983558639883995,
|
| 1141 |
+
"rewards/margins_std": 0.044014234095811844,
|
| 1142 |
+
"rewards/rejected": -0.00715703796595335,
|
| 1143 |
+
"step": 620
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 0.68,
|
| 1147 |
+
"grad_norm": 0.345703125,
|
| 1148 |
+
"learning_rate": 1.4080237670176453e-07,
|
| 1149 |
+
"logits/chosen": -1.5899397134780884,
|
| 1150 |
+
"logits/rejected": -0.9542080163955688,
|
| 1151 |
+
"logps/chosen": -250.47073364257812,
|
| 1152 |
+
"logps/rejected": -213.67672729492188,
|
| 1153 |
+
"loss": 0.6642,
|
| 1154 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1155 |
+
"rewards/chosen": 0.048125170171260834,
|
| 1156 |
+
"rewards/margins": 0.058349937200546265,
|
| 1157 |
+
"rewards/margins_max": 0.08438356220722198,
|
| 1158 |
+
"rewards/margins_min": 0.03231631591916084,
|
| 1159 |
+
"rewards/margins_std": 0.03681711107492447,
|
| 1160 |
+
"rewards/rejected": -0.010224771685898304,
|
| 1161 |
+
"step": 630
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 0.69,
|
| 1165 |
+
"grad_norm": 0.32421875,
|
| 1166 |
+
"learning_rate": 1.3241043700635352e-07,
|
| 1167 |
+
"logits/chosen": -1.4617611169815063,
|
| 1168 |
+
"logits/rejected": -0.715996265411377,
|
| 1169 |
+
"logps/chosen": -311.3377990722656,
|
| 1170 |
+
"logps/rejected": -235.23257446289062,
|
| 1171 |
+
"loss": 0.6615,
|
| 1172 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1173 |
+
"rewards/chosen": 0.049124039709568024,
|
| 1174 |
+
"rewards/margins": 0.06973692029714584,
|
| 1175 |
+
"rewards/margins_max": 0.09559544920921326,
|
| 1176 |
+
"rewards/margins_min": 0.04387838765978813,
|
| 1177 |
+
"rewards/margins_std": 0.03656948357820511,
|
| 1178 |
+
"rewards/rejected": -0.02061288245022297,
|
| 1179 |
+
"step": 640
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 0.7,
|
| 1183 |
+
"grad_norm": 0.341796875,
|
| 1184 |
+
"learning_rate": 1.2418533150446324e-07,
|
| 1185 |
+
"logits/chosen": -1.5678064823150635,
|
| 1186 |
+
"logits/rejected": -0.8574711680412292,
|
| 1187 |
+
"logps/chosen": -270.28997802734375,
|
| 1188 |
+
"logps/rejected": -229.6758575439453,
|
| 1189 |
+
"loss": 0.6643,
|
| 1190 |
+
"rewards/accuracies": 1.0,
|
| 1191 |
+
"rewards/chosen": 0.049732841551303864,
|
| 1192 |
+
"rewards/margins": 0.06726487725973129,
|
| 1193 |
+
"rewards/margins_max": 0.09402552247047424,
|
| 1194 |
+
"rewards/margins_min": 0.04050421714782715,
|
| 1195 |
+
"rewards/margins_std": 0.03784528002142906,
|
| 1196 |
+
"rewards/rejected": -0.01753203384578228,
|
| 1197 |
+
"step": 650
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 0.71,
|
| 1201 |
+
"grad_norm": 0.322265625,
|
| 1202 |
+
"learning_rate": 1.1613872984438628e-07,
|
| 1203 |
+
"logits/chosen": -1.5800104141235352,
|
| 1204 |
+
"logits/rejected": -0.9640630483627319,
|
| 1205 |
+
"logps/chosen": -217.96762084960938,
|
| 1206 |
+
"logps/rejected": -209.03237915039062,
|
| 1207 |
+
"loss": 0.6664,
|
| 1208 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1209 |
+
"rewards/chosen": 0.05414942651987076,
|
| 1210 |
+
"rewards/margins": 0.05594904348254204,
|
| 1211 |
+
"rewards/margins_max": 0.08529958873987198,
|
| 1212 |
+
"rewards/margins_min": 0.026598507538437843,
|
| 1213 |
+
"rewards/margins_std": 0.04150792956352234,
|
| 1214 |
+
"rewards/rejected": -0.0017996244132518768,
|
| 1215 |
+
"step": 660
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 0.72,
|
| 1219 |
+
"grad_norm": 0.412109375,
|
| 1220 |
+
"learning_rate": 1.0828204841601607e-07,
|
| 1221 |
+
"logits/chosen": -1.573081612586975,
|
| 1222 |
+
"logits/rejected": -1.0643428564071655,
|
| 1223 |
+
"logps/chosen": -267.5829772949219,
|
| 1224 |
+
"logps/rejected": -279.0192565917969,
|
| 1225 |
+
"loss": 0.6654,
|
| 1226 |
+
"rewards/accuracies": 1.0,
|
| 1227 |
+
"rewards/chosen": 0.05038607865571976,
|
| 1228 |
+
"rewards/margins": 0.059136874973773956,
|
| 1229 |
+
"rewards/margins_max": 0.08540179580450058,
|
| 1230 |
+
"rewards/margins_min": 0.032871946692466736,
|
| 1231 |
+
"rewards/margins_std": 0.037144217640161514,
|
| 1232 |
+
"rewards/rejected": -0.00875079445540905,
|
| 1233 |
+
"step": 670
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 0.73,
|
| 1237 |
+
"grad_norm": 0.3046875,
|
| 1238 |
+
"learning_rate": 1.0062643415345545e-07,
|
| 1239 |
+
"logits/chosen": -1.5550715923309326,
|
| 1240 |
+
"logits/rejected": -0.911180853843689,
|
| 1241 |
+
"logps/chosen": -234.7667999267578,
|
| 1242 |
+
"logps/rejected": -255.4190673828125,
|
| 1243 |
+
"loss": 0.6641,
|
| 1244 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1245 |
+
"rewards/chosen": 0.051492560654878616,
|
| 1246 |
+
"rewards/margins": 0.06578966230154037,
|
| 1247 |
+
"rewards/margins_max": 0.09979908168315887,
|
| 1248 |
+
"rewards/margins_min": 0.03178024664521217,
|
| 1249 |
+
"rewards/margins_std": 0.04809657856822014,
|
| 1250 |
+
"rewards/rejected": -0.014297107234597206,
|
| 1251 |
+
"step": 680
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 0.74,
|
| 1255 |
+
"grad_norm": 0.287109375,
|
| 1256 |
+
"learning_rate": 9.318274871992407e-08,
|
| 1257 |
+
"logits/chosen": -1.5272128582000732,
|
| 1258 |
+
"logits/rejected": -1.013564944267273,
|
| 1259 |
+
"logps/chosen": -241.36782836914062,
|
| 1260 |
+
"logps/rejected": -224.7821044921875,
|
| 1261 |
+
"loss": 0.6665,
|
| 1262 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1263 |
+
"rewards/chosen": 0.04677369445562363,
|
| 1264 |
+
"rewards/margins": 0.05521191284060478,
|
| 1265 |
+
"rewards/margins_max": 0.08911783993244171,
|
| 1266 |
+
"rewards/margins_min": 0.021305980160832405,
|
| 1267 |
+
"rewards/margins_std": 0.04795023053884506,
|
| 1268 |
+
"rewards/rejected": -0.008438214659690857,
|
| 1269 |
+
"step": 690
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 0.76,
|
| 1273 |
+
"grad_norm": 0.314453125,
|
| 1274 |
+
"learning_rate": 8.596155309740469e-08,
|
| 1275 |
+
"logits/chosen": -1.6033340692520142,
|
| 1276 |
+
"logits/rejected": -1.09574294090271,
|
| 1277 |
+
"logps/chosen": -246.16799926757812,
|
| 1278 |
+
"logps/rejected": -263.65142822265625,
|
| 1279 |
+
"loss": 0.6648,
|
| 1280 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1281 |
+
"rewards/chosen": 0.0439617782831192,
|
| 1282 |
+
"rewards/margins": 0.054900698363780975,
|
| 1283 |
+
"rewards/margins_max": 0.08236662298440933,
|
| 1284 |
+
"rewards/margins_min": 0.02743479050695896,
|
| 1285 |
+
"rewards/margins_std": 0.03884267061948776,
|
| 1286 |
+
"rewards/rejected": -0.010938925668597221,
|
| 1287 |
+
"step": 700
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 0.77,
|
| 1291 |
+
"grad_norm": 0.267578125,
|
| 1292 |
+
"learning_rate": 7.897309260289026e-08,
|
| 1293 |
+
"logits/chosen": -1.5482122898101807,
|
| 1294 |
+
"logits/rejected": -1.1510074138641357,
|
| 1295 |
+
"logps/chosen": -249.0963897705078,
|
| 1296 |
+
"logps/rejected": -259.163818359375,
|
| 1297 |
+
"loss": 0.6635,
|
| 1298 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1299 |
+
"rewards/chosen": 0.04723275452852249,
|
| 1300 |
+
"rewards/margins": 0.05302921682596207,
|
| 1301 |
+
"rewards/margins_max": 0.0721876472234726,
|
| 1302 |
+
"rewards/margins_min": 0.03387077525258064,
|
| 1303 |
+
"rewards/margins_std": 0.027094120159745216,
|
| 1304 |
+
"rewards/rejected": -0.005796459037810564,
|
| 1305 |
+
"step": 710
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 0.78,
|
| 1309 |
+
"grad_norm": 0.373046875,
|
| 1310 |
+
"learning_rate": 7.222728235249195e-08,
|
| 1311 |
+
"logits/chosen": -1.3384692668914795,
|
| 1312 |
+
"logits/rejected": -0.7518659830093384,
|
| 1313 |
+
"logps/chosen": -202.97393798828125,
|
| 1314 |
+
"logps/rejected": -189.1539764404297,
|
| 1315 |
+
"loss": 0.6669,
|
| 1316 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1317 |
+
"rewards/chosen": 0.03645472601056099,
|
| 1318 |
+
"rewards/margins": 0.05072823911905289,
|
| 1319 |
+
"rewards/margins_max": 0.0754196047782898,
|
| 1320 |
+
"rewards/margins_min": 0.026036862283945084,
|
| 1321 |
+
"rewards/margins_std": 0.03491886705160141,
|
| 1322 |
+
"rewards/rejected": -0.014273506589233875,
|
| 1323 |
+
"step": 720
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 0.79,
|
| 1327 |
+
"grad_norm": 0.28125,
|
| 1328 |
+
"learning_rate": 6.573369319403108e-08,
|
| 1329 |
+
"logits/chosen": -1.5845191478729248,
|
| 1330 |
+
"logits/rejected": -0.9599083065986633,
|
| 1331 |
+
"logps/chosen": -228.37417602539062,
|
| 1332 |
+
"logps/rejected": -237.9397430419922,
|
| 1333 |
+
"loss": 0.6648,
|
| 1334 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1335 |
+
"rewards/chosen": 0.05403770133852959,
|
| 1336 |
+
"rewards/margins": 0.06774094700813293,
|
| 1337 |
+
"rewards/margins_max": 0.09628431499004364,
|
| 1338 |
+
"rewards/margins_min": 0.03919757157564163,
|
| 1339 |
+
"rewards/margins_std": 0.04036641865968704,
|
| 1340 |
+
"rewards/rejected": -0.013703237287700176,
|
| 1341 |
+
"step": 730
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 0.8,
|
| 1345 |
+
"grad_norm": 0.357421875,
|
| 1346 |
+
"learning_rate": 5.9501538128072597e-08,
|
| 1347 |
+
"logits/chosen": -1.5676336288452148,
|
| 1348 |
+
"logits/rejected": -0.8547343015670776,
|
| 1349 |
+
"logps/chosen": -290.7181701660156,
|
| 1350 |
+
"logps/rejected": -234.4829559326172,
|
| 1351 |
+
"loss": 0.6666,
|
| 1352 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1353 |
+
"rewards/chosen": 0.046895284205675125,
|
| 1354 |
+
"rewards/margins": 0.059913188219070435,
|
| 1355 |
+
"rewards/margins_max": 0.08948854357004166,
|
| 1356 |
+
"rewards/margins_min": 0.030337834730744362,
|
| 1357 |
+
"rewards/margins_std": 0.04182586818933487,
|
| 1358 |
+
"rewards/rejected": -0.01301790215075016,
|
| 1359 |
+
"step": 740
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 0.81,
|
| 1363 |
+
"grad_norm": 0.34765625,
|
| 1364 |
+
"learning_rate": 5.353965923666742e-08,
|
| 1365 |
+
"logits/chosen": -1.3945667743682861,
|
| 1366 |
+
"logits/rejected": -0.8627855181694031,
|
| 1367 |
+
"logps/chosen": -309.19976806640625,
|
| 1368 |
+
"logps/rejected": -313.39263916015625,
|
| 1369 |
+
"loss": 0.6653,
|
| 1370 |
+
"rewards/accuracies": 1.0,
|
| 1371 |
+
"rewards/chosen": 0.052454281598329544,
|
| 1372 |
+
"rewards/margins": 0.06173131614923477,
|
| 1373 |
+
"rewards/margins_max": 0.08643205463886261,
|
| 1374 |
+
"rewards/margins_min": 0.037030577659606934,
|
| 1375 |
+
"rewards/margins_std": 0.03493211418390274,
|
| 1376 |
+
"rewards/rejected": -0.009277036413550377,
|
| 1377 |
+
"step": 750
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 0.82,
|
| 1381 |
+
"grad_norm": 0.3046875,
|
| 1382 |
+
"learning_rate": 4.7856515138347735e-08,
|
| 1383 |
+
"logits/chosen": -1.501372218132019,
|
| 1384 |
+
"logits/rejected": -0.781902015209198,
|
| 1385 |
+
"logps/chosen": -265.1309509277344,
|
| 1386 |
+
"logps/rejected": -230.6505126953125,
|
| 1387 |
+
"loss": 0.6627,
|
| 1388 |
+
"rewards/accuracies": 1.0,
|
| 1389 |
+
"rewards/chosen": 0.05485969036817551,
|
| 1390 |
+
"rewards/margins": 0.0646880641579628,
|
| 1391 |
+
"rewards/margins_max": 0.08704294264316559,
|
| 1392 |
+
"rewards/margins_min": 0.04233316332101822,
|
| 1393 |
+
"rewards/margins_std": 0.03161459416151047,
|
| 1394 |
+
"rewards/rejected": -0.009828361682593822,
|
| 1395 |
+
"step": 760
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 0.83,
|
| 1399 |
+
"grad_norm": 0.359375,
|
| 1400 |
+
"learning_rate": 4.2460168987173806e-08,
|
| 1401 |
+
"logits/chosen": -1.595336675643921,
|
| 1402 |
+
"logits/rejected": -0.9258917570114136,
|
| 1403 |
+
"logps/chosen": -295.4404602050781,
|
| 1404 |
+
"logps/rejected": -247.3825225830078,
|
| 1405 |
+
"loss": 0.6617,
|
| 1406 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1407 |
+
"rewards/chosen": 0.056455206125974655,
|
| 1408 |
+
"rewards/margins": 0.0680319294333458,
|
| 1409 |
+
"rewards/margins_max": 0.09642402827739716,
|
| 1410 |
+
"rewards/margins_min": 0.039639830589294434,
|
| 1411 |
+
"rewards/margins_std": 0.040152497589588165,
|
| 1412 |
+
"rewards/rejected": -0.011576727032661438,
|
| 1413 |
+
"step": 770
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"epoch": 0.84,
|
| 1417 |
+
"grad_norm": 0.326171875,
|
| 1418 |
+
"learning_rate": 3.7358277032860016e-08,
|
| 1419 |
+
"logits/chosen": -1.5608649253845215,
|
| 1420 |
+
"logits/rejected": -0.8827853202819824,
|
| 1421 |
+
"logps/chosen": -260.73944091796875,
|
| 1422 |
+
"logps/rejected": -277.09674072265625,
|
| 1423 |
+
"loss": 0.6638,
|
| 1424 |
+
"rewards/accuracies": 1.0,
|
| 1425 |
+
"rewards/chosen": 0.049758292734622955,
|
| 1426 |
+
"rewards/margins": 0.06226016953587532,
|
| 1427 |
+
"rewards/margins_max": 0.08762570470571518,
|
| 1428 |
+
"rewards/margins_min": 0.03689463064074516,
|
| 1429 |
+
"rewards/margins_std": 0.035872288048267365,
|
| 1430 |
+
"rewards/rejected": -0.012501873075962067,
|
| 1431 |
+
"step": 780
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 0.85,
|
| 1435 |
+
"grad_norm": 0.251953125,
|
| 1436 |
+
"learning_rate": 3.255807775821015e-08,
|
| 1437 |
+
"logits/chosen": -1.5679352283477783,
|
| 1438 |
+
"logits/rejected": -0.9218677282333374,
|
| 1439 |
+
"logps/chosen": -289.8990783691406,
|
| 1440 |
+
"logps/rejected": -239.71261596679688,
|
| 1441 |
+
"loss": 0.6657,
|
| 1442 |
+
"rewards/accuracies": 1.0,
|
| 1443 |
+
"rewards/chosen": 0.05259973928332329,
|
| 1444 |
+
"rewards/margins": 0.06116511672735214,
|
| 1445 |
+
"rewards/margins_max": 0.08521325886249542,
|
| 1446 |
+
"rewards/margins_min": 0.037116967141628265,
|
| 1447 |
+
"rewards/margins_std": 0.034009214490652084,
|
| 1448 |
+
"rewards/rejected": -0.00856537651270628,
|
| 1449 |
+
"step": 790
|
| 1450 |
+
},
|
| 1451 |
+
{
|
| 1452 |
+
"epoch": 0.86,
|
| 1453 |
+
"grad_norm": 0.306640625,
|
| 1454 |
+
"learning_rate": 2.8066381609273493e-08,
|
| 1455 |
+
"logits/chosen": -1.4278221130371094,
|
| 1456 |
+
"logits/rejected": -0.8260752558708191,
|
| 1457 |
+
"logps/chosen": -247.1492156982422,
|
| 1458 |
+
"logps/rejected": -225.48062133789062,
|
| 1459 |
+
"loss": 0.6659,
|
| 1460 |
+
"rewards/accuracies": 1.0,
|
| 1461 |
+
"rewards/chosen": 0.05115921422839165,
|
| 1462 |
+
"rewards/margins": 0.0625448077917099,
|
| 1463 |
+
"rewards/margins_max": 0.09311069548130035,
|
| 1464 |
+
"rewards/margins_min": 0.031978923827409744,
|
| 1465 |
+
"rewards/margins_std": 0.043226685374975204,
|
| 1466 |
+
"rewards/rejected": -0.011385595425963402,
|
| 1467 |
+
"step": 800
|
| 1468 |
+
},
|
| 1469 |
+
{
|
| 1470 |
+
"epoch": 0.87,
|
| 1471 |
+
"grad_norm": 0.3359375,
|
| 1472 |
+
"learning_rate": 2.3889561332792657e-08,
|
| 1473 |
+
"logits/chosen": -1.5297521352767944,
|
| 1474 |
+
"logits/rejected": -0.9847742319107056,
|
| 1475 |
+
"logps/chosen": -270.2022399902344,
|
| 1476 |
+
"logps/rejected": -246.3189239501953,
|
| 1477 |
+
"loss": 0.6656,
|
| 1478 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1479 |
+
"rewards/chosen": 0.04832616075873375,
|
| 1480 |
+
"rewards/margins": 0.057206034660339355,
|
| 1481 |
+
"rewards/margins_max": 0.08940082043409348,
|
| 1482 |
+
"rewards/margins_min": 0.025011247023940086,
|
| 1483 |
+
"rewards/margins_std": 0.04553030803799629,
|
| 1484 |
+
"rewards/rejected": -0.008879872970283031,
|
| 1485 |
+
"step": 810
|
| 1486 |
+
},
|
| 1487 |
+
{
|
| 1488 |
+
"epoch": 0.88,
|
| 1489 |
+
"grad_norm": 0.3203125,
|
| 1490 |
+
"learning_rate": 2.0033542934652675e-08,
|
| 1491 |
+
"logits/chosen": -1.5510307550430298,
|
| 1492 |
+
"logits/rejected": -0.856239914894104,
|
| 1493 |
+
"logps/chosen": -256.7835388183594,
|
| 1494 |
+
"logps/rejected": -223.23135375976562,
|
| 1495 |
+
"loss": 0.6654,
|
| 1496 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1497 |
+
"rewards/chosen": 0.052466489374637604,
|
| 1498 |
+
"rewards/margins": 0.058548521250486374,
|
| 1499 |
+
"rewards/margins_max": 0.08875375986099243,
|
| 1500 |
+
"rewards/margins_min": 0.028343280777335167,
|
| 1501 |
+
"rewards/margins_std": 0.042716652154922485,
|
| 1502 |
+
"rewards/rejected": -0.006082023028284311,
|
| 1503 |
+
"step": 820
|
| 1504 |
+
},
|
| 1505 |
+
{
|
| 1506 |
+
"epoch": 0.9,
|
| 1507 |
+
"grad_norm": 0.333984375,
|
| 1508 |
+
"learning_rate": 1.6503797272158282e-08,
|
| 1509 |
+
"logits/chosen": -1.4320178031921387,
|
| 1510 |
+
"logits/rejected": -1.0073983669281006,
|
| 1511 |
+
"logps/chosen": -241.7576904296875,
|
| 1512 |
+
"logps/rejected": -277.2112731933594,
|
| 1513 |
+
"loss": 0.6654,
|
| 1514 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1515 |
+
"rewards/chosen": 0.03875169903039932,
|
| 1516 |
+
"rewards/margins": 0.05952931568026543,
|
| 1517 |
+
"rewards/margins_max": 0.09110890328884125,
|
| 1518 |
+
"rewards/margins_min": 0.027949709445238113,
|
| 1519 |
+
"rewards/margins_std": 0.044660307466983795,
|
| 1520 |
+
"rewards/rejected": -0.020777616649866104,
|
| 1521 |
+
"step": 830
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 0.91,
|
| 1525 |
+
"grad_norm": 0.291015625,
|
| 1526 |
+
"learning_rate": 1.3305332292068705e-08,
|
| 1527 |
+
"logits/chosen": -1.5706042051315308,
|
| 1528 |
+
"logits/rejected": -1.0899040699005127,
|
| 1529 |
+
"logps/chosen": -276.74017333984375,
|
| 1530 |
+
"logps/rejected": -287.7333068847656,
|
| 1531 |
+
"loss": 0.6645,
|
| 1532 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1533 |
+
"rewards/chosen": 0.05088504031300545,
|
| 1534 |
+
"rewards/margins": 0.0576903410255909,
|
| 1535 |
+
"rewards/margins_max": 0.0889071449637413,
|
| 1536 |
+
"rewards/margins_min": 0.02647354081273079,
|
| 1537 |
+
"rewards/margins_std": 0.04414721950888634,
|
| 1538 |
+
"rewards/rejected": -0.006805300712585449,
|
| 1539 |
+
"step": 840
|
| 1540 |
+
},
|
| 1541 |
+
{
|
| 1542 |
+
"epoch": 0.92,
|
| 1543 |
+
"grad_norm": 0.33203125,
|
| 1544 |
+
"learning_rate": 1.0442685925403344e-08,
|
| 1545 |
+
"logits/chosen": -1.5475889444351196,
|
| 1546 |
+
"logits/rejected": -1.067118525505066,
|
| 1547 |
+
"logps/chosen": -247.9560089111328,
|
| 1548 |
+
"logps/rejected": -261.21209716796875,
|
| 1549 |
+
"loss": 0.6645,
|
| 1550 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1551 |
+
"rewards/chosen": 0.054555535316467285,
|
| 1552 |
+
"rewards/margins": 0.06061048060655594,
|
| 1553 |
+
"rewards/margins_max": 0.09440271556377411,
|
| 1554 |
+
"rewards/margins_min": 0.026818236336112022,
|
| 1555 |
+
"rewards/margins_std": 0.04778943955898285,
|
| 1556 |
+
"rewards/rejected": -0.00605494249612093,
|
| 1557 |
+
"step": 850
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 0.93,
|
| 1561 |
+
"grad_norm": 0.35546875,
|
| 1562 |
+
"learning_rate": 7.91991964909744e-09,
|
| 1563 |
+
"logits/chosen": -1.5974628925323486,
|
| 1564 |
+
"logits/rejected": -0.9376012682914734,
|
| 1565 |
+
"logps/chosen": -226.9661407470703,
|
| 1566 |
+
"logps/rejected": -215.2276153564453,
|
| 1567 |
+
"loss": 0.6649,
|
| 1568 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1569 |
+
"rewards/chosen": 0.04701418802142143,
|
| 1570 |
+
"rewards/margins": 0.06498473882675171,
|
| 1571 |
+
"rewards/margins_max": 0.08930108696222305,
|
| 1572 |
+
"rewards/margins_min": 0.040668390691280365,
|
| 1573 |
+
"rewards/margins_std": 0.034388504922389984,
|
| 1574 |
+
"rewards/rejected": -0.017970550805330276,
|
| 1575 |
+
"step": 860
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 0.94,
|
| 1579 |
+
"grad_norm": 0.259765625,
|
| 1580 |
+
"learning_rate": 5.740612723643401e-09,
|
| 1581 |
+
"logits/chosen": -1.5247188806533813,
|
| 1582 |
+
"logits/rejected": -1.0012189149856567,
|
| 1583 |
+
"logps/chosen": -215.85086059570312,
|
| 1584 |
+
"logps/rejected": -213.22756958007812,
|
| 1585 |
+
"loss": 0.6651,
|
| 1586 |
+
"rewards/accuracies": 0.925000011920929,
|
| 1587 |
+
"rewards/chosen": 0.03825182095170021,
|
| 1588 |
+
"rewards/margins": 0.05696084350347519,
|
| 1589 |
+
"rewards/margins_max": 0.08913502097129822,
|
| 1590 |
+
"rewards/margins_min": 0.024786660447716713,
|
| 1591 |
+
"rewards/margins_std": 0.045501161366701126,
|
| 1592 |
+
"rewards/rejected": -0.01870902255177498,
|
| 1593 |
+
"step": 870
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 0.95,
|
| 1597 |
+
"grad_norm": 0.3203125,
|
| 1598 |
+
"learning_rate": 3.907857114893359e-09,
|
| 1599 |
+
"logits/chosen": -1.526238203048706,
|
| 1600 |
+
"logits/rejected": -1.0228248834609985,
|
| 1601 |
+
"logps/chosen": -250.3122100830078,
|
| 1602 |
+
"logps/rejected": -268.1669616699219,
|
| 1603 |
+
"loss": 0.6645,
|
| 1604 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1605 |
+
"rewards/chosen": 0.05541209131479263,
|
| 1606 |
+
"rewards/margins": 0.06139000505208969,
|
| 1607 |
+
"rewards/margins_max": 0.08599219471216202,
|
| 1608 |
+
"rewards/margins_min": 0.03678782656788826,
|
| 1609 |
+
"rewards/margins_std": 0.03479274362325668,
|
| 1610 |
+
"rewards/rejected": -0.005977921187877655,
|
| 1611 |
+
"step": 880
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 0.96,
|
| 1615 |
+
"grad_norm": 0.390625,
|
| 1616 |
+
"learning_rate": 2.4242531072273255e-09,
|
| 1617 |
+
"logits/chosen": -1.6231883764266968,
|
| 1618 |
+
"logits/rejected": -1.0414937734603882,
|
| 1619 |
+
"logps/chosen": -242.6604766845703,
|
| 1620 |
+
"logps/rejected": -248.1051483154297,
|
| 1621 |
+
"loss": 0.6626,
|
| 1622 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1623 |
+
"rewards/chosen": 0.04938145726919174,
|
| 1624 |
+
"rewards/margins": 0.05951399728655815,
|
| 1625 |
+
"rewards/margins_max": 0.08650043606758118,
|
| 1626 |
+
"rewards/margins_min": 0.03252756968140602,
|
| 1627 |
+
"rewards/margins_std": 0.03816457465291023,
|
| 1628 |
+
"rewards/rejected": -0.01013254001736641,
|
| 1629 |
+
"step": 890
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 0.97,
|
| 1633 |
+
"grad_norm": 0.279296875,
|
| 1634 |
+
"learning_rate": 1.2919056143113061e-09,
|
| 1635 |
+
"logits/chosen": -1.6871249675750732,
|
| 1636 |
+
"logits/rejected": -1.0585591793060303,
|
| 1637 |
+
"logps/chosen": -217.6454620361328,
|
| 1638 |
+
"logps/rejected": -223.8054656982422,
|
| 1639 |
+
"loss": 0.6662,
|
| 1640 |
+
"rewards/accuracies": 0.949999988079071,
|
| 1641 |
+
"rewards/chosen": 0.048433009535074234,
|
| 1642 |
+
"rewards/margins": 0.06717512011528015,
|
| 1643 |
+
"rewards/margins_max": 0.09603999555110931,
|
| 1644 |
+
"rewards/margins_min": 0.03831023350358009,
|
| 1645 |
+
"rewards/margins_std": 0.04082110896706581,
|
| 1646 |
+
"rewards/rejected": -0.01874210312962532,
|
| 1647 |
+
"step": 900
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 0.98,
|
| 1651 |
+
"grad_norm": 0.326171875,
|
| 1652 |
+
"learning_rate": 5.124211926793575e-10,
|
| 1653 |
+
"logits/chosen": -1.413845419883728,
|
| 1654 |
+
"logits/rejected": -0.7665145993232727,
|
| 1655 |
+
"logps/chosen": -264.106689453125,
|
| 1656 |
+
"logps/rejected": -229.184326171875,
|
| 1657 |
+
"loss": 0.6668,
|
| 1658 |
+
"rewards/accuracies": 1.0,
|
| 1659 |
+
"rewards/chosen": 0.0459374263882637,
|
| 1660 |
+
"rewards/margins": 0.05880744382739067,
|
| 1661 |
+
"rewards/margins_max": 0.08354458957910538,
|
| 1662 |
+
"rewards/margins_min": 0.034070298075675964,
|
| 1663 |
+
"rewards/margins_std": 0.03498360887169838,
|
| 1664 |
+
"rewards/rejected": -0.012870019301772118,
|
| 1665 |
+
"step": 910
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 0.99,
|
| 1669 |
+
"grad_norm": 0.341796875,
|
| 1670 |
+
"learning_rate": 8.690576237688207e-11,
|
| 1671 |
+
"logits/chosen": -1.4011876583099365,
|
| 1672 |
+
"logits/rejected": -0.9543458819389343,
|
| 1673 |
+
"logps/chosen": -281.8406677246094,
|
| 1674 |
+
"logps/rejected": -222.5669708251953,
|
| 1675 |
+
"loss": 0.6665,
|
| 1676 |
+
"rewards/accuracies": 0.9750000238418579,
|
| 1677 |
+
"rewards/chosen": 0.045102305710315704,
|
| 1678 |
+
"rewards/margins": 0.05106315761804581,
|
| 1679 |
+
"rewards/margins_max": 0.07114427536725998,
|
| 1680 |
+
"rewards/margins_min": 0.03098202869296074,
|
| 1681 |
+
"rewards/margins_std": 0.028398994356393814,
|
| 1682 |
+
"rewards/rejected": -0.0059608458541333675,
|
| 1683 |
+
"step": 920
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 1.0,
|
| 1687 |
+
"eval_logits/chosen": -1.0221611261367798,
|
| 1688 |
+
"eval_logits/rejected": -0.8962497115135193,
|
| 1689 |
+
"eval_logps/chosen": -331.233642578125,
|
| 1690 |
+
"eval_logps/rejected": -327.55169677734375,
|
| 1691 |
+
"eval_loss": 0.6933034658432007,
|
| 1692 |
+
"eval_rewards/accuracies": 0.5360000133514404,
|
| 1693 |
+
"eval_rewards/chosen": 0.026766540482640266,
|
| 1694 |
+
"eval_rewards/margins": 0.0002464489371050149,
|
| 1695 |
+
"eval_rewards/margins_max": 0.06006291136145592,
|
| 1696 |
+
"eval_rewards/margins_min": -0.061876267194747925,
|
| 1697 |
+
"eval_rewards/margins_std": 0.04059867188334465,
|
| 1698 |
+
"eval_rewards/rejected": 0.02652009204030037,
|
| 1699 |
+
"eval_runtime": 750.1106,
|
| 1700 |
+
"eval_samples_per_second": 5.333,
|
| 1701 |
+
"eval_steps_per_second": 0.167,
|
| 1702 |
+
"step": 927
|
| 1703 |
+
},
|
| 1704 |
+
{
|
| 1705 |
+
"epoch": 1.0,
|
| 1706 |
+
"step": 927,
|
| 1707 |
+
"total_flos": 0.0,
|
| 1708 |
+
"train_loss": 0.6720605961327414,
|
| 1709 |
+
"train_runtime": 8514.5547,
|
| 1710 |
+
"train_samples_per_second": 1.741,
|
| 1711 |
+
"train_steps_per_second": 0.109
|
| 1712 |
+
}
|
| 1713 |
+
],
|
| 1714 |
+
"logging_steps": 10,
|
| 1715 |
+
"max_steps": 927,
|
| 1716 |
+
"num_input_tokens_seen": 0,
|
| 1717 |
+
"num_train_epochs": 1,
|
| 1718 |
+
"save_steps": 100,
|
| 1719 |
+
"total_flos": 0.0,
|
| 1720 |
+
"train_batch_size": 2,
|
| 1721 |
+
"trial_name": null,
|
| 1722 |
+
"trial_params": null
|
| 1723 |
+
}
|