Model save
Browse files- README.md +82 -0
- all_results.json +9 -0
- generation_config.json +6 -0
- train_results.json +9 -0
- trainer_state.json +1626 -0
README.md
ADDED
|
@@ -0,0 +1,82 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
license: apache-2.0
|
| 4 |
+
base_model: alignment-handbook/zephyr-7b-sft-full
|
| 5 |
+
tags:
|
| 6 |
+
- trl
|
| 7 |
+
- dpo
|
| 8 |
+
- generated_from_trainer
|
| 9 |
+
model-index:
|
| 10 |
+
- name: zephyr-7b-dpo-full
|
| 11 |
+
results: []
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 15 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 16 |
+
|
| 17 |
+
# zephyr-7b-dpo-full
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on the None dataset.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 0.7707
|
| 22 |
+
- Rewards/chosen: -1.5789
|
| 23 |
+
- Rewards/rejected: -2.6963
|
| 24 |
+
- Rewards/accuracies: 0.7857
|
| 25 |
+
- Rewards/margins: 1.1174
|
| 26 |
+
- Logps/rejected: -529.8352
|
| 27 |
+
- Logps/chosen: -439.8712
|
| 28 |
+
- Logits/rejected: 1.9213
|
| 29 |
+
- Logits/chosen: 0.5837
|
| 30 |
+
|
| 31 |
+
## Model description
|
| 32 |
+
|
| 33 |
+
More information needed
|
| 34 |
+
|
| 35 |
+
## Intended uses & limitations
|
| 36 |
+
|
| 37 |
+
More information needed
|
| 38 |
+
|
| 39 |
+
## Training and evaluation data
|
| 40 |
+
|
| 41 |
+
More information needed
|
| 42 |
+
|
| 43 |
+
## Training procedure
|
| 44 |
+
|
| 45 |
+
### Training hyperparameters
|
| 46 |
+
|
| 47 |
+
The following hyperparameters were used during training:
|
| 48 |
+
- learning_rate: 5e-07
|
| 49 |
+
- train_batch_size: 8
|
| 50 |
+
- eval_batch_size: 8
|
| 51 |
+
- seed: 42
|
| 52 |
+
- distributed_type: multi-GPU
|
| 53 |
+
- num_devices: 4
|
| 54 |
+
- gradient_accumulation_steps: 2
|
| 55 |
+
- total_train_batch_size: 64
|
| 56 |
+
- total_eval_batch_size: 32
|
| 57 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 58 |
+
- lr_scheduler_type: cosine
|
| 59 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 60 |
+
- num_epochs: 1
|
| 61 |
+
|
| 62 |
+
### Training results
|
| 63 |
+
|
| 64 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 65 |
+
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 66 |
+
| 0.9314 | 0.1047 | 100 | 0.9248 | -0.1341 | -0.3626 | 0.7103 | 0.2286 | -296.4655 | -295.3837 | -2.3941 | -2.4608 |
|
| 67 |
+
| 0.8722 | 0.2093 | 200 | 0.8865 | -0.5236 | -1.0726 | 0.7520 | 0.5489 | -367.4581 | -334.3411 | -1.6165 | -1.9828 |
|
| 68 |
+
| 0.8208 | 0.3140 | 300 | 0.8215 | -0.8865 | -1.6789 | 0.7639 | 0.7924 | -428.0927 | -370.6316 | 0.3061 | -0.4089 |
|
| 69 |
+
| 0.8208 | 0.4186 | 400 | 0.7982 | -1.1907 | -1.9986 | 0.7718 | 0.8079 | -460.0637 | -401.0516 | 0.5905 | -0.4262 |
|
| 70 |
+
| 0.7826 | 0.5233 | 500 | 0.7799 | -1.3975 | -2.4383 | 0.7758 | 1.0408 | -504.0349 | -421.7270 | 2.2339 | 1.0156 |
|
| 71 |
+
| 0.7546 | 0.6279 | 600 | 0.7723 | -1.5567 | -2.6664 | 0.7837 | 1.1097 | -526.8406 | -437.6459 | 1.6798 | 0.3290 |
|
| 72 |
+
| 0.7533 | 0.7326 | 700 | 0.7732 | -1.6247 | -2.7103 | 0.7837 | 1.0856 | -531.2306 | -444.4420 | 2.0190 | 0.6982 |
|
| 73 |
+
| 0.7498 | 0.8373 | 800 | 0.7710 | -1.5564 | -2.6320 | 0.7857 | 1.0756 | -523.4053 | -437.6152 | 1.7010 | 0.4246 |
|
| 74 |
+
| 0.7471 | 0.9419 | 900 | 0.7707 | -1.5789 | -2.6963 | 0.7857 | 1.1174 | -529.8352 | -439.8712 | 1.9213 | 0.5837 |
|
| 75 |
+
|
| 76 |
+
|
| 77 |
+
### Framework versions
|
| 78 |
+
|
| 79 |
+
- Transformers 4.44.2
|
| 80 |
+
- Pytorch 2.2.1+cu118
|
| 81 |
+
- Datasets 2.14.7
|
| 82 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9994767137624281,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.8142244146756477,
|
| 5 |
+
"train_runtime": 34283.0174,
|
| 6 |
+
"train_samples": 61134,
|
| 7 |
+
"train_samples_per_second": 1.783,
|
| 8 |
+
"train_steps_per_second": 0.028
|
| 9 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"eos_token_id": 2,
|
| 5 |
+
"transformers_version": "4.44.2"
|
| 6 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9994767137624281,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.8142244146756477,
|
| 5 |
+
"train_runtime": 34283.0174,
|
| 6 |
+
"train_samples": 61134,
|
| 7 |
+
"train_samples_per_second": 1.783,
|
| 8 |
+
"train_steps_per_second": 0.028
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1626 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9994767137624281,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 955,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0010465724751439038,
|
| 13 |
+
"grad_norm": 12.684210249452411,
|
| 14 |
+
"learning_rate": 5.208333333333333e-09,
|
| 15 |
+
"logits/chosen": -2.9233341217041016,
|
| 16 |
+
"logits/rejected": -2.7917747497558594,
|
| 17 |
+
"logps/chosen": -380.82366943359375,
|
| 18 |
+
"logps/rejected": -358.487060546875,
|
| 19 |
+
"loss": 1.0,
|
| 20 |
+
"rewards/accuracies": 0.0,
|
| 21 |
+
"rewards/chosen": 0.0,
|
| 22 |
+
"rewards/margins": 0.0,
|
| 23 |
+
"rewards/rejected": 0.0,
|
| 24 |
+
"step": 1
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.010465724751439037,
|
| 28 |
+
"grad_norm": 10.879665963860996,
|
| 29 |
+
"learning_rate": 5.208333333333333e-08,
|
| 30 |
+
"logits/chosen": -2.5956761837005615,
|
| 31 |
+
"logits/rejected": -2.569300413131714,
|
| 32 |
+
"logps/chosen": -256.6047058105469,
|
| 33 |
+
"logps/rejected": -234.93711853027344,
|
| 34 |
+
"loss": 0.9998,
|
| 35 |
+
"rewards/accuracies": 0.4722222089767456,
|
| 36 |
+
"rewards/chosen": 0.00017333232972305268,
|
| 37 |
+
"rewards/margins": 0.0007588959997519851,
|
| 38 |
+
"rewards/rejected": -0.0005855637136846781,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.020931449502878074,
|
| 43 |
+
"grad_norm": 12.660694439558682,
|
| 44 |
+
"learning_rate": 1.0416666666666667e-07,
|
| 45 |
+
"logits/chosen": -2.613247871398926,
|
| 46 |
+
"logits/rejected": -2.5758299827575684,
|
| 47 |
+
"logps/chosen": -283.08770751953125,
|
| 48 |
+
"logps/rejected": -282.29901123046875,
|
| 49 |
+
"loss": 0.9998,
|
| 50 |
+
"rewards/accuracies": 0.5,
|
| 51 |
+
"rewards/chosen": 0.00026421304210089147,
|
| 52 |
+
"rewards/margins": 0.00011151668149977922,
|
| 53 |
+
"rewards/rejected": 0.00015269630239345133,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.03139717425431711,
|
| 58 |
+
"grad_norm": 12.337952256825348,
|
| 59 |
+
"learning_rate": 1.5624999999999999e-07,
|
| 60 |
+
"logits/chosen": -2.691951036453247,
|
| 61 |
+
"logits/rejected": -2.667614698410034,
|
| 62 |
+
"logps/chosen": -270.21954345703125,
|
| 63 |
+
"logps/rejected": -276.70257568359375,
|
| 64 |
+
"loss": 0.9992,
|
| 65 |
+
"rewards/accuracies": 0.53125,
|
| 66 |
+
"rewards/chosen": 0.00038629298796877265,
|
| 67 |
+
"rewards/margins": 0.0007968698628246784,
|
| 68 |
+
"rewards/rejected": -0.00041057687485590577,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.04186289900575615,
|
| 73 |
+
"grad_norm": 11.876881014853103,
|
| 74 |
+
"learning_rate": 2.0833333333333333e-07,
|
| 75 |
+
"logits/chosen": -2.664839267730713,
|
| 76 |
+
"logits/rejected": -2.589832067489624,
|
| 77 |
+
"logps/chosen": -290.53656005859375,
|
| 78 |
+
"logps/rejected": -282.1867980957031,
|
| 79 |
+
"loss": 0.9971,
|
| 80 |
+
"rewards/accuracies": 0.6312500238418579,
|
| 81 |
+
"rewards/chosen": 0.004319709725677967,
|
| 82 |
+
"rewards/margins": 0.005323711317032576,
|
| 83 |
+
"rewards/rejected": -0.0010040017077699304,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.052328623757195186,
|
| 88 |
+
"grad_norm": 13.270256617763115,
|
| 89 |
+
"learning_rate": 2.604166666666667e-07,
|
| 90 |
+
"logits/chosen": -2.6715073585510254,
|
| 91 |
+
"logits/rejected": -2.5877907276153564,
|
| 92 |
+
"logps/chosen": -266.10845947265625,
|
| 93 |
+
"logps/rejected": -236.5567626953125,
|
| 94 |
+
"loss": 0.9929,
|
| 95 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 96 |
+
"rewards/chosen": 0.015197384171187878,
|
| 97 |
+
"rewards/margins": 0.015182172879576683,
|
| 98 |
+
"rewards/rejected": 1.5211687241389882e-05,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.06279434850863422,
|
| 103 |
+
"grad_norm": 11.894246837548359,
|
| 104 |
+
"learning_rate": 3.1249999999999997e-07,
|
| 105 |
+
"logits/chosen": -2.6269617080688477,
|
| 106 |
+
"logits/rejected": -2.5917298793792725,
|
| 107 |
+
"logps/chosen": -299.65716552734375,
|
| 108 |
+
"logps/rejected": -274.43682861328125,
|
| 109 |
+
"loss": 0.9862,
|
| 110 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 111 |
+
"rewards/chosen": 0.04350100830197334,
|
| 112 |
+
"rewards/margins": 0.029931485652923584,
|
| 113 |
+
"rewards/rejected": 0.013569521717727184,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.07326007326007326,
|
| 118 |
+
"grad_norm": 11.94629840214185,
|
| 119 |
+
"learning_rate": 3.645833333333333e-07,
|
| 120 |
+
"logits/chosen": -2.52927827835083,
|
| 121 |
+
"logits/rejected": -2.521303653717041,
|
| 122 |
+
"logps/chosen": -257.636474609375,
|
| 123 |
+
"logps/rejected": -262.7804260253906,
|
| 124 |
+
"loss": 0.9774,
|
| 125 |
+
"rewards/accuracies": 0.6875,
|
| 126 |
+
"rewards/chosen": 0.019934870302677155,
|
| 127 |
+
"rewards/margins": 0.0652446299791336,
|
| 128 |
+
"rewards/rejected": -0.04530975967645645,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.0837257980115123,
|
| 133 |
+
"grad_norm": 13.797354905847879,
|
| 134 |
+
"learning_rate": 4.1666666666666667e-07,
|
| 135 |
+
"logits/chosen": -2.543499708175659,
|
| 136 |
+
"logits/rejected": -2.4710183143615723,
|
| 137 |
+
"logps/chosen": -274.4893493652344,
|
| 138 |
+
"logps/rejected": -261.1844482421875,
|
| 139 |
+
"loss": 0.9567,
|
| 140 |
+
"rewards/accuracies": 0.706250011920929,
|
| 141 |
+
"rewards/chosen": 0.015994269400835037,
|
| 142 |
+
"rewards/margins": 0.11483701318502426,
|
| 143 |
+
"rewards/rejected": -0.09884275496006012,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.09419152276295134,
|
| 148 |
+
"grad_norm": 15.051298123886353,
|
| 149 |
+
"learning_rate": 4.6874999999999996e-07,
|
| 150 |
+
"logits/chosen": -2.483623504638672,
|
| 151 |
+
"logits/rejected": -2.4481821060180664,
|
| 152 |
+
"logps/chosen": -260.9582824707031,
|
| 153 |
+
"logps/rejected": -278.18963623046875,
|
| 154 |
+
"loss": 0.9481,
|
| 155 |
+
"rewards/accuracies": 0.668749988079071,
|
| 156 |
+
"rewards/chosen": -0.05396522209048271,
|
| 157 |
+
"rewards/margins": 0.08080872148275375,
|
| 158 |
+
"rewards/rejected": -0.13477393984794617,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.10465724751439037,
|
| 163 |
+
"grad_norm": 13.876516756747606,
|
| 164 |
+
"learning_rate": 4.999732492681437e-07,
|
| 165 |
+
"logits/chosen": -2.4570467472076416,
|
| 166 |
+
"logits/rejected": -2.381937026977539,
|
| 167 |
+
"logps/chosen": -317.0618591308594,
|
| 168 |
+
"logps/rejected": -315.6556396484375,
|
| 169 |
+
"loss": 0.9314,
|
| 170 |
+
"rewards/accuracies": 0.731249988079071,
|
| 171 |
+
"rewards/chosen": -0.17215368151664734,
|
| 172 |
+
"rewards/margins": 0.18427203595638275,
|
| 173 |
+
"rewards/rejected": -0.35642576217651367,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.10465724751439037,
|
| 178 |
+
"eval_logits/chosen": -2.4607555866241455,
|
| 179 |
+
"eval_logits/rejected": -2.3940787315368652,
|
| 180 |
+
"eval_logps/chosen": -295.3836669921875,
|
| 181 |
+
"eval_logps/rejected": -296.4655456542969,
|
| 182 |
+
"eval_loss": 0.9247981905937195,
|
| 183 |
+
"eval_rewards/accuracies": 0.7103174328804016,
|
| 184 |
+
"eval_rewards/chosen": -0.1340663731098175,
|
| 185 |
+
"eval_rewards/margins": 0.22856087982654572,
|
| 186 |
+
"eval_rewards/rejected": -0.3626272976398468,
|
| 187 |
+
"eval_runtime": 410.2228,
|
| 188 |
+
"eval_samples_per_second": 4.875,
|
| 189 |
+
"eval_steps_per_second": 0.154,
|
| 190 |
+
"step": 100
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.1151229722658294,
|
| 194 |
+
"grad_norm": 20.34786259548405,
|
| 195 |
+
"learning_rate": 4.996723692767926e-07,
|
| 196 |
+
"logits/chosen": -2.4748752117156982,
|
| 197 |
+
"logits/rejected": -2.420341730117798,
|
| 198 |
+
"logps/chosen": -263.1013488769531,
|
| 199 |
+
"logps/rejected": -280.06256103515625,
|
| 200 |
+
"loss": 0.9303,
|
| 201 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 202 |
+
"rewards/chosen": -0.21624498069286346,
|
| 203 |
+
"rewards/margins": 0.21136274933815002,
|
| 204 |
+
"rewards/rejected": -0.4276077151298523,
|
| 205 |
+
"step": 110
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.12558869701726844,
|
| 209 |
+
"grad_norm": 18.011468034651415,
|
| 210 |
+
"learning_rate": 4.990375746213598e-07,
|
| 211 |
+
"logits/chosen": -2.4168341159820557,
|
| 212 |
+
"logits/rejected": -2.353346109390259,
|
| 213 |
+
"logps/chosen": -284.01434326171875,
|
| 214 |
+
"logps/rejected": -333.86517333984375,
|
| 215 |
+
"loss": 0.9134,
|
| 216 |
+
"rewards/accuracies": 0.731249988079071,
|
| 217 |
+
"rewards/chosen": -0.2143399715423584,
|
| 218 |
+
"rewards/margins": 0.23967032134532928,
|
| 219 |
+
"rewards/rejected": -0.4540103077888489,
|
| 220 |
+
"step": 120
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.1360544217687075,
|
| 224 |
+
"grad_norm": 20.20595442453097,
|
| 225 |
+
"learning_rate": 4.980697142834314e-07,
|
| 226 |
+
"logits/chosen": -2.333416223526001,
|
| 227 |
+
"logits/rejected": -2.272174835205078,
|
| 228 |
+
"logps/chosen": -326.86810302734375,
|
| 229 |
+
"logps/rejected": -334.0741271972656,
|
| 230 |
+
"loss": 0.9077,
|
| 231 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 232 |
+
"rewards/chosen": -0.45602983236312866,
|
| 233 |
+
"rewards/margins": 0.34080833196640015,
|
| 234 |
+
"rewards/rejected": -0.7968382239341736,
|
| 235 |
+
"step": 130
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.14652014652014653,
|
| 239 |
+
"grad_norm": 25.245651601287875,
|
| 240 |
+
"learning_rate": 4.967700826904229e-07,
|
| 241 |
+
"logits/chosen": -1.9765796661376953,
|
| 242 |
+
"logits/rejected": -1.9096320867538452,
|
| 243 |
+
"logps/chosen": -344.718505859375,
|
| 244 |
+
"logps/rejected": -383.64544677734375,
|
| 245 |
+
"loss": 0.8734,
|
| 246 |
+
"rewards/accuracies": 0.737500011920929,
|
| 247 |
+
"rewards/chosen": -0.5360048413276672,
|
| 248 |
+
"rewards/margins": 0.41050204634666443,
|
| 249 |
+
"rewards/rejected": -0.9465070962905884,
|
| 250 |
+
"step": 140
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.15698587127158556,
|
| 254 |
+
"grad_norm": 21.269931947577092,
|
| 255 |
+
"learning_rate": 4.951404179843962e-07,
|
| 256 |
+
"logits/chosen": -2.0397307872772217,
|
| 257 |
+
"logits/rejected": -1.8855133056640625,
|
| 258 |
+
"logps/chosen": -348.33074951171875,
|
| 259 |
+
"logps/rejected": -355.48370361328125,
|
| 260 |
+
"loss": 0.8763,
|
| 261 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 262 |
+
"rewards/chosen": -0.6048094034194946,
|
| 263 |
+
"rewards/margins": 0.4798543453216553,
|
| 264 |
+
"rewards/rejected": -1.0846636295318604,
|
| 265 |
+
"step": 150
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.1674515960230246,
|
| 269 |
+
"grad_norm": 33.533617020269915,
|
| 270 |
+
"learning_rate": 4.931828996974498e-07,
|
| 271 |
+
"logits/chosen": -2.1307215690612793,
|
| 272 |
+
"logits/rejected": -1.8285764455795288,
|
| 273 |
+
"logps/chosen": -352.92236328125,
|
| 274 |
+
"logps/rejected": -371.9589538574219,
|
| 275 |
+
"loss": 0.8427,
|
| 276 |
+
"rewards/accuracies": 0.731249988079071,
|
| 277 |
+
"rewards/chosen": -0.5708774924278259,
|
| 278 |
+
"rewards/margins": 0.5884283781051636,
|
| 279 |
+
"rewards/rejected": -1.1593058109283447,
|
| 280 |
+
"step": 160
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.17791732077446362,
|
| 284 |
+
"grad_norm": 32.757872064459335,
|
| 285 |
+
"learning_rate": 4.909001458367866e-07,
|
| 286 |
+
"logits/chosen": -1.2587175369262695,
|
| 287 |
+
"logits/rejected": -0.8609746098518372,
|
| 288 |
+
"logps/chosen": -408.7366027832031,
|
| 289 |
+
"logps/rejected": -442.3734436035156,
|
| 290 |
+
"loss": 0.8506,
|
| 291 |
+
"rewards/accuracies": 0.731249988079071,
|
| 292 |
+
"rewards/chosen": -1.3020718097686768,
|
| 293 |
+
"rewards/margins": 0.713103175163269,
|
| 294 |
+
"rewards/rejected": -2.015174627304077,
|
| 295 |
+
"step": 170
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 0.18838304552590268,
|
| 299 |
+
"grad_norm": 39.16606354469095,
|
| 300 |
+
"learning_rate": 4.882952093833627e-07,
|
| 301 |
+
"logits/chosen": -1.010235071182251,
|
| 302 |
+
"logits/rejected": -0.6893966794013977,
|
| 303 |
+
"logps/chosen": -386.0320739746094,
|
| 304 |
+
"logps/rejected": -459.80596923828125,
|
| 305 |
+
"loss": 0.8371,
|
| 306 |
+
"rewards/accuracies": 0.731249988079071,
|
| 307 |
+
"rewards/chosen": -1.4610180854797363,
|
| 308 |
+
"rewards/margins": 0.6689072847366333,
|
| 309 |
+
"rewards/rejected": -2.12992525100708,
|
| 310 |
+
"step": 180
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.1988487702773417,
|
| 314 |
+
"grad_norm": 38.44785852081829,
|
| 315 |
+
"learning_rate": 4.853715742087946e-07,
|
| 316 |
+
"logits/chosen": -1.110741138458252,
|
| 317 |
+
"logits/rejected": -0.910930335521698,
|
| 318 |
+
"logps/chosen": -359.8037414550781,
|
| 319 |
+
"logps/rejected": -432.5801696777344,
|
| 320 |
+
"loss": 0.8426,
|
| 321 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 322 |
+
"rewards/chosen": -1.2385684251785278,
|
| 323 |
+
"rewards/margins": 0.5880638360977173,
|
| 324 |
+
"rewards/rejected": -1.8266319036483765,
|
| 325 |
+
"step": 190
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.20931449502878074,
|
| 329 |
+
"grad_norm": 29.500721336346615,
|
| 330 |
+
"learning_rate": 4.821331504159906e-07,
|
| 331 |
+
"logits/chosen": -1.7968833446502686,
|
| 332 |
+
"logits/rejected": -1.3632351160049438,
|
| 333 |
+
"logps/chosen": -372.0254211425781,
|
| 334 |
+
"logps/rejected": -369.52032470703125,
|
| 335 |
+
"loss": 0.8722,
|
| 336 |
+
"rewards/accuracies": 0.6875,
|
| 337 |
+
"rewards/chosen": -0.6137913465499878,
|
| 338 |
+
"rewards/margins": 0.5101683139801025,
|
| 339 |
+
"rewards/rejected": -1.1239596605300903,
|
| 340 |
+
"step": 200
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.20931449502878074,
|
| 344 |
+
"eval_logits/chosen": -1.9828360080718994,
|
| 345 |
+
"eval_logits/rejected": -1.6164638996124268,
|
| 346 |
+
"eval_logps/chosen": -334.34112548828125,
|
| 347 |
+
"eval_logps/rejected": -367.4581298828125,
|
| 348 |
+
"eval_loss": 0.8864663243293762,
|
| 349 |
+
"eval_rewards/accuracies": 0.7519841194152832,
|
| 350 |
+
"eval_rewards/chosen": -0.5236411094665527,
|
| 351 |
+
"eval_rewards/margins": 0.5489121079444885,
|
| 352 |
+
"eval_rewards/rejected": -1.072553277015686,
|
| 353 |
+
"eval_runtime": 405.4284,
|
| 354 |
+
"eval_samples_per_second": 4.933,
|
| 355 |
+
"eval_steps_per_second": 0.155,
|
| 356 |
+
"step": 200
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.21978021978021978,
|
| 360 |
+
"grad_norm": 33.64481516454493,
|
| 361 |
+
"learning_rate": 4.785842691097342e-07,
|
| 362 |
+
"logits/chosen": -1.8171882629394531,
|
| 363 |
+
"logits/rejected": -1.3618426322937012,
|
| 364 |
+
"logps/chosen": -348.45599365234375,
|
| 365 |
+
"logps/rejected": -389.9690856933594,
|
| 366 |
+
"loss": 0.8788,
|
| 367 |
+
"rewards/accuracies": 0.643750011920929,
|
| 368 |
+
"rewards/chosen": -0.6564497351646423,
|
| 369 |
+
"rewards/margins": 0.5203536152839661,
|
| 370 |
+
"rewards/rejected": -1.1768033504486084,
|
| 371 |
+
"step": 210
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 0.2302459445316588,
|
| 375 |
+
"grad_norm": 40.4541290817884,
|
| 376 |
+
"learning_rate": 4.7472967660421603e-07,
|
| 377 |
+
"logits/chosen": -1.0906522274017334,
|
| 378 |
+
"logits/rejected": -0.35459861159324646,
|
| 379 |
+
"logps/chosen": -383.7257080078125,
|
| 380 |
+
"logps/rejected": -398.504150390625,
|
| 381 |
+
"loss": 0.8394,
|
| 382 |
+
"rewards/accuracies": 0.731249988079071,
|
| 383 |
+
"rewards/chosen": -0.965796172618866,
|
| 384 |
+
"rewards/margins": 0.657662570476532,
|
| 385 |
+
"rewards/rejected": -1.6234586238861084,
|
| 386 |
+
"step": 220
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.24071166928309787,
|
| 390 |
+
"grad_norm": 35.38638904783933,
|
| 391 |
+
"learning_rate": 4.705745280752585e-07,
|
| 392 |
+
"logits/chosen": -1.0554046630859375,
|
| 393 |
+
"logits/rejected": -0.6607178449630737,
|
| 394 |
+
"logps/chosen": -392.6568603515625,
|
| 395 |
+
"logps/rejected": -424.40814208984375,
|
| 396 |
+
"loss": 0.8436,
|
| 397 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 398 |
+
"rewards/chosen": -1.0391901731491089,
|
| 399 |
+
"rewards/margins": 0.5193502902984619,
|
| 400 |
+
"rewards/rejected": -1.5585404634475708,
|
| 401 |
+
"step": 230
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.25117739403453687,
|
| 405 |
+
"grad_norm": 51.716672818453596,
|
| 406 |
+
"learning_rate": 4.6612438066572555e-07,
|
| 407 |
+
"logits/chosen": -0.8230894207954407,
|
| 408 |
+
"logits/rejected": -0.2124086320400238,
|
| 409 |
+
"logps/chosen": -360.98162841796875,
|
| 410 |
+
"logps/rejected": -425.2185974121094,
|
| 411 |
+
"loss": 0.8129,
|
| 412 |
+
"rewards/accuracies": 0.731249988079071,
|
| 413 |
+
"rewards/chosen": -0.9339190721511841,
|
| 414 |
+
"rewards/margins": 0.7763990163803101,
|
| 415 |
+
"rewards/rejected": -1.7103179693222046,
|
| 416 |
+
"step": 240
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.2616431187859759,
|
| 420 |
+
"grad_norm": 33.945957236730834,
|
| 421 |
+
"learning_rate": 4.6138518605333664e-07,
|
| 422 |
+
"logits/chosen": -0.6045997142791748,
|
| 423 |
+
"logits/rejected": -0.11657794564962387,
|
| 424 |
+
"logps/chosen": -377.04498291015625,
|
| 425 |
+
"logps/rejected": -422.51678466796875,
|
| 426 |
+
"loss": 0.7922,
|
| 427 |
+
"rewards/accuracies": 0.699999988079071,
|
| 428 |
+
"rewards/chosen": -0.9926928281784058,
|
| 429 |
+
"rewards/margins": 0.6563337445259094,
|
| 430 |
+
"rewards/rejected": -1.64902663230896,
|
| 431 |
+
"step": 250
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.272108843537415,
|
| 435 |
+
"grad_norm": 34.96189412822079,
|
| 436 |
+
"learning_rate": 4.5636328249082514e-07,
|
| 437 |
+
"logits/chosen": -0.0778447613120079,
|
| 438 |
+
"logits/rejected": 0.5611258149147034,
|
| 439 |
+
"logps/chosen": -424.87054443359375,
|
| 440 |
+
"logps/rejected": -477.55487060546875,
|
| 441 |
+
"loss": 0.8389,
|
| 442 |
+
"rewards/accuracies": 0.6875,
|
| 443 |
+
"rewards/chosen": -1.358944296836853,
|
| 444 |
+
"rewards/margins": 0.5935989618301392,
|
| 445 |
+
"rewards/rejected": -1.9525432586669922,
|
| 446 |
+
"step": 260
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.282574568288854,
|
| 450 |
+
"grad_norm": 31.898898777717502,
|
| 451 |
+
"learning_rate": 4.510653863290871e-07,
|
| 452 |
+
"logits/chosen": -1.0117652416229248,
|
| 453 |
+
"logits/rejected": -0.5077033042907715,
|
| 454 |
+
"logps/chosen": -378.0097961425781,
|
| 455 |
+
"logps/rejected": -400.56414794921875,
|
| 456 |
+
"loss": 0.798,
|
| 457 |
+
"rewards/accuracies": 0.75,
|
| 458 |
+
"rewards/chosen": -0.7833204865455627,
|
| 459 |
+
"rewards/margins": 0.6868009567260742,
|
| 460 |
+
"rewards/rejected": -1.4701216220855713,
|
| 461 |
+
"step": 270
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.29304029304029305,
|
| 465 |
+
"grad_norm": 32.916951386686485,
|
| 466 |
+
"learning_rate": 4.4549858303465737e-07,
|
| 467 |
+
"logits/chosen": -0.3384867310523987,
|
| 468 |
+
"logits/rejected": 0.2598015367984772,
|
| 469 |
+
"logps/chosen": -389.6136474609375,
|
| 470 |
+
"logps/rejected": -451.6962890625,
|
| 471 |
+
"loss": 0.846,
|
| 472 |
+
"rewards/accuracies": 0.762499988079071,
|
| 473 |
+
"rewards/chosen": -0.9335886240005493,
|
| 474 |
+
"rewards/margins": 0.7509840130805969,
|
| 475 |
+
"rewards/rejected": -1.6845725774765015,
|
| 476 |
+
"step": 280
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.3035060177917321,
|
| 480 |
+
"grad_norm": 37.72765713788291,
|
| 481 |
+
"learning_rate": 4.396703177135261e-07,
|
| 482 |
+
"logits/chosen": -0.23731651902198792,
|
| 483 |
+
"logits/rejected": 0.5401536822319031,
|
| 484 |
+
"logps/chosen": -377.9930114746094,
|
| 485 |
+
"logps/rejected": -415.17022705078125,
|
| 486 |
+
"loss": 0.8217,
|
| 487 |
+
"rewards/accuracies": 0.737500011920929,
|
| 488 |
+
"rewards/chosen": -1.052441120147705,
|
| 489 |
+
"rewards/margins": 0.7287407517433167,
|
| 490 |
+
"rewards/rejected": -1.7811816930770874,
|
| 491 |
+
"step": 290
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.3139717425431711,
|
| 495 |
+
"grad_norm": 32.913268476105344,
|
| 496 |
+
"learning_rate": 4.335883851539693e-07,
|
| 497 |
+
"logits/chosen": -0.5817835330963135,
|
| 498 |
+
"logits/rejected": -0.08701565116643906,
|
| 499 |
+
"logps/chosen": -362.9393615722656,
|
| 500 |
+
"logps/rejected": -424.712158203125,
|
| 501 |
+
"loss": 0.8208,
|
| 502 |
+
"rewards/accuracies": 0.699999988079071,
|
| 503 |
+
"rewards/chosen": -0.9393421411514282,
|
| 504 |
+
"rewards/margins": 0.6302340030670166,
|
| 505 |
+
"rewards/rejected": -1.5695761442184448,
|
| 506 |
+
"step": 300
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.3139717425431711,
|
| 510 |
+
"eval_logits/chosen": -0.4089333117008209,
|
| 511 |
+
"eval_logits/rejected": 0.30610814690589905,
|
| 512 |
+
"eval_logps/chosen": -370.631591796875,
|
| 513 |
+
"eval_logps/rejected": -428.09271240234375,
|
| 514 |
+
"eval_loss": 0.8214874267578125,
|
| 515 |
+
"eval_rewards/accuracies": 0.7638888955116272,
|
| 516 |
+
"eval_rewards/chosen": -0.8865460753440857,
|
| 517 |
+
"eval_rewards/margins": 0.792353093624115,
|
| 518 |
+
"eval_rewards/rejected": -1.6788992881774902,
|
| 519 |
+
"eval_runtime": 394.7768,
|
| 520 |
+
"eval_samples_per_second": 5.066,
|
| 521 |
+
"eval_steps_per_second": 0.16,
|
| 522 |
+
"step": 300
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.32443746729461015,
|
| 526 |
+
"grad_norm": 35.677999783537,
|
| 527 |
+
"learning_rate": 4.272609194017105e-07,
|
| 528 |
+
"logits/chosen": -0.5777407884597778,
|
| 529 |
+
"logits/rejected": 0.6635278463363647,
|
| 530 |
+
"logps/chosen": -369.7853088378906,
|
| 531 |
+
"logps/rejected": -404.04779052734375,
|
| 532 |
+
"loss": 0.7973,
|
| 533 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 534 |
+
"rewards/chosen": -0.8606454133987427,
|
| 535 |
+
"rewards/margins": 0.8526867628097534,
|
| 536 |
+
"rewards/rejected": -1.713331937789917,
|
| 537 |
+
"step": 310
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.3349031920460492,
|
| 541 |
+
"grad_norm": 32.27841917992221,
|
| 542 |
+
"learning_rate": 4.2069638288135547e-07,
|
| 543 |
+
"logits/chosen": 0.1319276988506317,
|
| 544 |
+
"logits/rejected": 0.9114134907722473,
|
| 545 |
+
"logps/chosen": -353.5772399902344,
|
| 546 |
+
"logps/rejected": -430.6465759277344,
|
| 547 |
+
"loss": 0.8394,
|
| 548 |
+
"rewards/accuracies": 0.71875,
|
| 549 |
+
"rewards/chosen": -1.0053209066390991,
|
| 550 |
+
"rewards/margins": 0.7951760292053223,
|
| 551 |
+
"rewards/rejected": -1.8004968166351318,
|
| 552 |
+
"step": 320
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 0.3453689167974882,
|
| 556 |
+
"grad_norm": 36.77279969745553,
|
| 557 |
+
"learning_rate": 4.139035550786494e-07,
|
| 558 |
+
"logits/chosen": 0.24769747257232666,
|
| 559 |
+
"logits/rejected": 0.9449674487113953,
|
| 560 |
+
"logps/chosen": -359.6546936035156,
|
| 561 |
+
"logps/rejected": -403.91668701171875,
|
| 562 |
+
"loss": 0.807,
|
| 563 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 564 |
+
"rewards/chosen": -1.0324417352676392,
|
| 565 |
+
"rewards/margins": 0.6857098937034607,
|
| 566 |
+
"rewards/rejected": -1.7181516885757446,
|
| 567 |
+
"step": 330
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.35583464154892724,
|
| 571 |
+
"grad_norm": 37.42709989098109,
|
| 572 |
+
"learning_rate": 4.0689152079869306e-07,
|
| 573 |
+
"logits/chosen": 0.4464758336544037,
|
| 574 |
+
"logits/rejected": 1.1670682430267334,
|
| 575 |
+
"logps/chosen": -367.1394348144531,
|
| 576 |
+
"logps/rejected": -414.1123046875,
|
| 577 |
+
"loss": 0.8191,
|
| 578 |
+
"rewards/accuracies": 0.706250011920929,
|
| 579 |
+
"rewards/chosen": -0.9761883020401001,
|
| 580 |
+
"rewards/margins": 0.8046461343765259,
|
| 581 |
+
"rewards/rejected": -1.7808345556259155,
|
| 582 |
+
"step": 340
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 0.3663003663003663,
|
| 586 |
+
"grad_norm": 36.82109712593396,
|
| 587 |
+
"learning_rate": 3.99669658015821e-07,
|
| 588 |
+
"logits/chosen": 0.28100523352622986,
|
| 589 |
+
"logits/rejected": 1.06730055809021,
|
| 590 |
+
"logps/chosen": -388.6217041015625,
|
| 591 |
+
"logps/rejected": -429.86822509765625,
|
| 592 |
+
"loss": 0.7899,
|
| 593 |
+
"rewards/accuracies": 0.737500011920929,
|
| 594 |
+
"rewards/chosen": -0.9743186831474304,
|
| 595 |
+
"rewards/margins": 0.7633944749832153,
|
| 596 |
+
"rewards/rejected": -1.7377132177352905,
|
| 597 |
+
"step": 350
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.37676609105180536,
|
| 601 |
+
"grad_norm": 45.084347178428885,
|
| 602 |
+
"learning_rate": 3.92247625331392e-07,
|
| 603 |
+
"logits/chosen": 0.001661926507949829,
|
| 604 |
+
"logits/rejected": 0.8640968203544617,
|
| 605 |
+
"logps/chosen": -382.6606750488281,
|
| 606 |
+
"logps/rejected": -439.5569763183594,
|
| 607 |
+
"loss": 0.7609,
|
| 608 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 609 |
+
"rewards/chosen": -0.967884361743927,
|
| 610 |
+
"rewards/margins": 0.8003349304199219,
|
| 611 |
+
"rewards/rejected": -1.7682193517684937,
|
| 612 |
+
"step": 360
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.3872318158032444,
|
| 616 |
+
"grad_norm": 47.00073205229117,
|
| 617 |
+
"learning_rate": 3.846353490562664e-07,
|
| 618 |
+
"logits/chosen": -0.05083969235420227,
|
| 619 |
+
"logits/rejected": 0.8022430539131165,
|
| 620 |
+
"logps/chosen": -356.86920166015625,
|
| 621 |
+
"logps/rejected": -404.42755126953125,
|
| 622 |
+
"loss": 0.8,
|
| 623 |
+
"rewards/accuracies": 0.731249988079071,
|
| 624 |
+
"rewards/chosen": -0.9355790019035339,
|
| 625 |
+
"rewards/margins": 0.7523115873336792,
|
| 626 |
+
"rewards/rejected": -1.687890648841858,
|
| 627 |
+
"step": 370
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.3976975405546834,
|
| 631 |
+
"grad_norm": 54.67041616108794,
|
| 632 |
+
"learning_rate": 3.768430099352445e-07,
|
| 633 |
+
"logits/chosen": 0.15414004027843475,
|
| 634 |
+
"logits/rejected": 1.1920832395553589,
|
| 635 |
+
"logps/chosen": -355.2337951660156,
|
| 636 |
+
"logps/rejected": -432.7115173339844,
|
| 637 |
+
"loss": 0.8138,
|
| 638 |
+
"rewards/accuracies": 0.762499988079071,
|
| 639 |
+
"rewards/chosen": -1.0056902170181274,
|
| 640 |
+
"rewards/margins": 0.8578664064407349,
|
| 641 |
+
"rewards/rejected": -1.8635566234588623,
|
| 642 |
+
"step": 380
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 0.40816326530612246,
|
| 646 |
+
"grad_norm": 36.59955698934101,
|
| 647 |
+
"learning_rate": 3.6888102953122304e-07,
|
| 648 |
+
"logits/chosen": -0.0811910480260849,
|
| 649 |
+
"logits/rejected": 0.8333920240402222,
|
| 650 |
+
"logps/chosen": -386.803955078125,
|
| 651 |
+
"logps/rejected": -451.043701171875,
|
| 652 |
+
"loss": 0.8125,
|
| 653 |
+
"rewards/accuracies": 0.737500011920929,
|
| 654 |
+
"rewards/chosen": -1.0521795749664307,
|
| 655 |
+
"rewards/margins": 0.7749984264373779,
|
| 656 |
+
"rewards/rejected": -1.8271780014038086,
|
| 657 |
+
"step": 390
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.4186289900575615,
|
| 661 |
+
"grad_norm": 44.12526971490869,
|
| 662 |
+
"learning_rate": 3.607600562872785e-07,
|
| 663 |
+
"logits/chosen": -0.16013869643211365,
|
| 664 |
+
"logits/rejected": 0.554747462272644,
|
| 665 |
+
"logps/chosen": -390.5782775878906,
|
| 666 |
+
"logps/rejected": -457.771484375,
|
| 667 |
+
"loss": 0.8208,
|
| 668 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 669 |
+
"rewards/chosen": -1.205999732017517,
|
| 670 |
+
"rewards/margins": 0.694991946220398,
|
| 671 |
+
"rewards/rejected": -1.9009917974472046,
|
| 672 |
+
"step": 400
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 0.4186289900575615,
|
| 676 |
+
"eval_logits/chosen": -0.42617201805114746,
|
| 677 |
+
"eval_logits/rejected": 0.5905421376228333,
|
| 678 |
+
"eval_logps/chosen": -401.0516052246094,
|
| 679 |
+
"eval_logps/rejected": -460.0636901855469,
|
| 680 |
+
"eval_loss": 0.7982370257377625,
|
| 681 |
+
"eval_rewards/accuracies": 0.77182537317276,
|
| 682 |
+
"eval_rewards/chosen": -1.1907460689544678,
|
| 683 |
+
"eval_rewards/margins": 0.8078626394271851,
|
| 684 |
+
"eval_rewards/rejected": -1.9986087083816528,
|
| 685 |
+
"eval_runtime": 407.5508,
|
| 686 |
+
"eval_samples_per_second": 4.907,
|
| 687 |
+
"eval_steps_per_second": 0.155,
|
| 688 |
+
"step": 400
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.4290947148090005,
|
| 692 |
+
"grad_norm": 27.932485684518436,
|
| 693 |
+
"learning_rate": 3.5249095128531856e-07,
|
| 694 |
+
"logits/chosen": 0.006985366344451904,
|
| 695 |
+
"logits/rejected": 0.9441334009170532,
|
| 696 |
+
"logps/chosen": -385.29931640625,
|
| 697 |
+
"logps/rejected": -432.65802001953125,
|
| 698 |
+
"loss": 0.8168,
|
| 699 |
+
"rewards/accuracies": 0.75,
|
| 700 |
+
"rewards/chosen": -1.135567307472229,
|
| 701 |
+
"rewards/margins": 0.7227233648300171,
|
| 702 |
+
"rewards/rejected": -1.858290672302246,
|
| 703 |
+
"step": 410
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.43956043956043955,
|
| 707 |
+
"grad_norm": 81.65756808013388,
|
| 708 |
+
"learning_rate": 3.4408477372034736e-07,
|
| 709 |
+
"logits/chosen": -0.2533782124519348,
|
| 710 |
+
"logits/rejected": 1.4089046716690063,
|
| 711 |
+
"logps/chosen": -373.7265625,
|
| 712 |
+
"logps/rejected": -432.09588623046875,
|
| 713 |
+
"loss": 0.8162,
|
| 714 |
+
"rewards/accuracies": 0.78125,
|
| 715 |
+
"rewards/chosen": -0.915477454662323,
|
| 716 |
+
"rewards/margins": 0.8855735659599304,
|
| 717 |
+
"rewards/rejected": -1.8010507822036743,
|
| 718 |
+
"step": 420
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 0.4500261643118786,
|
| 722 |
+
"grad_norm": 64.08330230011423,
|
| 723 |
+
"learning_rate": 3.3555276610977276e-07,
|
| 724 |
+
"logits/chosen": 0.9608128666877747,
|
| 725 |
+
"logits/rejected": 2.096973180770874,
|
| 726 |
+
"logps/chosen": -359.41241455078125,
|
| 727 |
+
"logps/rejected": -445.40631103515625,
|
| 728 |
+
"loss": 0.7851,
|
| 729 |
+
"rewards/accuracies": 0.75,
|
| 730 |
+
"rewards/chosen": -1.056495189666748,
|
| 731 |
+
"rewards/margins": 0.8955538868904114,
|
| 732 |
+
"rewards/rejected": -1.952048897743225,
|
| 733 |
+
"step": 430
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 0.4604918890633176,
|
| 737 |
+
"grad_norm": 45.75501310686222,
|
| 738 |
+
"learning_rate": 3.269063392575352e-07,
|
| 739 |
+
"logits/chosen": 0.31944066286087036,
|
| 740 |
+
"logits/rejected": 1.1516392230987549,
|
| 741 |
+
"logps/chosen": -387.1063537597656,
|
| 742 |
+
"logps/rejected": -415.3641052246094,
|
| 743 |
+
"loss": 0.8046,
|
| 744 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 745 |
+
"rewards/chosen": -0.9187418818473816,
|
| 746 |
+
"rewards/margins": 0.680588960647583,
|
| 747 |
+
"rewards/rejected": -1.5993306636810303,
|
| 748 |
+
"step": 440
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 0.47095761381475665,
|
| 752 |
+
"grad_norm": 40.092859728853355,
|
| 753 |
+
"learning_rate": 3.1815705699316964e-07,
|
| 754 |
+
"logits/chosen": -0.16375192999839783,
|
| 755 |
+
"logits/rejected": 0.5893057584762573,
|
| 756 |
+
"logps/chosen": -385.47412109375,
|
| 757 |
+
"logps/rejected": -477.24847412109375,
|
| 758 |
+
"loss": 0.797,
|
| 759 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 760 |
+
"rewards/chosen": -1.030901551246643,
|
| 761 |
+
"rewards/margins": 0.9582377672195435,
|
| 762 |
+
"rewards/rejected": -1.989139199256897,
|
| 763 |
+
"step": 450
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 0.48142333856619574,
|
| 767 |
+
"grad_norm": 41.71685104223482,
|
| 768 |
+
"learning_rate": 3.0931662070620794e-07,
|
| 769 |
+
"logits/chosen": 0.22953590750694275,
|
| 770 |
+
"logits/rejected": 1.0651460886001587,
|
| 771 |
+
"logps/chosen": -399.14300537109375,
|
| 772 |
+
"logps/rejected": -465.94744873046875,
|
| 773 |
+
"loss": 0.8182,
|
| 774 |
+
"rewards/accuracies": 0.75,
|
| 775 |
+
"rewards/chosen": -1.4160016775131226,
|
| 776 |
+
"rewards/margins": 0.7932834029197693,
|
| 777 |
+
"rewards/rejected": -2.209285259246826,
|
| 778 |
+
"step": 460
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 0.49188906331763477,
|
| 782 |
+
"grad_norm": 45.00081885012515,
|
| 783 |
+
"learning_rate": 3.003968536966078e-07,
|
| 784 |
+
"logits/chosen": 0.9409812092781067,
|
| 785 |
+
"logits/rejected": 1.9240789413452148,
|
| 786 |
+
"logps/chosen": -418.9644470214844,
|
| 787 |
+
"logps/rejected": -467.4033203125,
|
| 788 |
+
"loss": 0.7631,
|
| 789 |
+
"rewards/accuracies": 0.731249988079071,
|
| 790 |
+
"rewards/chosen": -1.3515257835388184,
|
| 791 |
+
"rewards/margins": 0.8228591084480286,
|
| 792 |
+
"rewards/rejected": -2.1743850708007812,
|
| 793 |
+
"step": 470
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.5023547880690737,
|
| 797 |
+
"grad_norm": 51.675134784509765,
|
| 798 |
+
"learning_rate": 2.9140968536213693e-07,
|
| 799 |
+
"logits/chosen": 1.075547456741333,
|
| 800 |
+
"logits/rejected": 2.493340015411377,
|
| 801 |
+
"logps/chosen": -390.1001892089844,
|
| 802 |
+
"logps/rejected": -482.7947692871094,
|
| 803 |
+
"loss": 0.7996,
|
| 804 |
+
"rewards/accuracies": 0.768750011920929,
|
| 805 |
+
"rewards/chosen": -1.228891134262085,
|
| 806 |
+
"rewards/margins": 1.0860393047332764,
|
| 807 |
+
"rewards/rejected": -2.3149304389953613,
|
| 808 |
+
"step": 480
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.5128205128205128,
|
| 812 |
+
"grad_norm": 36.865587472816905,
|
| 813 |
+
"learning_rate": 2.823671352438608e-07,
|
| 814 |
+
"logits/chosen": -0.14976122975349426,
|
| 815 |
+
"logits/rejected": 1.63918137550354,
|
| 816 |
+
"logps/chosen": -416.5582580566406,
|
| 817 |
+
"logps/rejected": -461.94415283203125,
|
| 818 |
+
"loss": 0.7795,
|
| 819 |
+
"rewards/accuracies": 0.8374999761581421,
|
| 820 |
+
"rewards/chosen": -1.046267032623291,
|
| 821 |
+
"rewards/margins": 1.0088865756988525,
|
| 822 |
+
"rewards/rejected": -2.0551538467407227,
|
| 823 |
+
"step": 490
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 0.5232862375719518,
|
| 827 |
+
"grad_norm": 41.30030890040398,
|
| 828 |
+
"learning_rate": 2.73281296951072e-07,
|
| 829 |
+
"logits/chosen": 1.0807673931121826,
|
| 830 |
+
"logits/rejected": 1.9634937047958374,
|
| 831 |
+
"logps/chosen": -394.7601318359375,
|
| 832 |
+
"logps/rejected": -450.50628662109375,
|
| 833 |
+
"loss": 0.7826,
|
| 834 |
+
"rewards/accuracies": 0.768750011920929,
|
| 835 |
+
"rewards/chosen": -1.3611440658569336,
|
| 836 |
+
"rewards/margins": 0.870000958442688,
|
| 837 |
+
"rewards/rejected": -2.231145143508911,
|
| 838 |
+
"step": 500
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 0.5232862375719518,
|
| 842 |
+
"eval_logits/chosen": 1.0156385898590088,
|
| 843 |
+
"eval_logits/rejected": 2.233916997909546,
|
| 844 |
+
"eval_logps/chosen": -421.72698974609375,
|
| 845 |
+
"eval_logps/rejected": -504.034912109375,
|
| 846 |
+
"eval_loss": 0.7799234390258789,
|
| 847 |
+
"eval_rewards/accuracies": 0.7757936716079712,
|
| 848 |
+
"eval_rewards/chosen": -1.397499680519104,
|
| 849 |
+
"eval_rewards/margins": 1.0408216714859009,
|
| 850 |
+
"eval_rewards/rejected": -2.438321590423584,
|
| 851 |
+
"eval_runtime": 404.3457,
|
| 852 |
+
"eval_samples_per_second": 4.946,
|
| 853 |
+
"eval_steps_per_second": 0.156,
|
| 854 |
+
"step": 500
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 0.533751962323391,
|
| 858 |
+
"grad_norm": 54.334410731861475,
|
| 859 |
+
"learning_rate": 2.641643219871597e-07,
|
| 860 |
+
"logits/chosen": 1.0641226768493652,
|
| 861 |
+
"logits/rejected": 2.259822130203247,
|
| 862 |
+
"logps/chosen": -399.194580078125,
|
| 863 |
+
"logps/rejected": -516.8468017578125,
|
| 864 |
+
"loss": 0.7452,
|
| 865 |
+
"rewards/accuracies": 0.793749988079071,
|
| 866 |
+
"rewards/chosen": -1.1524865627288818,
|
| 867 |
+
"rewards/margins": 1.3779429197311401,
|
| 868 |
+
"rewards/rejected": -2.5304293632507324,
|
| 869 |
+
"step": 510
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 0.54421768707483,
|
| 873 |
+
"grad_norm": 88.58051947279036,
|
| 874 |
+
"learning_rate": 2.550284034980507e-07,
|
| 875 |
+
"logits/chosen": 2.1809422969818115,
|
| 876 |
+
"logits/rejected": 3.113351345062256,
|
| 877 |
+
"logps/chosen": -471.63287353515625,
|
| 878 |
+
"logps/rejected": -541.3626708984375,
|
| 879 |
+
"loss": 0.8148,
|
| 880 |
+
"rewards/accuracies": 0.706250011920929,
|
| 881 |
+
"rewards/chosen": -1.9956514835357666,
|
| 882 |
+
"rewards/margins": 0.9180843234062195,
|
| 883 |
+
"rewards/rejected": -2.913735866546631,
|
| 884 |
+
"step": 520
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.554683411826269,
|
| 888 |
+
"grad_norm": 39.757328239690565,
|
| 889 |
+
"learning_rate": 2.4588575996495794e-07,
|
| 890 |
+
"logits/chosen": 0.6403165459632874,
|
| 891 |
+
"logits/rejected": 1.7489306926727295,
|
| 892 |
+
"logps/chosen": -415.21319580078125,
|
| 893 |
+
"logps/rejected": -499.917236328125,
|
| 894 |
+
"loss": 0.7868,
|
| 895 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 896 |
+
"rewards/chosen": -1.4644101858139038,
|
| 897 |
+
"rewards/margins": 0.9843127131462097,
|
| 898 |
+
"rewards/rejected": -2.4487228393554688,
|
| 899 |
+
"step": 530
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.565149136577708,
|
| 903 |
+
"grad_norm": 40.25852990858294,
|
| 904 |
+
"learning_rate": 2.367486188632446e-07,
|
| 905 |
+
"logits/chosen": -0.26262250542640686,
|
| 906 |
+
"logits/rejected": 0.8144134283065796,
|
| 907 |
+
"logps/chosen": -404.18792724609375,
|
| 908 |
+
"logps/rejected": -461.5171813964844,
|
| 909 |
+
"loss": 0.8012,
|
| 910 |
+
"rewards/accuracies": 0.737500011920929,
|
| 911 |
+
"rewards/chosen": -1.1386909484863281,
|
| 912 |
+
"rewards/margins": 0.8705164194107056,
|
| 913 |
+
"rewards/rejected": -2.009207248687744,
|
| 914 |
+
"step": 540
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 0.5756148613291471,
|
| 918 |
+
"grad_norm": 39.67017653274985,
|
| 919 |
+
"learning_rate": 2.276292003092593e-07,
|
| 920 |
+
"logits/chosen": 0.2649956941604614,
|
| 921 |
+
"logits/rejected": 1.6869157552719116,
|
| 922 |
+
"logps/chosen": -412.3680114746094,
|
| 923 |
+
"logps/rejected": -480.45892333984375,
|
| 924 |
+
"loss": 0.8035,
|
| 925 |
+
"rewards/accuracies": 0.768750011920929,
|
| 926 |
+
"rewards/chosen": -1.410407543182373,
|
| 927 |
+
"rewards/margins": 1.1013845205307007,
|
| 928 |
+
"rewards/rejected": -2.511791944503784,
|
| 929 |
+
"step": 550
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 0.5860805860805861,
|
| 933 |
+
"grad_norm": 48.8987231552702,
|
| 934 |
+
"learning_rate": 2.185397007170141e-07,
|
| 935 |
+
"logits/chosen": 0.45904502272605896,
|
| 936 |
+
"logits/rejected": 2.004812717437744,
|
| 937 |
+
"logps/chosen": -432.9220275878906,
|
| 938 |
+
"logps/rejected": -502.25054931640625,
|
| 939 |
+
"loss": 0.8156,
|
| 940 |
+
"rewards/accuracies": 0.75,
|
| 941 |
+
"rewards/chosen": -1.5010192394256592,
|
| 942 |
+
"rewards/margins": 0.9719829559326172,
|
| 943 |
+
"rewards/rejected": -2.4730021953582764,
|
| 944 |
+
"step": 560
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 0.5965463108320251,
|
| 948 |
+
"grad_norm": 46.52679933808478,
|
| 949 |
+
"learning_rate": 2.094922764865619e-07,
|
| 950 |
+
"logits/chosen": 0.06010497733950615,
|
| 951 |
+
"logits/rejected": 1.337308645248413,
|
| 952 |
+
"logps/chosen": -434.7491760253906,
|
| 953 |
+
"logps/rejected": -508.05413818359375,
|
| 954 |
+
"loss": 0.7681,
|
| 955 |
+
"rewards/accuracies": 0.737500011920929,
|
| 956 |
+
"rewards/chosen": -1.5524450540542603,
|
| 957 |
+
"rewards/margins": 0.8663631677627563,
|
| 958 |
+
"rewards/rejected": -2.4188084602355957,
|
| 959 |
+
"step": 570
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 0.6070120355834642,
|
| 963 |
+
"grad_norm": 49.619200203200066,
|
| 964 |
+
"learning_rate": 2.0049902774588797e-07,
|
| 965 |
+
"logits/chosen": 0.18441572785377502,
|
| 966 |
+
"logits/rejected": 1.5482475757598877,
|
| 967 |
+
"logps/chosen": -395.56878662109375,
|
| 968 |
+
"logps/rejected": -450.8467712402344,
|
| 969 |
+
"loss": 0.7832,
|
| 970 |
+
"rewards/accuracies": 0.78125,
|
| 971 |
+
"rewards/chosen": -1.4222776889801025,
|
| 972 |
+
"rewards/margins": 0.9578974843025208,
|
| 973 |
+
"rewards/rejected": -2.3801751136779785,
|
| 974 |
+
"step": 580
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 0.6174777603349032,
|
| 978 |
+
"grad_norm": 38.27541061990721,
|
| 979 |
+
"learning_rate": 1.9157198216806238e-07,
|
| 980 |
+
"logits/chosen": 0.08942364156246185,
|
| 981 |
+
"logits/rejected": 1.5633885860443115,
|
| 982 |
+
"logps/chosen": -431.2159118652344,
|
| 983 |
+
"logps/rejected": -519.74560546875,
|
| 984 |
+
"loss": 0.7602,
|
| 985 |
+
"rewards/accuracies": 0.699999988079071,
|
| 986 |
+
"rewards/chosen": -1.5320241451263428,
|
| 987 |
+
"rewards/margins": 0.9783884882926941,
|
| 988 |
+
"rewards/rejected": -2.5104126930236816,
|
| 989 |
+
"step": 590
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.6279434850863422,
|
| 993 |
+
"grad_norm": 43.240477495824976,
|
| 994 |
+
"learning_rate": 1.8272307888529274e-07,
|
| 995 |
+
"logits/chosen": 0.5950905084609985,
|
| 996 |
+
"logits/rejected": 1.837579369544983,
|
| 997 |
+
"logps/chosen": -435.34698486328125,
|
| 998 |
+
"logps/rejected": -513.4215698242188,
|
| 999 |
+
"loss": 0.7546,
|
| 1000 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1001 |
+
"rewards/chosen": -1.5277029275894165,
|
| 1002 |
+
"rewards/margins": 1.0060127973556519,
|
| 1003 |
+
"rewards/rejected": -2.5337159633636475,
|
| 1004 |
+
"step": 600
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.6279434850863422,
|
| 1008 |
+
"eval_logits/chosen": 0.3289608955383301,
|
| 1009 |
+
"eval_logits/rejected": 1.6797810792922974,
|
| 1010 |
+
"eval_logps/chosen": -437.6458740234375,
|
| 1011 |
+
"eval_logps/rejected": -526.840576171875,
|
| 1012 |
+
"eval_loss": 0.7723253965377808,
|
| 1013 |
+
"eval_rewards/accuracies": 0.783730149269104,
|
| 1014 |
+
"eval_rewards/chosen": -1.5566877126693726,
|
| 1015 |
+
"eval_rewards/margins": 1.1096901893615723,
|
| 1016 |
+
"eval_rewards/rejected": -2.6663780212402344,
|
| 1017 |
+
"eval_runtime": 437.2513,
|
| 1018 |
+
"eval_samples_per_second": 4.574,
|
| 1019 |
+
"eval_steps_per_second": 0.144,
|
| 1020 |
+
"step": 600
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 0.6384092098377813,
|
| 1024 |
+
"grad_norm": 52.91138230105047,
|
| 1025 |
+
"learning_rate": 1.7396415252139288e-07,
|
| 1026 |
+
"logits/chosen": 0.2616492509841919,
|
| 1027 |
+
"logits/rejected": 2.0162370204925537,
|
| 1028 |
+
"logps/chosen": -463.6256408691406,
|
| 1029 |
+
"logps/rejected": -519.8214111328125,
|
| 1030 |
+
"loss": 0.7861,
|
| 1031 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1032 |
+
"rewards/chosen": -1.580159306526184,
|
| 1033 |
+
"rewards/margins": 1.1713967323303223,
|
| 1034 |
+
"rewards/rejected": -2.751555919647217,
|
| 1035 |
+
"step": 610
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 0.6488749345892203,
|
| 1039 |
+
"grad_norm": 48.31537624152789,
|
| 1040 |
+
"learning_rate": 1.6530691736402316e-07,
|
| 1041 |
+
"logits/chosen": 0.6659356951713562,
|
| 1042 |
+
"logits/rejected": 1.7698688507080078,
|
| 1043 |
+
"logps/chosen": -429.9051818847656,
|
| 1044 |
+
"logps/rejected": -497.07025146484375,
|
| 1045 |
+
"loss": 0.7441,
|
| 1046 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1047 |
+
"rewards/chosen": -1.5888570547103882,
|
| 1048 |
+
"rewards/margins": 1.0183988809585571,
|
| 1049 |
+
"rewards/rejected": -2.607255697250366,
|
| 1050 |
+
"step": 620
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 0.6593406593406593,
|
| 1054 |
+
"grad_norm": 236.736070461938,
|
| 1055 |
+
"learning_rate": 1.5676295169786864e-07,
|
| 1056 |
+
"logits/chosen": 0.7470348477363586,
|
| 1057 |
+
"logits/rejected": 2.425363302230835,
|
| 1058 |
+
"logps/chosen": -437.38818359375,
|
| 1059 |
+
"logps/rejected": -515.9216918945312,
|
| 1060 |
+
"loss": 0.7726,
|
| 1061 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1062 |
+
"rewards/chosen": -1.6273107528686523,
|
| 1063 |
+
"rewards/margins": 1.1723902225494385,
|
| 1064 |
+
"rewards/rejected": -2.799700975418091,
|
| 1065 |
+
"step": 630
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 0.6698063840920984,
|
| 1069 |
+
"grad_norm": 52.12999592819006,
|
| 1070 |
+
"learning_rate": 1.483436823197092e-07,
|
| 1071 |
+
"logits/chosen": 1.6778907775878906,
|
| 1072 |
+
"logits/rejected": 3.183229446411133,
|
| 1073 |
+
"logps/chosen": -463.07952880859375,
|
| 1074 |
+
"logps/rejected": -530.0252685546875,
|
| 1075 |
+
"loss": 0.7562,
|
| 1076 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1077 |
+
"rewards/chosen": -1.960402488708496,
|
| 1078 |
+
"rewards/margins": 1.0529009103775024,
|
| 1079 |
+
"rewards/rejected": -3.013303279876709,
|
| 1080 |
+
"step": 640
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.6802721088435374,
|
| 1084 |
+
"grad_norm": 41.836074119152094,
|
| 1085 |
+
"learning_rate": 1.4006036925609243e-07,
|
| 1086 |
+
"logits/chosen": 0.887537956237793,
|
| 1087 |
+
"logits/rejected": 2.033219337463379,
|
| 1088 |
+
"logps/chosen": -427.80706787109375,
|
| 1089 |
+
"logps/rejected": -527.1192626953125,
|
| 1090 |
+
"loss": 0.7609,
|
| 1091 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1092 |
+
"rewards/chosen": -1.601722002029419,
|
| 1093 |
+
"rewards/margins": 0.9513872861862183,
|
| 1094 |
+
"rewards/rejected": -2.5531094074249268,
|
| 1095 |
+
"step": 650
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.6907378335949764,
|
| 1099 |
+
"grad_norm": 54.26043387708765,
|
| 1100 |
+
"learning_rate": 1.319240907040458e-07,
|
| 1101 |
+
"logits/chosen": 0.6521838903427124,
|
| 1102 |
+
"logits/rejected": 2.2099432945251465,
|
| 1103 |
+
"logps/chosen": -436.31256103515625,
|
| 1104 |
+
"logps/rejected": -529.8990478515625,
|
| 1105 |
+
"loss": 0.7569,
|
| 1106 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1107 |
+
"rewards/chosen": -1.5409561395645142,
|
| 1108 |
+
"rewards/margins": 1.183617353439331,
|
| 1109 |
+
"rewards/rejected": -2.7245736122131348,
|
| 1110 |
+
"step": 660
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 0.7012035583464155,
|
| 1114 |
+
"grad_norm": 51.79386928760391,
|
| 1115 |
+
"learning_rate": 1.239457282149695e-07,
|
| 1116 |
+
"logits/chosen": 0.2191120684146881,
|
| 1117 |
+
"logits/rejected": 1.4961636066436768,
|
| 1118 |
+
"logps/chosen": -430.20184326171875,
|
| 1119 |
+
"logps/rejected": -532.6630859375,
|
| 1120 |
+
"loss": 0.7425,
|
| 1121 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 1122 |
+
"rewards/chosen": -1.416535496711731,
|
| 1123 |
+
"rewards/margins": 1.0790010690689087,
|
| 1124 |
+
"rewards/rejected": -2.4955363273620605,
|
| 1125 |
+
"step": 670
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 0.7116692830978545,
|
| 1129 |
+
"grad_norm": 39.78182707020652,
|
| 1130 |
+
"learning_rate": 1.1613595214152711e-07,
|
| 1131 |
+
"logits/chosen": 0.693936288356781,
|
| 1132 |
+
"logits/rejected": 1.9294044971466064,
|
| 1133 |
+
"logps/chosen": -457.75494384765625,
|
| 1134 |
+
"logps/rejected": -548.4839477539062,
|
| 1135 |
+
"loss": 0.7871,
|
| 1136 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1137 |
+
"rewards/chosen": -1.460933804512024,
|
| 1138 |
+
"rewards/margins": 1.144832968711853,
|
| 1139 |
+
"rewards/rejected": -2.605767011642456,
|
| 1140 |
+
"step": 680
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 0.7221350078492935,
|
| 1144 |
+
"grad_norm": 39.2241916837222,
|
| 1145 |
+
"learning_rate": 1.0850520736699362e-07,
|
| 1146 |
+
"logits/chosen": 0.6683017015457153,
|
| 1147 |
+
"logits/rejected": 2.4072530269622803,
|
| 1148 |
+
"logps/chosen": -429.7069396972656,
|
| 1149 |
+
"logps/rejected": -508.4949645996094,
|
| 1150 |
+
"loss": 0.7309,
|
| 1151 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 1152 |
+
"rewards/chosen": -1.590958833694458,
|
| 1153 |
+
"rewards/margins": 1.1668089628219604,
|
| 1154 |
+
"rewards/rejected": -2.757768154144287,
|
| 1155 |
+
"step": 690
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 0.7326007326007326,
|
| 1159 |
+
"grad_norm": 48.29061638514223,
|
| 1160 |
+
"learning_rate": 1.0106369933615042e-07,
|
| 1161 |
+
"logits/chosen": 0.9930335283279419,
|
| 1162 |
+
"logits/rejected": 2.0346710681915283,
|
| 1163 |
+
"logps/chosen": -419.7799377441406,
|
| 1164 |
+
"logps/rejected": -497.60980224609375,
|
| 1165 |
+
"loss": 0.7533,
|
| 1166 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 1167 |
+
"rewards/chosen": -1.6569970846176147,
|
| 1168 |
+
"rewards/margins": 0.933269202709198,
|
| 1169 |
+
"rewards/rejected": -2.590266466140747,
|
| 1170 |
+
"step": 700
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 0.7326007326007326,
|
| 1174 |
+
"eval_logits/chosen": 0.698245108127594,
|
| 1175 |
+
"eval_logits/rejected": 2.0190064907073975,
|
| 1176 |
+
"eval_logps/chosen": -444.4420166015625,
|
| 1177 |
+
"eval_logps/rejected": -531.2305908203125,
|
| 1178 |
+
"eval_loss": 0.7731993198394775,
|
| 1179 |
+
"eval_rewards/accuracies": 0.783730149269104,
|
| 1180 |
+
"eval_rewards/chosen": -1.624650239944458,
|
| 1181 |
+
"eval_rewards/margins": 1.0856273174285889,
|
| 1182 |
+
"eval_rewards/rejected": -2.710277795791626,
|
| 1183 |
+
"eval_runtime": 434.4645,
|
| 1184 |
+
"eval_samples_per_second": 4.603,
|
| 1185 |
+
"eval_steps_per_second": 0.145,
|
| 1186 |
+
"step": 700
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.7430664573521716,
|
| 1190 |
+
"grad_norm": 53.90683395989049,
|
| 1191 |
+
"learning_rate": 9.382138040640714e-08,
|
| 1192 |
+
"logits/chosen": 0.8914741277694702,
|
| 1193 |
+
"logits/rejected": 2.3093373775482178,
|
| 1194 |
+
"logps/chosen": -432.30670166015625,
|
| 1195 |
+
"logps/rejected": -514.6910400390625,
|
| 1196 |
+
"loss": 0.8039,
|
| 1197 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1198 |
+
"rewards/chosen": -1.7386293411254883,
|
| 1199 |
+
"rewards/margins": 0.9785627126693726,
|
| 1200 |
+
"rewards/rejected": -2.7171919345855713,
|
| 1201 |
+
"step": 710
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 0.7535321821036107,
|
| 1205 |
+
"grad_norm": 42.68152796542752,
|
| 1206 |
+
"learning_rate": 8.678793653740632e-08,
|
| 1207 |
+
"logits/chosen": 0.6328099370002747,
|
| 1208 |
+
"logits/rejected": 1.9483941793441772,
|
| 1209 |
+
"logps/chosen": -434.7218322753906,
|
| 1210 |
+
"logps/rejected": -515.1497192382812,
|
| 1211 |
+
"loss": 0.7468,
|
| 1212 |
+
"rewards/accuracies": 0.75,
|
| 1213 |
+
"rewards/chosen": -1.5094399452209473,
|
| 1214 |
+
"rewards/margins": 1.1674511432647705,
|
| 1215 |
+
"rewards/rejected": -2.6768908500671387,
|
| 1216 |
+
"step": 720
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 0.7639979068550498,
|
| 1220 |
+
"grad_norm": 127.83230371998283,
|
| 1221 |
+
"learning_rate": 7.997277433690983e-08,
|
| 1222 |
+
"logits/chosen": 0.7204136848449707,
|
| 1223 |
+
"logits/rejected": 1.8154442310333252,
|
| 1224 |
+
"logps/chosen": -442.3927307128906,
|
| 1225 |
+
"logps/rejected": -500.26715087890625,
|
| 1226 |
+
"loss": 0.7779,
|
| 1227 |
+
"rewards/accuracies": 0.75,
|
| 1228 |
+
"rewards/chosen": -1.7704966068267822,
|
| 1229 |
+
"rewards/margins": 0.8622214198112488,
|
| 1230 |
+
"rewards/rejected": -2.6327178478240967,
|
| 1231 |
+
"step": 730
|
| 1232 |
+
},
|
| 1233 |
+
{
|
| 1234 |
+
"epoch": 0.7744636316064888,
|
| 1235 |
+
"grad_norm": 47.565336938926386,
|
| 1236 |
+
"learning_rate": 7.338500848029602e-08,
|
| 1237 |
+
"logits/chosen": 0.9852989912033081,
|
| 1238 |
+
"logits/rejected": 2.030043601989746,
|
| 1239 |
+
"logps/chosen": -424.19757080078125,
|
| 1240 |
+
"logps/rejected": -495.47698974609375,
|
| 1241 |
+
"loss": 0.7867,
|
| 1242 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1243 |
+
"rewards/chosen": -1.7129170894622803,
|
| 1244 |
+
"rewards/margins": 0.9195534586906433,
|
| 1245 |
+
"rewards/rejected": -2.6324708461761475,
|
| 1246 |
+
"step": 740
|
| 1247 |
+
},
|
| 1248 |
+
{
|
| 1249 |
+
"epoch": 0.7849293563579278,
|
| 1250 |
+
"grad_norm": 44.5701608685198,
|
| 1251 |
+
"learning_rate": 6.70334495204884e-08,
|
| 1252 |
+
"logits/chosen": 0.9759708642959595,
|
| 1253 |
+
"logits/rejected": 2.1048731803894043,
|
| 1254 |
+
"logps/chosen": -432.25665283203125,
|
| 1255 |
+
"logps/rejected": -540.9370727539062,
|
| 1256 |
+
"loss": 0.7385,
|
| 1257 |
+
"rewards/accuracies": 0.71875,
|
| 1258 |
+
"rewards/chosen": -1.8538835048675537,
|
| 1259 |
+
"rewards/margins": 1.118869662284851,
|
| 1260 |
+
"rewards/rejected": -2.9727535247802734,
|
| 1261 |
+
"step": 750
|
| 1262 |
+
},
|
| 1263 |
+
{
|
| 1264 |
+
"epoch": 0.7953950811093669,
|
| 1265 |
+
"grad_norm": 41.44381898960517,
|
| 1266 |
+
"learning_rate": 6.092659210462231e-08,
|
| 1267 |
+
"logits/chosen": 1.2645413875579834,
|
| 1268 |
+
"logits/rejected": 2.389688014984131,
|
| 1269 |
+
"logps/chosen": -435.0619201660156,
|
| 1270 |
+
"logps/rejected": -540.019287109375,
|
| 1271 |
+
"loss": 0.7955,
|
| 1272 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 1273 |
+
"rewards/chosen": -1.9370710849761963,
|
| 1274 |
+
"rewards/margins": 1.0329408645629883,
|
| 1275 |
+
"rewards/rejected": -2.9700117111206055,
|
| 1276 |
+
"step": 760
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 0.8058608058608059,
|
| 1280 |
+
"grad_norm": 30.769792434058846,
|
| 1281 |
+
"learning_rate": 5.507260361320737e-08,
|
| 1282 |
+
"logits/chosen": 0.6200178861618042,
|
| 1283 |
+
"logits/rejected": 1.4624145030975342,
|
| 1284 |
+
"logps/chosen": -481.1289978027344,
|
| 1285 |
+
"logps/rejected": -598.737060546875,
|
| 1286 |
+
"loss": 0.7364,
|
| 1287 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1288 |
+
"rewards/chosen": -1.7775561809539795,
|
| 1289 |
+
"rewards/margins": 0.9905630350112915,
|
| 1290 |
+
"rewards/rejected": -2.7681193351745605,
|
| 1291 |
+
"step": 770
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 0.8163265306122449,
|
| 1295 |
+
"grad_norm": 41.661722006018714,
|
| 1296 |
+
"learning_rate": 4.947931323697982e-08,
|
| 1297 |
+
"logits/chosen": 0.542585015296936,
|
| 1298 |
+
"logits/rejected": 1.6623198986053467,
|
| 1299 |
+
"logps/chosen": -482.098388671875,
|
| 1300 |
+
"logps/rejected": -534.7088623046875,
|
| 1301 |
+
"loss": 0.7831,
|
| 1302 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1303 |
+
"rewards/chosen": -1.63140869140625,
|
| 1304 |
+
"rewards/margins": 1.0396373271942139,
|
| 1305 |
+
"rewards/rejected": -2.671046257019043,
|
| 1306 |
+
"step": 780
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 0.826792255363684,
|
| 1310 |
+
"grad_norm": 58.38987202011464,
|
| 1311 |
+
"learning_rate": 4.415420150605398e-08,
|
| 1312 |
+
"logits/chosen": 0.6806478500366211,
|
| 1313 |
+
"logits/rejected": 2.5227348804473877,
|
| 1314 |
+
"logps/chosen": -446.99407958984375,
|
| 1315 |
+
"logps/rejected": -531.7085571289062,
|
| 1316 |
+
"loss": 0.8047,
|
| 1317 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 1318 |
+
"rewards/chosen": -1.6840887069702148,
|
| 1319 |
+
"rewards/margins": 1.1803264617919922,
|
| 1320 |
+
"rewards/rejected": -2.864415407180786,
|
| 1321 |
+
"step": 790
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"epoch": 0.837257980115123,
|
| 1325 |
+
"grad_norm": 54.92650887953424,
|
| 1326 |
+
"learning_rate": 3.9104390285376374e-08,
|
| 1327 |
+
"logits/chosen": 0.9064415097236633,
|
| 1328 |
+
"logits/rejected": 2.0099093914031982,
|
| 1329 |
+
"logps/chosen": -438.2958068847656,
|
| 1330 |
+
"logps/rejected": -536.7066650390625,
|
| 1331 |
+
"loss": 0.7498,
|
| 1332 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1333 |
+
"rewards/chosen": -1.7221300601959229,
|
| 1334 |
+
"rewards/margins": 1.0116078853607178,
|
| 1335 |
+
"rewards/rejected": -2.7337381839752197,
|
| 1336 |
+
"step": 800
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 0.837257980115123,
|
| 1340 |
+
"eval_logits/chosen": 0.42459264397621155,
|
| 1341 |
+
"eval_logits/rejected": 1.701023817062378,
|
| 1342 |
+
"eval_logps/chosen": -437.61517333984375,
|
| 1343 |
+
"eval_logps/rejected": -523.4053344726562,
|
| 1344 |
+
"eval_loss": 0.770976722240448,
|
| 1345 |
+
"eval_rewards/accuracies": 0.7857142686843872,
|
| 1346 |
+
"eval_rewards/chosen": -1.5563815832138062,
|
| 1347 |
+
"eval_rewards/margins": 1.07564377784729,
|
| 1348 |
+
"eval_rewards/rejected": -2.6320252418518066,
|
| 1349 |
+
"eval_runtime": 433.8899,
|
| 1350 |
+
"eval_samples_per_second": 4.609,
|
| 1351 |
+
"eval_steps_per_second": 0.145,
|
| 1352 |
+
"step": 800
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 0.847723704866562,
|
| 1356 |
+
"grad_norm": 43.87215946696443,
|
| 1357 |
+
"learning_rate": 3.433663324986208e-08,
|
| 1358 |
+
"logits/chosen": 0.48925551772117615,
|
| 1359 |
+
"logits/rejected": 1.9570659399032593,
|
| 1360 |
+
"logps/chosen": -447.86956787109375,
|
| 1361 |
+
"logps/rejected": -505.90460205078125,
|
| 1362 |
+
"loss": 0.7762,
|
| 1363 |
+
"rewards/accuracies": 0.78125,
|
| 1364 |
+
"rewards/chosen": -1.5664446353912354,
|
| 1365 |
+
"rewards/margins": 1.0283935070037842,
|
| 1366 |
+
"rewards/rejected": -2.5948386192321777,
|
| 1367 |
+
"step": 810
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 0.858189429618001,
|
| 1371 |
+
"grad_norm": 51.45847719694428,
|
| 1372 |
+
"learning_rate": 2.9857306851953897e-08,
|
| 1373 |
+
"logits/chosen": 0.5297014117240906,
|
| 1374 |
+
"logits/rejected": 1.7246748208999634,
|
| 1375 |
+
"logps/chosen": -460.77532958984375,
|
| 1376 |
+
"logps/rejected": -539.4356689453125,
|
| 1377 |
+
"loss": 0.7816,
|
| 1378 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 1379 |
+
"rewards/chosen": -1.5598516464233398,
|
| 1380 |
+
"rewards/margins": 1.0026460886001587,
|
| 1381 |
+
"rewards/rejected": -2.562497615814209,
|
| 1382 |
+
"step": 820
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 0.8686551543694401,
|
| 1386 |
+
"grad_norm": 42.163832663541584,
|
| 1387 |
+
"learning_rate": 2.567240179368185e-08,
|
| 1388 |
+
"logits/chosen": 0.6047491431236267,
|
| 1389 |
+
"logits/rejected": 1.7334954738616943,
|
| 1390 |
+
"logps/chosen": -419.39410400390625,
|
| 1391 |
+
"logps/rejected": -516.76171875,
|
| 1392 |
+
"loss": 0.7858,
|
| 1393 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1394 |
+
"rewards/chosen": -1.612444281578064,
|
| 1395 |
+
"rewards/margins": 1.0573890209197998,
|
| 1396 |
+
"rewards/rejected": -2.669833183288574,
|
| 1397 |
+
"step": 830
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 0.8791208791208791,
|
| 1401 |
+
"grad_norm": 39.25207064131512,
|
| 1402 |
+
"learning_rate": 2.1787515014630357e-08,
|
| 1403 |
+
"logits/chosen": 0.570409893989563,
|
| 1404 |
+
"logits/rejected": 2.050328493118286,
|
| 1405 |
+
"logps/chosen": -446.327392578125,
|
| 1406 |
+
"logps/rejected": -537.5885009765625,
|
| 1407 |
+
"loss": 0.7303,
|
| 1408 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1409 |
+
"rewards/chosen": -1.6405481100082397,
|
| 1410 |
+
"rewards/margins": 1.0744378566741943,
|
| 1411 |
+
"rewards/rejected": -2.7149858474731445,
|
| 1412 |
+
"step": 840
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 0.8895866038723181,
|
| 1416 |
+
"grad_norm": 38.654218714989135,
|
| 1417 |
+
"learning_rate": 1.820784220652766e-08,
|
| 1418 |
+
"logits/chosen": 0.9261777997016907,
|
| 1419 |
+
"logits/rejected": 2.4772746562957764,
|
| 1420 |
+
"logps/chosen": -441.26593017578125,
|
| 1421 |
+
"logps/rejected": -515.775634765625,
|
| 1422 |
+
"loss": 0.7198,
|
| 1423 |
+
"rewards/accuracies": 0.8062499761581421,
|
| 1424 |
+
"rewards/chosen": -1.6039543151855469,
|
| 1425 |
+
"rewards/margins": 1.2189046144485474,
|
| 1426 |
+
"rewards/rejected": -2.822859048843384,
|
| 1427 |
+
"step": 850
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 0.9000523286237572,
|
| 1431 |
+
"grad_norm": 40.349585259848205,
|
| 1432 |
+
"learning_rate": 1.4938170864468636e-08,
|
| 1433 |
+
"logits/chosen": 0.8169757127761841,
|
| 1434 |
+
"logits/rejected": 2.297128200531006,
|
| 1435 |
+
"logps/chosen": -451.08831787109375,
|
| 1436 |
+
"logps/rejected": -525.7896118164062,
|
| 1437 |
+
"loss": 0.7621,
|
| 1438 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1439 |
+
"rewards/chosen": -1.6278215646743774,
|
| 1440 |
+
"rewards/margins": 1.0165573358535767,
|
| 1441 |
+
"rewards/rejected": -2.644378662109375,
|
| 1442 |
+
"step": 860
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 0.9105180533751962,
|
| 1446 |
+
"grad_norm": 51.17316772784962,
|
| 1447 |
+
"learning_rate": 1.1982873884064465e-08,
|
| 1448 |
+
"logits/chosen": 0.41630926728248596,
|
| 1449 |
+
"logits/rejected": 1.8632510900497437,
|
| 1450 |
+
"logps/chosen": -424.0140686035156,
|
| 1451 |
+
"logps/rejected": -504.92547607421875,
|
| 1452 |
+
"loss": 0.7708,
|
| 1453 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 1454 |
+
"rewards/chosen": -1.5223877429962158,
|
| 1455 |
+
"rewards/margins": 1.0619755983352661,
|
| 1456 |
+
"rewards/rejected": -2.5843632221221924,
|
| 1457 |
+
"step": 870
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 0.9209837781266352,
|
| 1461 |
+
"grad_norm": 101.12783488833479,
|
| 1462 |
+
"learning_rate": 9.345903713082304e-09,
|
| 1463 |
+
"logits/chosen": 0.549272894859314,
|
| 1464 |
+
"logits/rejected": 1.6597293615341187,
|
| 1465 |
+
"logps/chosen": -451.48077392578125,
|
| 1466 |
+
"logps/rejected": -547.0665893554688,
|
| 1467 |
+
"loss": 0.765,
|
| 1468 |
+
"rewards/accuracies": 0.75,
|
| 1469 |
+
"rewards/chosen": -1.473332166671753,
|
| 1470 |
+
"rewards/margins": 1.0507396459579468,
|
| 1471 |
+
"rewards/rejected": -2.52407169342041,
|
| 1472 |
+
"step": 880
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 0.9314495028780743,
|
| 1476 |
+
"grad_norm": 38.221448327358026,
|
| 1477 |
+
"learning_rate": 7.030787065396865e-09,
|
| 1478 |
+
"logits/chosen": 0.5618971586227417,
|
| 1479 |
+
"logits/rejected": 1.767998456954956,
|
| 1480 |
+
"logps/chosen": -447.612060546875,
|
| 1481 |
+
"logps/rejected": -547.6981201171875,
|
| 1482 |
+
"loss": 0.763,
|
| 1483 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1484 |
+
"rewards/chosen": -1.5987756252288818,
|
| 1485 |
+
"rewards/margins": 1.075731873512268,
|
| 1486 |
+
"rewards/rejected": -2.6745076179504395,
|
| 1487 |
+
"step": 890
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 0.9419152276295133,
|
| 1491 |
+
"grad_norm": 44.6705960527571,
|
| 1492 |
+
"learning_rate": 5.04062020432286e-09,
|
| 1493 |
+
"logits/chosen": 0.39258813858032227,
|
| 1494 |
+
"logits/rejected": 1.4880520105361938,
|
| 1495 |
+
"logps/chosen": -429.32666015625,
|
| 1496 |
+
"logps/rejected": -534.9125366210938,
|
| 1497 |
+
"loss": 0.7471,
|
| 1498 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1499 |
+
"rewards/chosen": -1.4897866249084473,
|
| 1500 |
+
"rewards/margins": 1.0451949834823608,
|
| 1501 |
+
"rewards/rejected": -2.5349814891815186,
|
| 1502 |
+
"step": 900
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 0.9419152276295133,
|
| 1506 |
+
"eval_logits/chosen": 0.5837222337722778,
|
| 1507 |
+
"eval_logits/rejected": 1.9212509393692017,
|
| 1508 |
+
"eval_logps/chosen": -439.8711853027344,
|
| 1509 |
+
"eval_logps/rejected": -529.835205078125,
|
| 1510 |
+
"eval_loss": 0.7707245945930481,
|
| 1511 |
+
"eval_rewards/accuracies": 0.7857142686843872,
|
| 1512 |
+
"eval_rewards/chosen": -1.5789415836334229,
|
| 1513 |
+
"eval_rewards/margins": 1.1173815727233887,
|
| 1514 |
+
"eval_rewards/rejected": -2.6963229179382324,
|
| 1515 |
+
"eval_runtime": 434.4266,
|
| 1516 |
+
"eval_samples_per_second": 4.604,
|
| 1517 |
+
"eval_steps_per_second": 0.145,
|
| 1518 |
+
"step": 900
|
| 1519 |
+
},
|
| 1520 |
+
{
|
| 1521 |
+
"epoch": 0.9523809523809523,
|
| 1522 |
+
"grad_norm": 67.71518744315276,
|
| 1523 |
+
"learning_rate": 3.3780648016376866e-09,
|
| 1524 |
+
"logits/chosen": 0.7767223119735718,
|
| 1525 |
+
"logits/rejected": 1.9748680591583252,
|
| 1526 |
+
"logps/chosen": -378.9118347167969,
|
| 1527 |
+
"logps/rejected": -469.43475341796875,
|
| 1528 |
+
"loss": 0.7758,
|
| 1529 |
+
"rewards/accuracies": 0.78125,
|
| 1530 |
+
"rewards/chosen": -1.5541870594024658,
|
| 1531 |
+
"rewards/margins": 1.01143479347229,
|
| 1532 |
+
"rewards/rejected": -2.565621852874756,
|
| 1533 |
+
"step": 910
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 0.9628466771323915,
|
| 1537 |
+
"grad_norm": 41.38648210926084,
|
| 1538 |
+
"learning_rate": 2.0453443778310766e-09,
|
| 1539 |
+
"logits/chosen": 0.674950897693634,
|
| 1540 |
+
"logits/rejected": 1.9640384912490845,
|
| 1541 |
+
"logps/chosen": -441.17425537109375,
|
| 1542 |
+
"logps/rejected": -512.7354736328125,
|
| 1543 |
+
"loss": 0.7815,
|
| 1544 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1545 |
+
"rewards/chosen": -1.5449568033218384,
|
| 1546 |
+
"rewards/margins": 1.0664222240447998,
|
| 1547 |
+
"rewards/rejected": -2.6113791465759277,
|
| 1548 |
+
"step": 920
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 0.9733124018838305,
|
| 1552 |
+
"grad_norm": 44.32409040550577,
|
| 1553 |
+
"learning_rate": 1.0442413283435758e-09,
|
| 1554 |
+
"logits/chosen": 0.3402867913246155,
|
| 1555 |
+
"logits/rejected": 2.12355375289917,
|
| 1556 |
+
"logps/chosen": -458.4666442871094,
|
| 1557 |
+
"logps/rejected": -520.189697265625,
|
| 1558 |
+
"loss": 0.7506,
|
| 1559 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1560 |
+
"rewards/chosen": -1.4366506338119507,
|
| 1561 |
+
"rewards/margins": 1.1228129863739014,
|
| 1562 |
+
"rewards/rejected": -2.5594632625579834,
|
| 1563 |
+
"step": 930
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 0.9837781266352695,
|
| 1567 |
+
"grad_norm": 37.11757505544597,
|
| 1568 |
+
"learning_rate": 3.760945397705828e-10,
|
| 1569 |
+
"logits/chosen": 0.5025678873062134,
|
| 1570 |
+
"logits/rejected": 1.861249327659607,
|
| 1571 |
+
"logps/chosen": -410.0967712402344,
|
| 1572 |
+
"logps/rejected": -514.7454223632812,
|
| 1573 |
+
"loss": 0.7376,
|
| 1574 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1575 |
+
"rewards/chosen": -1.468609094619751,
|
| 1576 |
+
"rewards/margins": 1.0813006162643433,
|
| 1577 |
+
"rewards/rejected": -2.549909830093384,
|
| 1578 |
+
"step": 940
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 0.9942438513867086,
|
| 1582 |
+
"grad_norm": 43.09246402370499,
|
| 1583 |
+
"learning_rate": 4.17975992204056e-11,
|
| 1584 |
+
"logits/chosen": 0.6304216384887695,
|
| 1585 |
+
"logits/rejected": 1.4559619426727295,
|
| 1586 |
+
"logps/chosen": -462.8529357910156,
|
| 1587 |
+
"logps/rejected": -536.662109375,
|
| 1588 |
+
"loss": 0.757,
|
| 1589 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1590 |
+
"rewards/chosen": -1.6002047061920166,
|
| 1591 |
+
"rewards/margins": 0.9165245890617371,
|
| 1592 |
+
"rewards/rejected": -2.5167293548583984,
|
| 1593 |
+
"step": 950
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 0.9994767137624281,
|
| 1597 |
+
"step": 955,
|
| 1598 |
+
"total_flos": 0.0,
|
| 1599 |
+
"train_loss": 0.8142244146756477,
|
| 1600 |
+
"train_runtime": 34283.0174,
|
| 1601 |
+
"train_samples_per_second": 1.783,
|
| 1602 |
+
"train_steps_per_second": 0.028
|
| 1603 |
+
}
|
| 1604 |
+
],
|
| 1605 |
+
"logging_steps": 10,
|
| 1606 |
+
"max_steps": 955,
|
| 1607 |
+
"num_input_tokens_seen": 0,
|
| 1608 |
+
"num_train_epochs": 1,
|
| 1609 |
+
"save_steps": 100,
|
| 1610 |
+
"stateful_callbacks": {
|
| 1611 |
+
"TrainerControl": {
|
| 1612 |
+
"args": {
|
| 1613 |
+
"should_epoch_stop": false,
|
| 1614 |
+
"should_evaluate": false,
|
| 1615 |
+
"should_log": false,
|
| 1616 |
+
"should_save": true,
|
| 1617 |
+
"should_training_stop": true
|
| 1618 |
+
},
|
| 1619 |
+
"attributes": {}
|
| 1620 |
+
}
|
| 1621 |
+
},
|
| 1622 |
+
"total_flos": 0.0,
|
| 1623 |
+
"train_batch_size": 8,
|
| 1624 |
+
"trial_name": null,
|
| 1625 |
+
"trial_params": null
|
| 1626 |
+
}
|