Model save
Browse files- README.md +82 -0
- all_results.json +9 -0
- train_results.json +9 -0
- trainer_state.json +1626 -0
README.md
ADDED
|
@@ -0,0 +1,82 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
base_model: princeton-nlp/Llama-3-Base-8B-SFT
|
| 3 |
+
library_name: peft
|
| 4 |
+
tags:
|
| 5 |
+
- trl
|
| 6 |
+
- dpo
|
| 7 |
+
- generated_from_trainer
|
| 8 |
+
model-index:
|
| 9 |
+
- name: llama3-dpo-lora
|
| 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 |
+
# llama3-dpo-lora
|
| 17 |
+
|
| 18 |
+
This model is a fine-tuned version of [princeton-nlp/Llama-3-Base-8B-SFT](https://huggingface.co/princeton-nlp/Llama-3-Base-8B-SFT) on the None dataset.
|
| 19 |
+
It achieves the following results on the evaluation set:
|
| 20 |
+
- Loss: 0.5190
|
| 21 |
+
- Rewards/chosen: -0.1399
|
| 22 |
+
- Rewards/rejected: -0.9459
|
| 23 |
+
- Rewards/accuracies: 0.7260
|
| 24 |
+
- Rewards/margins: 0.8060
|
| 25 |
+
- Logps/rejected: -283.9163
|
| 26 |
+
- Logps/chosen: -291.1600
|
| 27 |
+
- Logits/rejected: -0.3917
|
| 28 |
+
- Logits/chosen: -0.4219
|
| 29 |
+
|
| 30 |
+
## Model description
|
| 31 |
+
|
| 32 |
+
More information needed
|
| 33 |
+
|
| 34 |
+
## Intended uses & limitations
|
| 35 |
+
|
| 36 |
+
More information needed
|
| 37 |
+
|
| 38 |
+
## Training and evaluation data
|
| 39 |
+
|
| 40 |
+
More information needed
|
| 41 |
+
|
| 42 |
+
## Training procedure
|
| 43 |
+
|
| 44 |
+
### Training hyperparameters
|
| 45 |
+
|
| 46 |
+
The following hyperparameters were used during training:
|
| 47 |
+
- learning_rate: 5e-06
|
| 48 |
+
- train_batch_size: 1
|
| 49 |
+
- eval_batch_size: 4
|
| 50 |
+
- seed: 42
|
| 51 |
+
- distributed_type: multi-GPU
|
| 52 |
+
- num_devices: 4
|
| 53 |
+
- gradient_accumulation_steps: 16
|
| 54 |
+
- total_train_batch_size: 64
|
| 55 |
+
- total_eval_batch_size: 16
|
| 56 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 57 |
+
- lr_scheduler_type: cosine
|
| 58 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 59 |
+
- num_epochs: 1
|
| 60 |
+
|
| 61 |
+
### Training results
|
| 62 |
+
|
| 63 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 64 |
+
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 65 |
+
| 0.6297 | 0.1047 | 100 | 0.6140 | 0.1358 | -0.1277 | 0.6960 | 0.2634 | -275.7340 | -288.4034 | -0.5479 | -0.5526 |
|
| 66 |
+
| 0.5676 | 0.2094 | 200 | 0.5569 | -0.1144 | -0.6599 | 0.7000 | 0.5455 | -281.0560 | -290.9051 | -0.4945 | -0.5116 |
|
| 67 |
+
| 0.5414 | 0.3141 | 300 | 0.5403 | -0.3808 | -1.0461 | 0.7260 | 0.6652 | -284.9180 | -293.5698 | -0.4540 | -0.4775 |
|
| 68 |
+
| 0.5124 | 0.4187 | 400 | 0.5341 | -0.2337 | -0.9896 | 0.7040 | 0.7559 | -284.3532 | -292.0986 | -0.4243 | -0.4516 |
|
| 69 |
+
| 0.5529 | 0.5234 | 500 | 0.5260 | -0.2177 | -1.0037 | 0.7240 | 0.7861 | -284.4948 | -291.9380 | -0.3995 | -0.4290 |
|
| 70 |
+
| 0.53 | 0.6281 | 600 | 0.5244 | -0.0687 | -0.8583 | 0.7200 | 0.7895 | -283.0403 | -290.4489 | -0.4028 | -0.4317 |
|
| 71 |
+
| 0.5028 | 0.7328 | 700 | 0.5190 | -0.3357 | -1.1360 | 0.7320 | 0.8003 | -285.8177 | -293.1184 | -0.3874 | -0.4179 |
|
| 72 |
+
| 0.5347 | 0.8375 | 800 | 0.5191 | -0.1404 | -0.9419 | 0.7320 | 0.8015 | -283.8760 | -291.1650 | -0.3924 | -0.4225 |
|
| 73 |
+
| 0.4783 | 0.9422 | 900 | 0.5190 | -0.1399 | -0.9459 | 0.7260 | 0.8060 | -283.9163 | -291.1600 | -0.3917 | -0.4219 |
|
| 74 |
+
|
| 75 |
+
|
| 76 |
+
### Framework versions
|
| 77 |
+
|
| 78 |
+
- PEFT 0.7.1
|
| 79 |
+
- Transformers 4.44.2
|
| 80 |
+
- Pytorch 2.2.1+cu121
|
| 81 |
+
- Datasets 2.14.6
|
| 82 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9997382884061764,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.543355998318857,
|
| 5 |
+
"train_runtime": 19333.2554,
|
| 6 |
+
"train_samples": 61135,
|
| 7 |
+
"train_samples_per_second": 3.162,
|
| 8 |
+
"train_steps_per_second": 0.049
|
| 9 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 0.9997382884061764,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.543355998318857,
|
| 5 |
+
"train_runtime": 19333.2554,
|
| 6 |
+
"train_samples": 61135,
|
| 7 |
+
"train_samples_per_second": 3.162,
|
| 8 |
+
"train_steps_per_second": 0.049
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1626 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9997382884061764,
|
| 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": 0.0010468463752944255,
|
| 13 |
+
"grad_norm": 5.5,
|
| 14 |
+
"learning_rate": 5.208333333333333e-08,
|
| 15 |
+
"logits/chosen": -0.32969313859939575,
|
| 16 |
+
"logits/rejected": -0.3478139042854309,
|
| 17 |
+
"logps/chosen": -282.7626953125,
|
| 18 |
+
"logps/rejected": -210.6118621826172,
|
| 19 |
+
"loss": 0.6931,
|
| 20 |
+
"rewards/accuracies": 0.0,
|
| 21 |
+
"rewards/chosen": 0.0,
|
| 22 |
+
"rewards/margins": 0.0,
|
| 23 |
+
"rewards/rejected": 0.0,
|
| 24 |
+
"step": 1
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.010468463752944255,
|
| 28 |
+
"grad_norm": 6.0,
|
| 29 |
+
"learning_rate": 5.208333333333334e-07,
|
| 30 |
+
"logits/chosen": -0.5201655030250549,
|
| 31 |
+
"logits/rejected": -0.5292737483978271,
|
| 32 |
+
"logps/chosen": -312.750244140625,
|
| 33 |
+
"logps/rejected": -275.4716491699219,
|
| 34 |
+
"loss": 0.6919,
|
| 35 |
+
"rewards/accuracies": 0.4652777910232544,
|
| 36 |
+
"rewards/chosen": 0.0034895986318588257,
|
| 37 |
+
"rewards/margins": 0.002287701005116105,
|
| 38 |
+
"rewards/rejected": 0.001201898674480617,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.02093692750588851,
|
| 43 |
+
"grad_norm": 5.5,
|
| 44 |
+
"learning_rate": 1.0416666666666667e-06,
|
| 45 |
+
"logits/chosen": -0.4813454747200012,
|
| 46 |
+
"logits/rejected": -0.5021841526031494,
|
| 47 |
+
"logps/chosen": -304.3456726074219,
|
| 48 |
+
"logps/rejected": -268.962158203125,
|
| 49 |
+
"loss": 0.6934,
|
| 50 |
+
"rewards/accuracies": 0.5375000238418579,
|
| 51 |
+
"rewards/chosen": -0.0017953884089365602,
|
| 52 |
+
"rewards/margins": 0.0013109362917020917,
|
| 53 |
+
"rewards/rejected": -0.003106324467808008,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.031405391258832765,
|
| 58 |
+
"grad_norm": 5.78125,
|
| 59 |
+
"learning_rate": 1.5625e-06,
|
| 60 |
+
"logits/chosen": -0.4880186915397644,
|
| 61 |
+
"logits/rejected": -0.4969080984592438,
|
| 62 |
+
"logps/chosen": -287.50286865234375,
|
| 63 |
+
"logps/rejected": -249.9375457763672,
|
| 64 |
+
"loss": 0.688,
|
| 65 |
+
"rewards/accuracies": 0.5562499761581421,
|
| 66 |
+
"rewards/chosen": 0.010473029688000679,
|
| 67 |
+
"rewards/margins": 0.01431797631084919,
|
| 68 |
+
"rewards/rejected": -0.003844945225864649,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.04187385501177702,
|
| 73 |
+
"grad_norm": 5.375,
|
| 74 |
+
"learning_rate": 2.0833333333333334e-06,
|
| 75 |
+
"logits/chosen": -0.46122869849205017,
|
| 76 |
+
"logits/rejected": -0.4958384037017822,
|
| 77 |
+
"logps/chosen": -303.1500549316406,
|
| 78 |
+
"logps/rejected": -241.998046875,
|
| 79 |
+
"loss": 0.683,
|
| 80 |
+
"rewards/accuracies": 0.6499999761581421,
|
| 81 |
+
"rewards/chosen": 0.03339455649256706,
|
| 82 |
+
"rewards/margins": 0.0322079211473465,
|
| 83 |
+
"rewards/rejected": 0.0011866316199302673,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.05234231876472128,
|
| 88 |
+
"grad_norm": 4.59375,
|
| 89 |
+
"learning_rate": 2.604166666666667e-06,
|
| 90 |
+
"logits/chosen": -0.5217183828353882,
|
| 91 |
+
"logits/rejected": -0.5492846369743347,
|
| 92 |
+
"logps/chosen": -302.4110412597656,
|
| 93 |
+
"logps/rejected": -280.4798278808594,
|
| 94 |
+
"loss": 0.6784,
|
| 95 |
+
"rewards/accuracies": 0.625,
|
| 96 |
+
"rewards/chosen": 0.07218249142169952,
|
| 97 |
+
"rewards/margins": 0.04362189769744873,
|
| 98 |
+
"rewards/rejected": 0.028560593724250793,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.06281078251766553,
|
| 103 |
+
"grad_norm": 5.03125,
|
| 104 |
+
"learning_rate": 3.125e-06,
|
| 105 |
+
"logits/chosen": -0.5444208979606628,
|
| 106 |
+
"logits/rejected": -0.5450385212898254,
|
| 107 |
+
"logps/chosen": -288.3908996582031,
|
| 108 |
+
"logps/rejected": -252.4159393310547,
|
| 109 |
+
"loss": 0.662,
|
| 110 |
+
"rewards/accuracies": 0.625,
|
| 111 |
+
"rewards/chosen": 0.10959690809249878,
|
| 112 |
+
"rewards/margins": 0.04304002225399017,
|
| 113 |
+
"rewards/rejected": 0.0665568858385086,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.07327924627060979,
|
| 118 |
+
"grad_norm": 5.21875,
|
| 119 |
+
"learning_rate": 3.6458333333333333e-06,
|
| 120 |
+
"logits/chosen": -0.5294634699821472,
|
| 121 |
+
"logits/rejected": -0.5446035265922546,
|
| 122 |
+
"logps/chosen": -284.2558288574219,
|
| 123 |
+
"logps/rejected": -261.0161437988281,
|
| 124 |
+
"loss": 0.6465,
|
| 125 |
+
"rewards/accuracies": 0.6875,
|
| 126 |
+
"rewards/chosen": 0.1622026413679123,
|
| 127 |
+
"rewards/margins": 0.1133759468793869,
|
| 128 |
+
"rewards/rejected": 0.04882669448852539,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.08374771002355404,
|
| 133 |
+
"grad_norm": 4.78125,
|
| 134 |
+
"learning_rate": 4.166666666666667e-06,
|
| 135 |
+
"logits/chosen": -0.46931973099708557,
|
| 136 |
+
"logits/rejected": -0.4959704279899597,
|
| 137 |
+
"logps/chosen": -285.55169677734375,
|
| 138 |
+
"logps/rejected": -271.7218017578125,
|
| 139 |
+
"loss": 0.6312,
|
| 140 |
+
"rewards/accuracies": 0.706250011920929,
|
| 141 |
+
"rewards/chosen": 0.2240675389766693,
|
| 142 |
+
"rewards/margins": 0.17874710261821747,
|
| 143 |
+
"rewards/rejected": 0.04532044008374214,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.0942161737764983,
|
| 148 |
+
"grad_norm": 5.90625,
|
| 149 |
+
"learning_rate": 4.6875000000000004e-06,
|
| 150 |
+
"logits/chosen": -0.5021494626998901,
|
| 151 |
+
"logits/rejected": -0.5237849950790405,
|
| 152 |
+
"logps/chosen": -328.7523193359375,
|
| 153 |
+
"logps/rejected": -293.5586853027344,
|
| 154 |
+
"loss": 0.6197,
|
| 155 |
+
"rewards/accuracies": 0.71875,
|
| 156 |
+
"rewards/chosen": 0.19337503612041473,
|
| 157 |
+
"rewards/margins": 0.22977392375469208,
|
| 158 |
+
"rewards/rejected": -0.036398887634277344,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.10468463752944256,
|
| 163 |
+
"grad_norm": 5.65625,
|
| 164 |
+
"learning_rate": 4.9997324926814375e-06,
|
| 165 |
+
"logits/chosen": -0.5387491583824158,
|
| 166 |
+
"logits/rejected": -0.5343273282051086,
|
| 167 |
+
"logps/chosen": -274.4442443847656,
|
| 168 |
+
"logps/rejected": -288.30035400390625,
|
| 169 |
+
"loss": 0.6297,
|
| 170 |
+
"rewards/accuracies": 0.637499988079071,
|
| 171 |
+
"rewards/chosen": 0.221399188041687,
|
| 172 |
+
"rewards/margins": 0.2429426610469818,
|
| 173 |
+
"rewards/rejected": -0.0215434692800045,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.10468463752944256,
|
| 178 |
+
"eval_logits/chosen": -0.5525645613670349,
|
| 179 |
+
"eval_logits/rejected": -0.5478540062904358,
|
| 180 |
+
"eval_logps/chosen": -288.4034118652344,
|
| 181 |
+
"eval_logps/rejected": -275.7340087890625,
|
| 182 |
+
"eval_loss": 0.6140495538711548,
|
| 183 |
+
"eval_rewards/accuracies": 0.6959999799728394,
|
| 184 |
+
"eval_rewards/chosen": 0.13576927781105042,
|
| 185 |
+
"eval_rewards/margins": 0.26342061161994934,
|
| 186 |
+
"eval_rewards/rejected": -0.1276513636112213,
|
| 187 |
+
"eval_runtime": 351.3471,
|
| 188 |
+
"eval_samples_per_second": 5.692,
|
| 189 |
+
"eval_steps_per_second": 0.356,
|
| 190 |
+
"step": 100
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.11515310128238682,
|
| 194 |
+
"grad_norm": 4.25,
|
| 195 |
+
"learning_rate": 4.996723692767927e-06,
|
| 196 |
+
"logits/chosen": -0.5775566101074219,
|
| 197 |
+
"logits/rejected": -0.603683590888977,
|
| 198 |
+
"logps/chosen": -287.82733154296875,
|
| 199 |
+
"logps/rejected": -275.31671142578125,
|
| 200 |
+
"loss": 0.606,
|
| 201 |
+
"rewards/accuracies": 0.71875,
|
| 202 |
+
"rewards/chosen": 0.15195702016353607,
|
| 203 |
+
"rewards/margins": 0.34537211060523987,
|
| 204 |
+
"rewards/rejected": -0.1934150755405426,
|
| 205 |
+
"step": 110
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.12562156503533106,
|
| 209 |
+
"grad_norm": 4.96875,
|
| 210 |
+
"learning_rate": 4.9903757462135984e-06,
|
| 211 |
+
"logits/chosen": -0.5295812487602234,
|
| 212 |
+
"logits/rejected": -0.5440601110458374,
|
| 213 |
+
"logps/chosen": -260.30078125,
|
| 214 |
+
"logps/rejected": -251.58401489257812,
|
| 215 |
+
"loss": 0.5924,
|
| 216 |
+
"rewards/accuracies": 0.6625000238418579,
|
| 217 |
+
"rewards/chosen": 0.02338511124253273,
|
| 218 |
+
"rewards/margins": 0.281913697719574,
|
| 219 |
+
"rewards/rejected": -0.25852862000465393,
|
| 220 |
+
"step": 120
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.1360900287882753,
|
| 224 |
+
"grad_norm": 5.46875,
|
| 225 |
+
"learning_rate": 4.980697142834315e-06,
|
| 226 |
+
"logits/chosen": -0.49980980157852173,
|
| 227 |
+
"logits/rejected": -0.5171049237251282,
|
| 228 |
+
"logps/chosen": -300.2707214355469,
|
| 229 |
+
"logps/rejected": -336.19036865234375,
|
| 230 |
+
"loss": 0.5933,
|
| 231 |
+
"rewards/accuracies": 0.65625,
|
| 232 |
+
"rewards/chosen": 0.02033509686589241,
|
| 233 |
+
"rewards/margins": 0.3162926733493805,
|
| 234 |
+
"rewards/rejected": -0.2959575653076172,
|
| 235 |
+
"step": 130
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.14655849254121958,
|
| 239 |
+
"grad_norm": 4.75,
|
| 240 |
+
"learning_rate": 4.967700826904229e-06,
|
| 241 |
+
"logits/chosen": -0.5714845061302185,
|
| 242 |
+
"logits/rejected": -0.5790292024612427,
|
| 243 |
+
"logps/chosen": -280.66845703125,
|
| 244 |
+
"logps/rejected": -273.6711120605469,
|
| 245 |
+
"loss": 0.5811,
|
| 246 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 247 |
+
"rewards/chosen": -0.004461909644305706,
|
| 248 |
+
"rewards/margins": 0.407376229763031,
|
| 249 |
+
"rewards/rejected": -0.4118381440639496,
|
| 250 |
+
"step": 140
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.15702695629416383,
|
| 254 |
+
"grad_norm": 4.84375,
|
| 255 |
+
"learning_rate": 4.951404179843963e-06,
|
| 256 |
+
"logits/chosen": -0.5678334832191467,
|
| 257 |
+
"logits/rejected": -0.5395095944404602,
|
| 258 |
+
"logps/chosen": -306.054443359375,
|
| 259 |
+
"logps/rejected": -276.5906066894531,
|
| 260 |
+
"loss": 0.5777,
|
| 261 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 262 |
+
"rewards/chosen": 0.13024172186851501,
|
| 263 |
+
"rewards/margins": 0.4546026587486267,
|
| 264 |
+
"rewards/rejected": -0.3243609070777893,
|
| 265 |
+
"step": 150
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.16749542004710807,
|
| 269 |
+
"grad_norm": 4.375,
|
| 270 |
+
"learning_rate": 4.931828996974498e-06,
|
| 271 |
+
"logits/chosen": -0.5105162858963013,
|
| 272 |
+
"logits/rejected": -0.5018196702003479,
|
| 273 |
+
"logps/chosen": -296.04840087890625,
|
| 274 |
+
"logps/rejected": -270.5509033203125,
|
| 275 |
+
"loss": 0.5478,
|
| 276 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 277 |
+
"rewards/chosen": 0.04980681464076042,
|
| 278 |
+
"rewards/margins": 0.5175832509994507,
|
| 279 |
+
"rewards/rejected": -0.4677763879299164,
|
| 280 |
+
"step": 160
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.17796388380005235,
|
| 284 |
+
"grad_norm": 4.78125,
|
| 285 |
+
"learning_rate": 4.909001458367867e-06,
|
| 286 |
+
"logits/chosen": -0.5939355492591858,
|
| 287 |
+
"logits/rejected": -0.5804780125617981,
|
| 288 |
+
"logps/chosen": -287.94708251953125,
|
| 289 |
+
"logps/rejected": -275.73980712890625,
|
| 290 |
+
"loss": 0.5789,
|
| 291 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 292 |
+
"rewards/chosen": -0.19239839911460876,
|
| 293 |
+
"rewards/margins": 0.4369952082633972,
|
| 294 |
+
"rewards/rejected": -0.6293936967849731,
|
| 295 |
+
"step": 170
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 0.1884323475529966,
|
| 299 |
+
"grad_norm": 5.15625,
|
| 300 |
+
"learning_rate": 4.882952093833628e-06,
|
| 301 |
+
"logits/chosen": -0.5993765592575073,
|
| 302 |
+
"logits/rejected": -0.5707160830497742,
|
| 303 |
+
"logps/chosen": -301.9080505371094,
|
| 304 |
+
"logps/rejected": -265.2618713378906,
|
| 305 |
+
"loss": 0.555,
|
| 306 |
+
"rewards/accuracies": 0.706250011920929,
|
| 307 |
+
"rewards/chosen": -0.19447076320648193,
|
| 308 |
+
"rewards/margins": 0.48417288064956665,
|
| 309 |
+
"rewards/rejected": -0.6786437630653381,
|
| 310 |
+
"step": 180
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.19890081130594087,
|
| 314 |
+
"grad_norm": 5.90625,
|
| 315 |
+
"learning_rate": 4.853715742087947e-06,
|
| 316 |
+
"logits/chosen": -0.5384653806686401,
|
| 317 |
+
"logits/rejected": -0.5204354524612427,
|
| 318 |
+
"logps/chosen": -275.14886474609375,
|
| 319 |
+
"logps/rejected": -281.9067687988281,
|
| 320 |
+
"loss": 0.5706,
|
| 321 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 322 |
+
"rewards/chosen": -0.033171653747558594,
|
| 323 |
+
"rewards/margins": 0.469027042388916,
|
| 324 |
+
"rewards/rejected": -0.5021986961364746,
|
| 325 |
+
"step": 190
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.2093692750588851,
|
| 329 |
+
"grad_norm": 6.5625,
|
| 330 |
+
"learning_rate": 4.821331504159906e-06,
|
| 331 |
+
"logits/chosen": -0.5440086722373962,
|
| 332 |
+
"logits/rejected": -0.5563145279884338,
|
| 333 |
+
"logps/chosen": -296.43060302734375,
|
| 334 |
+
"logps/rejected": -255.0513916015625,
|
| 335 |
+
"loss": 0.5676,
|
| 336 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 337 |
+
"rewards/chosen": -0.12640251219272614,
|
| 338 |
+
"rewards/margins": 0.5171942710876465,
|
| 339 |
+
"rewards/rejected": -0.6435968279838562,
|
| 340 |
+
"step": 200
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.2093692750588851,
|
| 344 |
+
"eval_logits/chosen": -0.5116367340087891,
|
| 345 |
+
"eval_logits/rejected": -0.4945087134838104,
|
| 346 |
+
"eval_logps/chosen": -290.90509033203125,
|
| 347 |
+
"eval_logps/rejected": -281.0560302734375,
|
| 348 |
+
"eval_loss": 0.5569251179695129,
|
| 349 |
+
"eval_rewards/accuracies": 0.699999988079071,
|
| 350 |
+
"eval_rewards/chosen": -0.11436504870653152,
|
| 351 |
+
"eval_rewards/margins": 0.5454893708229065,
|
| 352 |
+
"eval_rewards/rejected": -0.6598544716835022,
|
| 353 |
+
"eval_runtime": 349.6273,
|
| 354 |
+
"eval_samples_per_second": 5.72,
|
| 355 |
+
"eval_steps_per_second": 0.358,
|
| 356 |
+
"step": 200
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.21983773881182936,
|
| 360 |
+
"grad_norm": 6.15625,
|
| 361 |
+
"learning_rate": 4.7858426910973435e-06,
|
| 362 |
+
"logits/chosen": -0.5827375650405884,
|
| 363 |
+
"logits/rejected": -0.5831678509712219,
|
| 364 |
+
"logps/chosen": -278.0469970703125,
|
| 365 |
+
"logps/rejected": -271.9605407714844,
|
| 366 |
+
"loss": 0.5508,
|
| 367 |
+
"rewards/accuracies": 0.737500011920929,
|
| 368 |
+
"rewards/chosen": -0.025282684713602066,
|
| 369 |
+
"rewards/margins": 0.5209798216819763,
|
| 370 |
+
"rewards/rejected": -0.5462625622749329,
|
| 371 |
+
"step": 210
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 0.23030620256477363,
|
| 375 |
+
"grad_norm": 5.125,
|
| 376 |
+
"learning_rate": 4.747296766042161e-06,
|
| 377 |
+
"logits/chosen": -0.5575278997421265,
|
| 378 |
+
"logits/rejected": -0.5416348576545715,
|
| 379 |
+
"logps/chosen": -317.4739990234375,
|
| 380 |
+
"logps/rejected": -269.8050231933594,
|
| 381 |
+
"loss": 0.5368,
|
| 382 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 383 |
+
"rewards/chosen": 0.03362050652503967,
|
| 384 |
+
"rewards/margins": 0.569665789604187,
|
| 385 |
+
"rewards/rejected": -0.536045253276825,
|
| 386 |
+
"step": 220
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.24077466631771788,
|
| 390 |
+
"grad_norm": 5.6875,
|
| 391 |
+
"learning_rate": 4.705745280752586e-06,
|
| 392 |
+
"logits/chosen": -0.589636504650116,
|
| 393 |
+
"logits/rejected": -0.5510541200637817,
|
| 394 |
+
"logps/chosen": -290.55206298828125,
|
| 395 |
+
"logps/rejected": -287.7290954589844,
|
| 396 |
+
"loss": 0.5581,
|
| 397 |
+
"rewards/accuracies": 0.706250011920929,
|
| 398 |
+
"rewards/chosen": -0.08335064351558685,
|
| 399 |
+
"rewards/margins": 0.5679007768630981,
|
| 400 |
+
"rewards/rejected": -0.6512514352798462,
|
| 401 |
+
"step": 230
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.2512431300706621,
|
| 405 |
+
"grad_norm": 4.96875,
|
| 406 |
+
"learning_rate": 4.661243806657256e-06,
|
| 407 |
+
"logits/chosen": -0.5663197636604309,
|
| 408 |
+
"logits/rejected": -0.525894045829773,
|
| 409 |
+
"logps/chosen": -299.3938903808594,
|
| 410 |
+
"logps/rejected": -263.598876953125,
|
| 411 |
+
"loss": 0.5563,
|
| 412 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 413 |
+
"rewards/chosen": -0.1382952779531479,
|
| 414 |
+
"rewards/margins": 0.5307878255844116,
|
| 415 |
+
"rewards/rejected": -0.6690832376480103,
|
| 416 |
+
"step": 240
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.26171159382360637,
|
| 420 |
+
"grad_norm": 5.6875,
|
| 421 |
+
"learning_rate": 4.613851860533367e-06,
|
| 422 |
+
"logits/chosen": -0.5768588781356812,
|
| 423 |
+
"logits/rejected": -0.5346202850341797,
|
| 424 |
+
"logps/chosen": -293.30059814453125,
|
| 425 |
+
"logps/rejected": -259.82177734375,
|
| 426 |
+
"loss": 0.5789,
|
| 427 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 428 |
+
"rewards/chosen": -0.08053269237279892,
|
| 429 |
+
"rewards/margins": 0.5023098587989807,
|
| 430 |
+
"rewards/rejected": -0.5828425288200378,
|
| 431 |
+
"step": 250
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.2721800575765506,
|
| 435 |
+
"grad_norm": 5.6875,
|
| 436 |
+
"learning_rate": 4.563632824908252e-06,
|
| 437 |
+
"logits/chosen": -0.5458053946495056,
|
| 438 |
+
"logits/rejected": -0.5105060935020447,
|
| 439 |
+
"logps/chosen": -292.6302185058594,
|
| 440 |
+
"logps/rejected": -279.1827392578125,
|
| 441 |
+
"loss": 0.5404,
|
| 442 |
+
"rewards/accuracies": 0.78125,
|
| 443 |
+
"rewards/chosen": 0.12764385342597961,
|
| 444 |
+
"rewards/margins": 0.8003277778625488,
|
| 445 |
+
"rewards/rejected": -0.6726840138435364,
|
| 446 |
+
"step": 260
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.2826485213294949,
|
| 450 |
+
"grad_norm": 5.78125,
|
| 451 |
+
"learning_rate": 4.510653863290871e-06,
|
| 452 |
+
"logits/chosen": -0.545439600944519,
|
| 453 |
+
"logits/rejected": -0.5205134749412537,
|
| 454 |
+
"logps/chosen": -294.71783447265625,
|
| 455 |
+
"logps/rejected": -302.39599609375,
|
| 456 |
+
"loss": 0.5465,
|
| 457 |
+
"rewards/accuracies": 0.699999988079071,
|
| 458 |
+
"rewards/chosen": -0.015889674425125122,
|
| 459 |
+
"rewards/margins": 0.5484660267829895,
|
| 460 |
+
"rewards/rejected": -0.564355731010437,
|
| 461 |
+
"step": 270
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.29311698508243916,
|
| 465 |
+
"grad_norm": 4.78125,
|
| 466 |
+
"learning_rate": 4.454985830346574e-06,
|
| 467 |
+
"logits/chosen": -0.6070719957351685,
|
| 468 |
+
"logits/rejected": -0.5709980726242065,
|
| 469 |
+
"logps/chosen": -300.3115234375,
|
| 470 |
+
"logps/rejected": -283.37689208984375,
|
| 471 |
+
"loss": 0.5681,
|
| 472 |
+
"rewards/accuracies": 0.6187499761581421,
|
| 473 |
+
"rewards/chosen": -0.04909560829401016,
|
| 474 |
+
"rewards/margins": 0.496112197637558,
|
| 475 |
+
"rewards/rejected": -0.5452078580856323,
|
| 476 |
+
"step": 280
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.3035854488353834,
|
| 480 |
+
"grad_norm": 4.65625,
|
| 481 |
+
"learning_rate": 4.396703177135262e-06,
|
| 482 |
+
"logits/chosen": -0.5642537474632263,
|
| 483 |
+
"logits/rejected": -0.5349927544593811,
|
| 484 |
+
"logps/chosen": -285.5408935546875,
|
| 485 |
+
"logps/rejected": -256.77691650390625,
|
| 486 |
+
"loss": 0.5249,
|
| 487 |
+
"rewards/accuracies": 0.75,
|
| 488 |
+
"rewards/chosen": 0.10653576999902725,
|
| 489 |
+
"rewards/margins": 0.6360087990760803,
|
| 490 |
+
"rewards/rejected": -0.5294729471206665,
|
| 491 |
+
"step": 290
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.31405391258832765,
|
| 495 |
+
"grad_norm": 5.75,
|
| 496 |
+
"learning_rate": 4.335883851539693e-06,
|
| 497 |
+
"logits/chosen": -0.5834770798683167,
|
| 498 |
+
"logits/rejected": -0.544857382774353,
|
| 499 |
+
"logps/chosen": -296.65936279296875,
|
| 500 |
+
"logps/rejected": -293.7464599609375,
|
| 501 |
+
"loss": 0.5414,
|
| 502 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 503 |
+
"rewards/chosen": -0.30995655059814453,
|
| 504 |
+
"rewards/margins": 0.6452124714851379,
|
| 505 |
+
"rewards/rejected": -0.9551690220832825,
|
| 506 |
+
"step": 300
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.31405391258832765,
|
| 510 |
+
"eval_logits/chosen": -0.4775179624557495,
|
| 511 |
+
"eval_logits/rejected": -0.45395201444625854,
|
| 512 |
+
"eval_logps/chosen": -293.56982421875,
|
| 513 |
+
"eval_logps/rejected": -284.91802978515625,
|
| 514 |
+
"eval_loss": 0.5403068661689758,
|
| 515 |
+
"eval_rewards/accuracies": 0.7260000109672546,
|
| 516 |
+
"eval_rewards/chosen": -0.38083866238594055,
|
| 517 |
+
"eval_rewards/margins": 0.6652126312255859,
|
| 518 |
+
"eval_rewards/rejected": -1.0460513830184937,
|
| 519 |
+
"eval_runtime": 349.5769,
|
| 520 |
+
"eval_samples_per_second": 5.721,
|
| 521 |
+
"eval_steps_per_second": 0.358,
|
| 522 |
+
"step": 300
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.3245223763412719,
|
| 526 |
+
"grad_norm": 6.28125,
|
| 527 |
+
"learning_rate": 4.2726091940171055e-06,
|
| 528 |
+
"logits/chosen": -0.5361314415931702,
|
| 529 |
+
"logits/rejected": -0.5678786635398865,
|
| 530 |
+
"logps/chosen": -296.285400390625,
|
| 531 |
+
"logps/rejected": -342.7236022949219,
|
| 532 |
+
"loss": 0.5056,
|
| 533 |
+
"rewards/accuracies": 0.793749988079071,
|
| 534 |
+
"rewards/chosen": -0.19761790335178375,
|
| 535 |
+
"rewards/margins": 0.8299154043197632,
|
| 536 |
+
"rewards/rejected": -1.0275332927703857,
|
| 537 |
+
"step": 310
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.33499084009421615,
|
| 541 |
+
"grad_norm": 5.21875,
|
| 542 |
+
"learning_rate": 4.206963828813555e-06,
|
| 543 |
+
"logits/chosen": -0.5878556370735168,
|
| 544 |
+
"logits/rejected": -0.5477277040481567,
|
| 545 |
+
"logps/chosen": -296.2107849121094,
|
| 546 |
+
"logps/rejected": -280.1604919433594,
|
| 547 |
+
"loss": 0.524,
|
| 548 |
+
"rewards/accuracies": 0.737500011920929,
|
| 549 |
+
"rewards/chosen": -0.1949264109134674,
|
| 550 |
+
"rewards/margins": 0.7900466918945312,
|
| 551 |
+
"rewards/rejected": -0.9849729537963867,
|
| 552 |
+
"step": 320
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 0.34545930384716045,
|
| 556 |
+
"grad_norm": 5.0625,
|
| 557 |
+
"learning_rate": 4.139035550786495e-06,
|
| 558 |
+
"logits/chosen": -0.6225862503051758,
|
| 559 |
+
"logits/rejected": -0.5590678453445435,
|
| 560 |
+
"logps/chosen": -289.2728576660156,
|
| 561 |
+
"logps/rejected": -260.7643737792969,
|
| 562 |
+
"loss": 0.5217,
|
| 563 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 564 |
+
"rewards/chosen": -0.2666059136390686,
|
| 565 |
+
"rewards/margins": 0.6325812339782715,
|
| 566 |
+
"rewards/rejected": -0.8991872072219849,
|
| 567 |
+
"step": 330
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.3559277676001047,
|
| 571 |
+
"grad_norm": 5.96875,
|
| 572 |
+
"learning_rate": 4.068915207986931e-06,
|
| 573 |
+
"logits/chosen": -0.5779263377189636,
|
| 574 |
+
"logits/rejected": -0.5189130902290344,
|
| 575 |
+
"logps/chosen": -298.147705078125,
|
| 576 |
+
"logps/rejected": -258.82391357421875,
|
| 577 |
+
"loss": 0.5224,
|
| 578 |
+
"rewards/accuracies": 0.71875,
|
| 579 |
+
"rewards/chosen": -0.4627823233604431,
|
| 580 |
+
"rewards/margins": 0.7694815397262573,
|
| 581 |
+
"rewards/rejected": -1.2322638034820557,
|
| 582 |
+
"step": 340
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 0.36639623135304894,
|
| 586 |
+
"grad_norm": 6.21875,
|
| 587 |
+
"learning_rate": 3.996696580158211e-06,
|
| 588 |
+
"logits/chosen": -0.5267971158027649,
|
| 589 |
+
"logits/rejected": -0.5050491690635681,
|
| 590 |
+
"logps/chosen": -335.9749755859375,
|
| 591 |
+
"logps/rejected": -290.8840026855469,
|
| 592 |
+
"loss": 0.5206,
|
| 593 |
+
"rewards/accuracies": 0.737500011920929,
|
| 594 |
+
"rewards/chosen": -0.16786900162696838,
|
| 595 |
+
"rewards/margins": 0.7010205984115601,
|
| 596 |
+
"rewards/rejected": -0.8688896894454956,
|
| 597 |
+
"step": 350
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.3768646951059932,
|
| 601 |
+
"grad_norm": 5.3125,
|
| 602 |
+
"learning_rate": 3.922476253313921e-06,
|
| 603 |
+
"logits/chosen": -0.5035579800605774,
|
| 604 |
+
"logits/rejected": -0.5138710141181946,
|
| 605 |
+
"logps/chosen": -273.709716796875,
|
| 606 |
+
"logps/rejected": -296.85650634765625,
|
| 607 |
+
"loss": 0.5587,
|
| 608 |
+
"rewards/accuracies": 0.643750011920929,
|
| 609 |
+
"rewards/chosen": -0.28165164589881897,
|
| 610 |
+
"rewards/margins": 0.6714268922805786,
|
| 611 |
+
"rewards/rejected": -0.9530784487724304,
|
| 612 |
+
"step": 360
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 0.38733315885893743,
|
| 616 |
+
"grad_norm": 5.875,
|
| 617 |
+
"learning_rate": 3.846353490562664e-06,
|
| 618 |
+
"logits/chosen": -0.5329638123512268,
|
| 619 |
+
"logits/rejected": -0.5389483571052551,
|
| 620 |
+
"logps/chosen": -288.07745361328125,
|
| 621 |
+
"logps/rejected": -262.96124267578125,
|
| 622 |
+
"loss": 0.5038,
|
| 623 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 624 |
+
"rewards/chosen": -0.05612688139081001,
|
| 625 |
+
"rewards/margins": 0.8792274594306946,
|
| 626 |
+
"rewards/rejected": -0.9353543519973755,
|
| 627 |
+
"step": 370
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.39780162261188173,
|
| 631 |
+
"grad_norm": 5.25,
|
| 632 |
+
"learning_rate": 3.768430099352445e-06,
|
| 633 |
+
"logits/chosen": -0.5536895990371704,
|
| 634 |
+
"logits/rejected": -0.5606611371040344,
|
| 635 |
+
"logps/chosen": -304.82574462890625,
|
| 636 |
+
"logps/rejected": -277.5342712402344,
|
| 637 |
+
"loss": 0.5089,
|
| 638 |
+
"rewards/accuracies": 0.75,
|
| 639 |
+
"rewards/chosen": -0.04281007498502731,
|
| 640 |
+
"rewards/margins": 0.9195000529289246,
|
| 641 |
+
"rewards/rejected": -0.9623101353645325,
|
| 642 |
+
"step": 380
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 0.408270086364826,
|
| 646 |
+
"grad_norm": 6.5,
|
| 647 |
+
"learning_rate": 3.6888102953122307e-06,
|
| 648 |
+
"logits/chosen": -0.5975215435028076,
|
| 649 |
+
"logits/rejected": -0.5678552389144897,
|
| 650 |
+
"logps/chosen": -261.8155212402344,
|
| 651 |
+
"logps/rejected": -262.86151123046875,
|
| 652 |
+
"loss": 0.5554,
|
| 653 |
+
"rewards/accuracies": 0.6937500238418579,
|
| 654 |
+
"rewards/chosen": -0.049895767122507095,
|
| 655 |
+
"rewards/margins": 0.72125244140625,
|
| 656 |
+
"rewards/rejected": -0.7711483240127563,
|
| 657 |
+
"step": 390
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 0.4187385501177702,
|
| 661 |
+
"grad_norm": 5.40625,
|
| 662 |
+
"learning_rate": 3.607600562872785e-06,
|
| 663 |
+
"logits/chosen": -0.5524119138717651,
|
| 664 |
+
"logits/rejected": -0.5214421153068542,
|
| 665 |
+
"logps/chosen": -284.4479675292969,
|
| 666 |
+
"logps/rejected": -274.91790771484375,
|
| 667 |
+
"loss": 0.5124,
|
| 668 |
+
"rewards/accuracies": 0.71875,
|
| 669 |
+
"rewards/chosen": -0.08701231330633163,
|
| 670 |
+
"rewards/margins": 0.7274686694145203,
|
| 671 |
+
"rewards/rejected": -0.8144810795783997,
|
| 672 |
+
"step": 400
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 0.4187385501177702,
|
| 676 |
+
"eval_logits/chosen": -0.45163384079933167,
|
| 677 |
+
"eval_logits/rejected": -0.42426639795303345,
|
| 678 |
+
"eval_logps/chosen": -292.09857177734375,
|
| 679 |
+
"eval_logps/rejected": -284.3532409667969,
|
| 680 |
+
"eval_loss": 0.5341187715530396,
|
| 681 |
+
"eval_rewards/accuracies": 0.7039999961853027,
|
| 682 |
+
"eval_rewards/chosen": -0.23370328545570374,
|
| 683 |
+
"eval_rewards/margins": 0.7558707594871521,
|
| 684 |
+
"eval_rewards/rejected": -0.9895740151405334,
|
| 685 |
+
"eval_runtime": 349.3488,
|
| 686 |
+
"eval_samples_per_second": 5.725,
|
| 687 |
+
"eval_steps_per_second": 0.358,
|
| 688 |
+
"step": 400
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.42920701387071447,
|
| 692 |
+
"grad_norm": 5.1875,
|
| 693 |
+
"learning_rate": 3.5249095128531863e-06,
|
| 694 |
+
"logits/chosen": -0.5735105872154236,
|
| 695 |
+
"logits/rejected": -0.5243286490440369,
|
| 696 |
+
"logps/chosen": -278.07611083984375,
|
| 697 |
+
"logps/rejected": -275.5641784667969,
|
| 698 |
+
"loss": 0.5147,
|
| 699 |
+
"rewards/accuracies": 0.75,
|
| 700 |
+
"rewards/chosen": -0.19002243876457214,
|
| 701 |
+
"rewards/margins": 0.7818821668624878,
|
| 702 |
+
"rewards/rejected": -0.9719046354293823,
|
| 703 |
+
"step": 410
|
| 704 |
+
},
|
| 705 |
+
{
|
| 706 |
+
"epoch": 0.4396754776236587,
|
| 707 |
+
"grad_norm": 5.75,
|
| 708 |
+
"learning_rate": 3.4408477372034743e-06,
|
| 709 |
+
"logits/chosen": -0.5684117078781128,
|
| 710 |
+
"logits/rejected": -0.5453870296478271,
|
| 711 |
+
"logps/chosen": -308.50128173828125,
|
| 712 |
+
"logps/rejected": -296.2371826171875,
|
| 713 |
+
"loss": 0.5608,
|
| 714 |
+
"rewards/accuracies": 0.768750011920929,
|
| 715 |
+
"rewards/chosen": -0.22137188911437988,
|
| 716 |
+
"rewards/margins": 0.63337242603302,
|
| 717 |
+
"rewards/rejected": -0.8547442555427551,
|
| 718 |
+
"step": 420
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 0.45014394137660296,
|
| 722 |
+
"grad_norm": 6.125,
|
| 723 |
+
"learning_rate": 3.355527661097728e-06,
|
| 724 |
+
"logits/chosen": -0.5637370347976685,
|
| 725 |
+
"logits/rejected": -0.5657171010971069,
|
| 726 |
+
"logps/chosen": -280.50177001953125,
|
| 727 |
+
"logps/rejected": -281.47467041015625,
|
| 728 |
+
"loss": 0.5229,
|
| 729 |
+
"rewards/accuracies": 0.71875,
|
| 730 |
+
"rewards/chosen": -0.39310961961746216,
|
| 731 |
+
"rewards/margins": 0.6110748648643494,
|
| 732 |
+
"rewards/rejected": -1.0041844844818115,
|
| 733 |
+
"step": 430
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 0.46061240512954726,
|
| 737 |
+
"grad_norm": 5.53125,
|
| 738 |
+
"learning_rate": 3.269063392575352e-06,
|
| 739 |
+
"logits/chosen": -0.5321542024612427,
|
| 740 |
+
"logits/rejected": -0.5234124660491943,
|
| 741 |
+
"logps/chosen": -328.39447021484375,
|
| 742 |
+
"logps/rejected": -307.01556396484375,
|
| 743 |
+
"loss": 0.5135,
|
| 744 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 745 |
+
"rewards/chosen": -0.2074432671070099,
|
| 746 |
+
"rewards/margins": 0.7093298435211182,
|
| 747 |
+
"rewards/rejected": -0.9167732000350952,
|
| 748 |
+
"step": 440
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 0.4710808688824915,
|
| 752 |
+
"grad_norm": 6.625,
|
| 753 |
+
"learning_rate": 3.181570569931697e-06,
|
| 754 |
+
"logits/chosen": -0.5547568202018738,
|
| 755 |
+
"logits/rejected": -0.5429819822311401,
|
| 756 |
+
"logps/chosen": -286.77447509765625,
|
| 757 |
+
"logps/rejected": -282.605224609375,
|
| 758 |
+
"loss": 0.5,
|
| 759 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 760 |
+
"rewards/chosen": -0.41850847005844116,
|
| 761 |
+
"rewards/margins": 0.6984038352966309,
|
| 762 |
+
"rewards/rejected": -1.1169124841690063,
|
| 763 |
+
"step": 450
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 0.48154933263543576,
|
| 767 |
+
"grad_norm": 5.125,
|
| 768 |
+
"learning_rate": 3.09316620706208e-06,
|
| 769 |
+
"logits/chosen": -0.5107685327529907,
|
| 770 |
+
"logits/rejected": -0.5204188227653503,
|
| 771 |
+
"logps/chosen": -307.8090515136719,
|
| 772 |
+
"logps/rejected": -288.7105407714844,
|
| 773 |
+
"loss": 0.4984,
|
| 774 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 775 |
+
"rewards/chosen": -0.21866807341575623,
|
| 776 |
+
"rewards/margins": 0.888043999671936,
|
| 777 |
+
"rewards/rejected": -1.106711983680725,
|
| 778 |
+
"step": 460
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 0.49201779638838,
|
| 782 |
+
"grad_norm": 5.3125,
|
| 783 |
+
"learning_rate": 3.0039685369660785e-06,
|
| 784 |
+
"logits/chosen": -0.5120314359664917,
|
| 785 |
+
"logits/rejected": -0.46821707487106323,
|
| 786 |
+
"logps/chosen": -281.59722900390625,
|
| 787 |
+
"logps/rejected": -266.38134765625,
|
| 788 |
+
"loss": 0.5331,
|
| 789 |
+
"rewards/accuracies": 0.699999988079071,
|
| 790 |
+
"rewards/chosen": -0.03815505653619766,
|
| 791 |
+
"rewards/margins": 0.8301553726196289,
|
| 792 |
+
"rewards/rejected": -0.8683103322982788,
|
| 793 |
+
"step": 470
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.5024862601413242,
|
| 797 |
+
"grad_norm": 5.53125,
|
| 798 |
+
"learning_rate": 2.91409685362137e-06,
|
| 799 |
+
"logits/chosen": -0.5207866430282593,
|
| 800 |
+
"logits/rejected": -0.5130727887153625,
|
| 801 |
+
"logps/chosen": -278.73504638671875,
|
| 802 |
+
"logps/rejected": -276.3402404785156,
|
| 803 |
+
"loss": 0.5098,
|
| 804 |
+
"rewards/accuracies": 0.71875,
|
| 805 |
+
"rewards/chosen": -0.15987971425056458,
|
| 806 |
+
"rewards/margins": 0.769777774810791,
|
| 807 |
+
"rewards/rejected": -0.9296576380729675,
|
| 808 |
+
"step": 480
|
| 809 |
+
},
|
| 810 |
+
{
|
| 811 |
+
"epoch": 0.5129547238942685,
|
| 812 |
+
"grad_norm": 4.9375,
|
| 813 |
+
"learning_rate": 2.8236713524386085e-06,
|
| 814 |
+
"logits/chosen": -0.5813151597976685,
|
| 815 |
+
"logits/rejected": -0.5460097193717957,
|
| 816 |
+
"logps/chosen": -278.7942810058594,
|
| 817 |
+
"logps/rejected": -256.22735595703125,
|
| 818 |
+
"loss": 0.498,
|
| 819 |
+
"rewards/accuracies": 0.731249988079071,
|
| 820 |
+
"rewards/chosen": -0.02987014129757881,
|
| 821 |
+
"rewards/margins": 0.7739948034286499,
|
| 822 |
+
"rewards/rejected": -0.8038649559020996,
|
| 823 |
+
"step": 490
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 0.5234231876472127,
|
| 827 |
+
"grad_norm": 5.59375,
|
| 828 |
+
"learning_rate": 2.7328129695107205e-06,
|
| 829 |
+
"logits/chosen": -0.4917908310890198,
|
| 830 |
+
"logits/rejected": -0.5057145357131958,
|
| 831 |
+
"logps/chosen": -264.3317565917969,
|
| 832 |
+
"logps/rejected": -273.6480712890625,
|
| 833 |
+
"loss": 0.5529,
|
| 834 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 835 |
+
"rewards/chosen": -0.05067148059606552,
|
| 836 |
+
"rewards/margins": 0.9685169458389282,
|
| 837 |
+
"rewards/rejected": -1.0191885232925415,
|
| 838 |
+
"step": 500
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 0.5234231876472127,
|
| 842 |
+
"eval_logits/chosen": -0.4290330708026886,
|
| 843 |
+
"eval_logits/rejected": -0.3994762897491455,
|
| 844 |
+
"eval_logps/chosen": -291.93804931640625,
|
| 845 |
+
"eval_logps/rejected": -284.49481201171875,
|
| 846 |
+
"eval_loss": 0.5259639620780945,
|
| 847 |
+
"eval_rewards/accuracies": 0.7239999771118164,
|
| 848 |
+
"eval_rewards/chosen": -0.2176663726568222,
|
| 849 |
+
"eval_rewards/margins": 0.7860648036003113,
|
| 850 |
+
"eval_rewards/rejected": -1.0037312507629395,
|
| 851 |
+
"eval_runtime": 349.4608,
|
| 852 |
+
"eval_samples_per_second": 5.723,
|
| 853 |
+
"eval_steps_per_second": 0.358,
|
| 854 |
+
"step": 500
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 0.533891651400157,
|
| 858 |
+
"grad_norm": 4.75,
|
| 859 |
+
"learning_rate": 2.641643219871597e-06,
|
| 860 |
+
"logits/chosen": -0.5275259017944336,
|
| 861 |
+
"logits/rejected": -0.49047979712486267,
|
| 862 |
+
"logps/chosen": -313.4814147949219,
|
| 863 |
+
"logps/rejected": -297.6827697753906,
|
| 864 |
+
"loss": 0.5085,
|
| 865 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 866 |
+
"rewards/chosen": -0.3123074471950531,
|
| 867 |
+
"rewards/margins": 0.7204921841621399,
|
| 868 |
+
"rewards/rejected": -1.0327996015548706,
|
| 869 |
+
"step": 510
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 0.5443601151531012,
|
| 873 |
+
"grad_norm": 6.90625,
|
| 874 |
+
"learning_rate": 2.5502840349805074e-06,
|
| 875 |
+
"logits/chosen": -0.5200182199478149,
|
| 876 |
+
"logits/rejected": -0.5098154544830322,
|
| 877 |
+
"logps/chosen": -310.73944091796875,
|
| 878 |
+
"logps/rejected": -297.84326171875,
|
| 879 |
+
"loss": 0.5471,
|
| 880 |
+
"rewards/accuracies": 0.706250011920929,
|
| 881 |
+
"rewards/chosen": -0.353024423122406,
|
| 882 |
+
"rewards/margins": 0.8153821229934692,
|
| 883 |
+
"rewards/rejected": -1.1684064865112305,
|
| 884 |
+
"step": 520
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.5548285789060455,
|
| 888 |
+
"grad_norm": 5.03125,
|
| 889 |
+
"learning_rate": 2.4588575996495797e-06,
|
| 890 |
+
"logits/chosen": -0.47494420409202576,
|
| 891 |
+
"logits/rejected": -0.48031002283096313,
|
| 892 |
+
"logps/chosen": -272.15692138671875,
|
| 893 |
+
"logps/rejected": -262.4747009277344,
|
| 894 |
+
"loss": 0.5205,
|
| 895 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 896 |
+
"rewards/chosen": -0.41294246912002563,
|
| 897 |
+
"rewards/margins": 0.8126422166824341,
|
| 898 |
+
"rewards/rejected": -1.225584626197815,
|
| 899 |
+
"step": 530
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 0.5652970426589898,
|
| 903 |
+
"grad_norm": 5.53125,
|
| 904 |
+
"learning_rate": 2.367486188632446e-06,
|
| 905 |
+
"logits/chosen": -0.5167185068130493,
|
| 906 |
+
"logits/rejected": -0.514365017414093,
|
| 907 |
+
"logps/chosen": -285.01080322265625,
|
| 908 |
+
"logps/rejected": -324.41302490234375,
|
| 909 |
+
"loss": 0.5084,
|
| 910 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 911 |
+
"rewards/chosen": -0.18025778234004974,
|
| 912 |
+
"rewards/margins": 0.8430238962173462,
|
| 913 |
+
"rewards/rejected": -1.0232816934585571,
|
| 914 |
+
"step": 540
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 0.575765506411934,
|
| 918 |
+
"grad_norm": 4.1875,
|
| 919 |
+
"learning_rate": 2.276292003092593e-06,
|
| 920 |
+
"logits/chosen": -0.5328477621078491,
|
| 921 |
+
"logits/rejected": -0.514901340007782,
|
| 922 |
+
"logps/chosen": -257.83953857421875,
|
| 923 |
+
"logps/rejected": -265.4601135253906,
|
| 924 |
+
"loss": 0.4871,
|
| 925 |
+
"rewards/accuracies": 0.768750011920929,
|
| 926 |
+
"rewards/chosen": -0.24479135870933533,
|
| 927 |
+
"rewards/margins": 0.8982963562011719,
|
| 928 |
+
"rewards/rejected": -1.1430877447128296,
|
| 929 |
+
"step": 550
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 0.5862339701648783,
|
| 933 |
+
"grad_norm": 4.875,
|
| 934 |
+
"learning_rate": 2.1853970071701415e-06,
|
| 935 |
+
"logits/chosen": -0.5274088978767395,
|
| 936 |
+
"logits/rejected": -0.4985182285308838,
|
| 937 |
+
"logps/chosen": -278.2223815917969,
|
| 938 |
+
"logps/rejected": -279.3916931152344,
|
| 939 |
+
"loss": 0.5206,
|
| 940 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 941 |
+
"rewards/chosen": -0.12979185581207275,
|
| 942 |
+
"rewards/margins": 0.8115113973617554,
|
| 943 |
+
"rewards/rejected": -0.9413032531738281,
|
| 944 |
+
"step": 560
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 0.5967024339178225,
|
| 948 |
+
"grad_norm": 5.84375,
|
| 949 |
+
"learning_rate": 2.0949227648656194e-06,
|
| 950 |
+
"logits/chosen": -0.5541775822639465,
|
| 951 |
+
"logits/rejected": -0.5307848453521729,
|
| 952 |
+
"logps/chosen": -294.7757263183594,
|
| 953 |
+
"logps/rejected": -261.8687438964844,
|
| 954 |
+
"loss": 0.5405,
|
| 955 |
+
"rewards/accuracies": 0.731249988079071,
|
| 956 |
+
"rewards/chosen": -0.2242761105298996,
|
| 957 |
+
"rewards/margins": 0.8702728152275085,
|
| 958 |
+
"rewards/rejected": -1.0945489406585693,
|
| 959 |
+
"step": 570
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 0.6071708976707668,
|
| 963 |
+
"grad_norm": 5.78125,
|
| 964 |
+
"learning_rate": 2.00499027745888e-06,
|
| 965 |
+
"logits/chosen": -0.5282704830169678,
|
| 966 |
+
"logits/rejected": -0.5080757141113281,
|
| 967 |
+
"logps/chosen": -299.1280212402344,
|
| 968 |
+
"logps/rejected": -297.7095031738281,
|
| 969 |
+
"loss": 0.5509,
|
| 970 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 971 |
+
"rewards/chosen": -0.1550193578004837,
|
| 972 |
+
"rewards/margins": 0.7501784563064575,
|
| 973 |
+
"rewards/rejected": -0.9051979184150696,
|
| 974 |
+
"step": 580
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 0.6176393614237111,
|
| 978 |
+
"grad_norm": 5.65625,
|
| 979 |
+
"learning_rate": 1.915719821680624e-06,
|
| 980 |
+
"logits/chosen": -0.5544322729110718,
|
| 981 |
+
"logits/rejected": -0.5079981088638306,
|
| 982 |
+
"logps/chosen": -285.83575439453125,
|
| 983 |
+
"logps/rejected": -281.488037109375,
|
| 984 |
+
"loss": 0.5198,
|
| 985 |
+
"rewards/accuracies": 0.793749988079071,
|
| 986 |
+
"rewards/chosen": 0.1644931137561798,
|
| 987 |
+
"rewards/margins": 0.8637059330940247,
|
| 988 |
+
"rewards/rejected": -0.6992128491401672,
|
| 989 |
+
"step": 590
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 0.6281078251766553,
|
| 993 |
+
"grad_norm": 5.34375,
|
| 994 |
+
"learning_rate": 1.8272307888529276e-06,
|
| 995 |
+
"logits/chosen": -0.47845038771629333,
|
| 996 |
+
"logits/rejected": -0.44052690267562866,
|
| 997 |
+
"logps/chosen": -261.1684265136719,
|
| 998 |
+
"logps/rejected": -279.1067810058594,
|
| 999 |
+
"loss": 0.53,
|
| 1000 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 1001 |
+
"rewards/chosen": -0.013519972562789917,
|
| 1002 |
+
"rewards/margins": 0.8593659400939941,
|
| 1003 |
+
"rewards/rejected": -0.8728858828544617,
|
| 1004 |
+
"step": 600
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.6281078251766553,
|
| 1008 |
+
"eval_logits/chosen": -0.4317292273044586,
|
| 1009 |
+
"eval_logits/rejected": -0.4027506709098816,
|
| 1010 |
+
"eval_logps/chosen": -290.4488830566406,
|
| 1011 |
+
"eval_logps/rejected": -283.0403137207031,
|
| 1012 |
+
"eval_loss": 0.5244117379188538,
|
| 1013 |
+
"eval_rewards/accuracies": 0.7200000286102295,
|
| 1014 |
+
"eval_rewards/chosen": -0.06874838471412659,
|
| 1015 |
+
"eval_rewards/margins": 0.7895320653915405,
|
| 1016 |
+
"eval_rewards/rejected": -0.8582804203033447,
|
| 1017 |
+
"eval_runtime": 349.48,
|
| 1018 |
+
"eval_samples_per_second": 5.723,
|
| 1019 |
+
"eval_steps_per_second": 0.358,
|
| 1020 |
+
"step": 600
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 0.6385762889295996,
|
| 1024 |
+
"grad_norm": 4.625,
|
| 1025 |
+
"learning_rate": 1.739641525213929e-06,
|
| 1026 |
+
"logits/chosen": -0.510543942451477,
|
| 1027 |
+
"logits/rejected": -0.5007991194725037,
|
| 1028 |
+
"logps/chosen": -264.67572021484375,
|
| 1029 |
+
"logps/rejected": -270.05438232421875,
|
| 1030 |
+
"loss": 0.4957,
|
| 1031 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1032 |
+
"rewards/chosen": -0.04250194877386093,
|
| 1033 |
+
"rewards/margins": 0.8683522343635559,
|
| 1034 |
+
"rewards/rejected": -0.9108540415763855,
|
| 1035 |
+
"step": 610
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 0.6490447526825438,
|
| 1039 |
+
"grad_norm": 4.5625,
|
| 1040 |
+
"learning_rate": 1.6530691736402317e-06,
|
| 1041 |
+
"logits/chosen": -0.5225728750228882,
|
| 1042 |
+
"logits/rejected": -0.49780869483947754,
|
| 1043 |
+
"logps/chosen": -291.7324523925781,
|
| 1044 |
+
"logps/rejected": -280.8247985839844,
|
| 1045 |
+
"loss": 0.5072,
|
| 1046 |
+
"rewards/accuracies": 0.7875000238418579,
|
| 1047 |
+
"rewards/chosen": -0.15649111568927765,
|
| 1048 |
+
"rewards/margins": 0.8370261192321777,
|
| 1049 |
+
"rewards/rejected": -0.9935172200202942,
|
| 1050 |
+
"step": 620
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 0.6595132164354881,
|
| 1054 |
+
"grad_norm": 5.34375,
|
| 1055 |
+
"learning_rate": 1.5676295169786864e-06,
|
| 1056 |
+
"logits/chosen": -0.5341150164604187,
|
| 1057 |
+
"logits/rejected": -0.4966842532157898,
|
| 1058 |
+
"logps/chosen": -283.0377502441406,
|
| 1059 |
+
"logps/rejected": -271.4356994628906,
|
| 1060 |
+
"loss": 0.5294,
|
| 1061 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1062 |
+
"rewards/chosen": -0.335578590631485,
|
| 1063 |
+
"rewards/margins": 0.8176633715629578,
|
| 1064 |
+
"rewards/rejected": -1.1532419919967651,
|
| 1065 |
+
"step": 630
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 0.6699816801884323,
|
| 1069 |
+
"grad_norm": 5.25,
|
| 1070 |
+
"learning_rate": 1.4834368231970922e-06,
|
| 1071 |
+
"logits/chosen": -0.5592583417892456,
|
| 1072 |
+
"logits/rejected": -0.5109054446220398,
|
| 1073 |
+
"logps/chosen": -287.51068115234375,
|
| 1074 |
+
"logps/rejected": -272.8183288574219,
|
| 1075 |
+
"loss": 0.5,
|
| 1076 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1077 |
+
"rewards/chosen": -0.3240619897842407,
|
| 1078 |
+
"rewards/margins": 0.7095167636871338,
|
| 1079 |
+
"rewards/rejected": -1.0335787534713745,
|
| 1080 |
+
"step": 640
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 0.6804501439413766,
|
| 1084 |
+
"grad_norm": 5.28125,
|
| 1085 |
+
"learning_rate": 1.4006036925609245e-06,
|
| 1086 |
+
"logits/chosen": -0.5300047993659973,
|
| 1087 |
+
"logits/rejected": -0.49794530868530273,
|
| 1088 |
+
"logps/chosen": -300.1281433105469,
|
| 1089 |
+
"logps/rejected": -249.81936645507812,
|
| 1090 |
+
"loss": 0.5413,
|
| 1091 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1092 |
+
"rewards/chosen": -0.32234618067741394,
|
| 1093 |
+
"rewards/margins": 0.8122032880783081,
|
| 1094 |
+
"rewards/rejected": -1.1345494985580444,
|
| 1095 |
+
"step": 650
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 0.6909186076943209,
|
| 1099 |
+
"grad_norm": 5.15625,
|
| 1100 |
+
"learning_rate": 1.3192409070404582e-06,
|
| 1101 |
+
"logits/chosen": -0.550862193107605,
|
| 1102 |
+
"logits/rejected": -0.532629668712616,
|
| 1103 |
+
"logps/chosen": -303.74505615234375,
|
| 1104 |
+
"logps/rejected": -305.6463928222656,
|
| 1105 |
+
"loss": 0.5143,
|
| 1106 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1107 |
+
"rewards/chosen": -0.20528188347816467,
|
| 1108 |
+
"rewards/margins": 0.7565289735794067,
|
| 1109 |
+
"rewards/rejected": -0.9618108868598938,
|
| 1110 |
+
"step": 660
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 0.7013870714472651,
|
| 1114 |
+
"grad_norm": 6.59375,
|
| 1115 |
+
"learning_rate": 1.2394572821496953e-06,
|
| 1116 |
+
"logits/chosen": -0.5544015169143677,
|
| 1117 |
+
"logits/rejected": -0.5221393704414368,
|
| 1118 |
+
"logps/chosen": -277.2763977050781,
|
| 1119 |
+
"logps/rejected": -258.49676513671875,
|
| 1120 |
+
"loss": 0.5181,
|
| 1121 |
+
"rewards/accuracies": 0.71875,
|
| 1122 |
+
"rewards/chosen": -0.20379216969013214,
|
| 1123 |
+
"rewards/margins": 0.7780588269233704,
|
| 1124 |
+
"rewards/rejected": -0.9818509817123413,
|
| 1125 |
+
"step": 670
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 0.7118555352002094,
|
| 1129 |
+
"grad_norm": 5.09375,
|
| 1130 |
+
"learning_rate": 1.1613595214152713e-06,
|
| 1131 |
+
"logits/chosen": -0.5822625160217285,
|
| 1132 |
+
"logits/rejected": -0.540503203868866,
|
| 1133 |
+
"logps/chosen": -286.454833984375,
|
| 1134 |
+
"logps/rejected": -275.0696105957031,
|
| 1135 |
+
"loss": 0.5043,
|
| 1136 |
+
"rewards/accuracies": 0.7562500238418579,
|
| 1137 |
+
"rewards/chosen": -0.32862305641174316,
|
| 1138 |
+
"rewards/margins": 0.7822999954223633,
|
| 1139 |
+
"rewards/rejected": -1.1109230518341064,
|
| 1140 |
+
"step": 680
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 0.7223239989531536,
|
| 1144 |
+
"grad_norm": 4.3125,
|
| 1145 |
+
"learning_rate": 1.0850520736699362e-06,
|
| 1146 |
+
"logits/chosen": -0.5231102705001831,
|
| 1147 |
+
"logits/rejected": -0.49847880005836487,
|
| 1148 |
+
"logps/chosen": -340.5146484375,
|
| 1149 |
+
"logps/rejected": -315.625,
|
| 1150 |
+
"loss": 0.5348,
|
| 1151 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1152 |
+
"rewards/chosen": -0.29148069024086,
|
| 1153 |
+
"rewards/margins": 0.8814651370048523,
|
| 1154 |
+
"rewards/rejected": -1.1729459762573242,
|
| 1155 |
+
"step": 690
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 0.7327924627060979,
|
| 1159 |
+
"grad_norm": 5.625,
|
| 1160 |
+
"learning_rate": 1.0106369933615043e-06,
|
| 1161 |
+
"logits/chosen": -0.5652047991752625,
|
| 1162 |
+
"logits/rejected": -0.5295856595039368,
|
| 1163 |
+
"logps/chosen": -315.13458251953125,
|
| 1164 |
+
"logps/rejected": -262.6495361328125,
|
| 1165 |
+
"loss": 0.5028,
|
| 1166 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1167 |
+
"rewards/chosen": -0.3104090690612793,
|
| 1168 |
+
"rewards/margins": 0.7127448320388794,
|
| 1169 |
+
"rewards/rejected": -1.0231537818908691,
|
| 1170 |
+
"step": 700
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 0.7327924627060979,
|
| 1174 |
+
"eval_logits/chosen": -0.417924702167511,
|
| 1175 |
+
"eval_logits/rejected": -0.38744810223579407,
|
| 1176 |
+
"eval_logps/chosen": -293.1184387207031,
|
| 1177 |
+
"eval_logps/rejected": -285.8177185058594,
|
| 1178 |
+
"eval_loss": 0.5190439224243164,
|
| 1179 |
+
"eval_rewards/accuracies": 0.7319999933242798,
|
| 1180 |
+
"eval_rewards/chosen": -0.33567702770233154,
|
| 1181 |
+
"eval_rewards/margins": 0.8003441095352173,
|
| 1182 |
+
"eval_rewards/rejected": -1.1360211372375488,
|
| 1183 |
+
"eval_runtime": 349.5304,
|
| 1184 |
+
"eval_samples_per_second": 5.722,
|
| 1185 |
+
"eval_steps_per_second": 0.358,
|
| 1186 |
+
"step": 700
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 0.7432609264590422,
|
| 1190 |
+
"grad_norm": 6.1875,
|
| 1191 |
+
"learning_rate": 9.382138040640714e-07,
|
| 1192 |
+
"logits/chosen": -0.5737341046333313,
|
| 1193 |
+
"logits/rejected": -0.5333853363990784,
|
| 1194 |
+
"logps/chosen": -264.9908752441406,
|
| 1195 |
+
"logps/rejected": -278.7557067871094,
|
| 1196 |
+
"loss": 0.5569,
|
| 1197 |
+
"rewards/accuracies": 0.7250000238418579,
|
| 1198 |
+
"rewards/chosen": -0.3085269331932068,
|
| 1199 |
+
"rewards/margins": 0.818738579750061,
|
| 1200 |
+
"rewards/rejected": -1.1272655725479126,
|
| 1201 |
+
"step": 710
|
| 1202 |
+
},
|
| 1203 |
+
{
|
| 1204 |
+
"epoch": 0.7537293902119864,
|
| 1205 |
+
"grad_norm": 5.34375,
|
| 1206 |
+
"learning_rate": 8.678793653740633e-07,
|
| 1207 |
+
"logits/chosen": -0.4978073537349701,
|
| 1208 |
+
"logits/rejected": -0.49744582176208496,
|
| 1209 |
+
"logps/chosen": -263.29010009765625,
|
| 1210 |
+
"logps/rejected": -263.02618408203125,
|
| 1211 |
+
"loss": 0.5204,
|
| 1212 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1213 |
+
"rewards/chosen": -0.2587752938270569,
|
| 1214 |
+
"rewards/margins": 0.8211803436279297,
|
| 1215 |
+
"rewards/rejected": -1.0799555778503418,
|
| 1216 |
+
"step": 720
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 0.7641978539649307,
|
| 1220 |
+
"grad_norm": 4.75,
|
| 1221 |
+
"learning_rate": 7.997277433690984e-07,
|
| 1222 |
+
"logits/chosen": -0.5180292725563049,
|
| 1223 |
+
"logits/rejected": -0.47244367003440857,
|
| 1224 |
+
"logps/chosen": -301.2723693847656,
|
| 1225 |
+
"logps/rejected": -287.05584716796875,
|
| 1226 |
+
"loss": 0.5153,
|
| 1227 |
+
"rewards/accuracies": 0.800000011920929,
|
| 1228 |
+
"rewards/chosen": -0.28081485629081726,
|
| 1229 |
+
"rewards/margins": 0.7739724516868591,
|
| 1230 |
+
"rewards/rejected": -1.054787278175354,
|
| 1231 |
+
"step": 730
|
| 1232 |
+
},
|
| 1233 |
+
{
|
| 1234 |
+
"epoch": 0.7746663177178749,
|
| 1235 |
+
"grad_norm": 4.65625,
|
| 1236 |
+
"learning_rate": 7.338500848029603e-07,
|
| 1237 |
+
"logits/chosen": -0.47029241919517517,
|
| 1238 |
+
"logits/rejected": -0.48912668228149414,
|
| 1239 |
+
"logps/chosen": -290.9686584472656,
|
| 1240 |
+
"logps/rejected": -274.376953125,
|
| 1241 |
+
"loss": 0.4939,
|
| 1242 |
+
"rewards/accuracies": 0.768750011920929,
|
| 1243 |
+
"rewards/chosen": -0.0831630602478981,
|
| 1244 |
+
"rewards/margins": 0.7876149415969849,
|
| 1245 |
+
"rewards/rejected": -0.8707780838012695,
|
| 1246 |
+
"step": 740
|
| 1247 |
+
},
|
| 1248 |
+
{
|
| 1249 |
+
"epoch": 0.7851347814708192,
|
| 1250 |
+
"grad_norm": 5.09375,
|
| 1251 |
+
"learning_rate": 6.70334495204884e-07,
|
| 1252 |
+
"logits/chosen": -0.5157723426818848,
|
| 1253 |
+
"logits/rejected": -0.49868693947792053,
|
| 1254 |
+
"logps/chosen": -323.7619323730469,
|
| 1255 |
+
"logps/rejected": -285.82916259765625,
|
| 1256 |
+
"loss": 0.5023,
|
| 1257 |
+
"rewards/accuracies": 0.75,
|
| 1258 |
+
"rewards/chosen": -0.11948978900909424,
|
| 1259 |
+
"rewards/margins": 0.8135945200920105,
|
| 1260 |
+
"rewards/rejected": -0.9330843687057495,
|
| 1261 |
+
"step": 750
|
| 1262 |
+
},
|
| 1263 |
+
{
|
| 1264 |
+
"epoch": 0.7956032452237635,
|
| 1265 |
+
"grad_norm": 5.46875,
|
| 1266 |
+
"learning_rate": 6.092659210462232e-07,
|
| 1267 |
+
"logits/chosen": -0.5377079248428345,
|
| 1268 |
+
"logits/rejected": -0.5193135142326355,
|
| 1269 |
+
"logps/chosen": -268.98455810546875,
|
| 1270 |
+
"logps/rejected": -268.32025146484375,
|
| 1271 |
+
"loss": 0.5455,
|
| 1272 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1273 |
+
"rewards/chosen": -0.2502976357936859,
|
| 1274 |
+
"rewards/margins": 0.698966920375824,
|
| 1275 |
+
"rewards/rejected": -0.949264645576477,
|
| 1276 |
+
"step": 760
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 0.8060717089767077,
|
| 1280 |
+
"grad_norm": 5.28125,
|
| 1281 |
+
"learning_rate": 5.507260361320738e-07,
|
| 1282 |
+
"logits/chosen": -0.5280566215515137,
|
| 1283 |
+
"logits/rejected": -0.5302770137786865,
|
| 1284 |
+
"logps/chosen": -284.72637939453125,
|
| 1285 |
+
"logps/rejected": -279.0828552246094,
|
| 1286 |
+
"loss": 0.5093,
|
| 1287 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1288 |
+
"rewards/chosen": -0.12712079286575317,
|
| 1289 |
+
"rewards/margins": 0.6862561106681824,
|
| 1290 |
+
"rewards/rejected": -0.8133770227432251,
|
| 1291 |
+
"step": 770
|
| 1292 |
+
},
|
| 1293 |
+
{
|
| 1294 |
+
"epoch": 0.816540172729652,
|
| 1295 |
+
"grad_norm": 5.21875,
|
| 1296 |
+
"learning_rate": 4.947931323697983e-07,
|
| 1297 |
+
"logits/chosen": -0.5104556083679199,
|
| 1298 |
+
"logits/rejected": -0.48463624715805054,
|
| 1299 |
+
"logps/chosen": -286.16925048828125,
|
| 1300 |
+
"logps/rejected": -279.46295166015625,
|
| 1301 |
+
"loss": 0.5278,
|
| 1302 |
+
"rewards/accuracies": 0.6875,
|
| 1303 |
+
"rewards/chosen": -0.2269764244556427,
|
| 1304 |
+
"rewards/margins": 0.6587584018707275,
|
| 1305 |
+
"rewards/rejected": -0.8857349157333374,
|
| 1306 |
+
"step": 780
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 0.8270086364825961,
|
| 1310 |
+
"grad_norm": 5.375,
|
| 1311 |
+
"learning_rate": 4.4154201506053985e-07,
|
| 1312 |
+
"logits/chosen": -0.5380119681358337,
|
| 1313 |
+
"logits/rejected": -0.5271115303039551,
|
| 1314 |
+
"logps/chosen": -300.27313232421875,
|
| 1315 |
+
"logps/rejected": -263.89984130859375,
|
| 1316 |
+
"loss": 0.511,
|
| 1317 |
+
"rewards/accuracies": 0.8187500238418579,
|
| 1318 |
+
"rewards/chosen": -0.15483184158802032,
|
| 1319 |
+
"rewards/margins": 0.8044347763061523,
|
| 1320 |
+
"rewards/rejected": -0.9592665433883667,
|
| 1321 |
+
"step": 790
|
| 1322 |
+
},
|
| 1323 |
+
{
|
| 1324 |
+
"epoch": 0.8374771002355405,
|
| 1325 |
+
"grad_norm": 4.84375,
|
| 1326 |
+
"learning_rate": 3.910439028537638e-07,
|
| 1327 |
+
"logits/chosen": -0.5491563677787781,
|
| 1328 |
+
"logits/rejected": -0.5051991939544678,
|
| 1329 |
+
"logps/chosen": -347.644287109375,
|
| 1330 |
+
"logps/rejected": -302.0996398925781,
|
| 1331 |
+
"loss": 0.5347,
|
| 1332 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1333 |
+
"rewards/chosen": -0.08171938359737396,
|
| 1334 |
+
"rewards/margins": 0.7037029266357422,
|
| 1335 |
+
"rewards/rejected": -0.7854223251342773,
|
| 1336 |
+
"step": 800
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 0.8374771002355405,
|
| 1340 |
+
"eval_logits/chosen": -0.42248398065567017,
|
| 1341 |
+
"eval_logits/rejected": -0.39241039752960205,
|
| 1342 |
+
"eval_logps/chosen": -291.1650085449219,
|
| 1343 |
+
"eval_logps/rejected": -283.87603759765625,
|
| 1344 |
+
"eval_loss": 0.5191403031349182,
|
| 1345 |
+
"eval_rewards/accuracies": 0.7319999933242798,
|
| 1346 |
+
"eval_rewards/chosen": -0.14035792648792267,
|
| 1347 |
+
"eval_rewards/margins": 0.8014954924583435,
|
| 1348 |
+
"eval_rewards/rejected": -0.9418535232543945,
|
| 1349 |
+
"eval_runtime": 349.5564,
|
| 1350 |
+
"eval_samples_per_second": 5.722,
|
| 1351 |
+
"eval_steps_per_second": 0.358,
|
| 1352 |
+
"step": 800
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 0.8479455639884846,
|
| 1356 |
+
"grad_norm": 4.6875,
|
| 1357 |
+
"learning_rate": 3.4336633249862084e-07,
|
| 1358 |
+
"logits/chosen": -0.5698053240776062,
|
| 1359 |
+
"logits/rejected": -0.5044723153114319,
|
| 1360 |
+
"logps/chosen": -319.55126953125,
|
| 1361 |
+
"logps/rejected": -288.7781066894531,
|
| 1362 |
+
"loss": 0.5021,
|
| 1363 |
+
"rewards/accuracies": 0.7437499761581421,
|
| 1364 |
+
"rewards/chosen": -0.17804962396621704,
|
| 1365 |
+
"rewards/margins": 0.8044875860214233,
|
| 1366 |
+
"rewards/rejected": -0.9825371503829956,
|
| 1367 |
+
"step": 810
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 0.8584140277414289,
|
| 1371 |
+
"grad_norm": 5.25,
|
| 1372 |
+
"learning_rate": 2.98573068519539e-07,
|
| 1373 |
+
"logits/chosen": -0.5468612909317017,
|
| 1374 |
+
"logits/rejected": -0.5319810509681702,
|
| 1375 |
+
"logps/chosen": -307.5137939453125,
|
| 1376 |
+
"logps/rejected": -293.56561279296875,
|
| 1377 |
+
"loss": 0.5358,
|
| 1378 |
+
"rewards/accuracies": 0.706250011920929,
|
| 1379 |
+
"rewards/chosen": -0.15299567580223083,
|
| 1380 |
+
"rewards/margins": 0.7571539282798767,
|
| 1381 |
+
"rewards/rejected": -0.9101495742797852,
|
| 1382 |
+
"step": 820
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 0.8688824914943732,
|
| 1386 |
+
"grad_norm": 5.375,
|
| 1387 |
+
"learning_rate": 2.5672401793681854e-07,
|
| 1388 |
+
"logits/chosen": -0.565108597278595,
|
| 1389 |
+
"logits/rejected": -0.5492611527442932,
|
| 1390 |
+
"logps/chosen": -274.61260986328125,
|
| 1391 |
+
"logps/rejected": -269.25439453125,
|
| 1392 |
+
"loss": 0.5048,
|
| 1393 |
+
"rewards/accuracies": 0.737500011920929,
|
| 1394 |
+
"rewards/chosen": -0.17146927118301392,
|
| 1395 |
+
"rewards/margins": 0.8092424273490906,
|
| 1396 |
+
"rewards/rejected": -0.9807117581367493,
|
| 1397 |
+
"step": 830
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 0.8793509552473174,
|
| 1401 |
+
"grad_norm": 5.1875,
|
| 1402 |
+
"learning_rate": 2.178751501463036e-07,
|
| 1403 |
+
"logits/chosen": -0.533309280872345,
|
| 1404 |
+
"logits/rejected": -0.5187536478042603,
|
| 1405 |
+
"logps/chosen": -313.8269958496094,
|
| 1406 |
+
"logps/rejected": -307.17987060546875,
|
| 1407 |
+
"loss": 0.517,
|
| 1408 |
+
"rewards/accuracies": 0.6812499761581421,
|
| 1409 |
+
"rewards/chosen": -0.20717601478099823,
|
| 1410 |
+
"rewards/margins": 0.7723677158355713,
|
| 1411 |
+
"rewards/rejected": -0.9795438051223755,
|
| 1412 |
+
"step": 840
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 0.8898194190002617,
|
| 1416 |
+
"grad_norm": 5.4375,
|
| 1417 |
+
"learning_rate": 1.820784220652766e-07,
|
| 1418 |
+
"logits/chosen": -0.5832973122596741,
|
| 1419 |
+
"logits/rejected": -0.5479104518890381,
|
| 1420 |
+
"logps/chosen": -345.9014587402344,
|
| 1421 |
+
"logps/rejected": -279.53985595703125,
|
| 1422 |
+
"loss": 0.5083,
|
| 1423 |
+
"rewards/accuracies": 0.71875,
|
| 1424 |
+
"rewards/chosen": -0.0073674083687365055,
|
| 1425 |
+
"rewards/margins": 0.875013530254364,
|
| 1426 |
+
"rewards/rejected": -0.8823810815811157,
|
| 1427 |
+
"step": 850
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 0.9002878827532059,
|
| 1431 |
+
"grad_norm": 5.375,
|
| 1432 |
+
"learning_rate": 1.4938170864468636e-07,
|
| 1433 |
+
"logits/chosen": -0.5060287117958069,
|
| 1434 |
+
"logits/rejected": -0.4896390438079834,
|
| 1435 |
+
"logps/chosen": -289.37042236328125,
|
| 1436 |
+
"logps/rejected": -268.93487548828125,
|
| 1437 |
+
"loss": 0.4806,
|
| 1438 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1439 |
+
"rewards/chosen": -0.11767210066318512,
|
| 1440 |
+
"rewards/margins": 0.9038281440734863,
|
| 1441 |
+
"rewards/rejected": -1.0215001106262207,
|
| 1442 |
+
"step": 860
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 0.9107563465061502,
|
| 1446 |
+
"grad_norm": 5.53125,
|
| 1447 |
+
"learning_rate": 1.1982873884064466e-07,
|
| 1448 |
+
"logits/chosen": -0.4754942059516907,
|
| 1449 |
+
"logits/rejected": -0.4765293598175049,
|
| 1450 |
+
"logps/chosen": -287.2554931640625,
|
| 1451 |
+
"logps/rejected": -276.70794677734375,
|
| 1452 |
+
"loss": 0.5263,
|
| 1453 |
+
"rewards/accuracies": 0.7124999761581421,
|
| 1454 |
+
"rewards/chosen": -0.16617272794246674,
|
| 1455 |
+
"rewards/margins": 0.6353666186332703,
|
| 1456 |
+
"rewards/rejected": -0.801539421081543,
|
| 1457 |
+
"step": 870
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 0.9212248102590945,
|
| 1461 |
+
"grad_norm": 5.75,
|
| 1462 |
+
"learning_rate": 9.345903713082305e-08,
|
| 1463 |
+
"logits/chosen": -0.5580819845199585,
|
| 1464 |
+
"logits/rejected": -0.5499614477157593,
|
| 1465 |
+
"logps/chosen": -315.15478515625,
|
| 1466 |
+
"logps/rejected": -280.8697814941406,
|
| 1467 |
+
"loss": 0.5463,
|
| 1468 |
+
"rewards/accuracies": 0.625,
|
| 1469 |
+
"rewards/chosen": -0.1284041702747345,
|
| 1470 |
+
"rewards/margins": 0.6124891042709351,
|
| 1471 |
+
"rewards/rejected": -0.7408932447433472,
|
| 1472 |
+
"step": 880
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 0.9316932740120387,
|
| 1476 |
+
"grad_norm": 5.125,
|
| 1477 |
+
"learning_rate": 7.030787065396866e-08,
|
| 1478 |
+
"logits/chosen": -0.5307958722114563,
|
| 1479 |
+
"logits/rejected": -0.49479198455810547,
|
| 1480 |
+
"logps/chosen": -318.7244567871094,
|
| 1481 |
+
"logps/rejected": -293.06219482421875,
|
| 1482 |
+
"loss": 0.5198,
|
| 1483 |
+
"rewards/accuracies": 0.71875,
|
| 1484 |
+
"rewards/chosen": -0.11201741546392441,
|
| 1485 |
+
"rewards/margins": 0.724387526512146,
|
| 1486 |
+
"rewards/rejected": -0.8364049792289734,
|
| 1487 |
+
"step": 890
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 0.942161737764983,
|
| 1491 |
+
"grad_norm": 4.96875,
|
| 1492 |
+
"learning_rate": 5.0406202043228604e-08,
|
| 1493 |
+
"logits/chosen": -0.541701078414917,
|
| 1494 |
+
"logits/rejected": -0.5375146865844727,
|
| 1495 |
+
"logps/chosen": -333.6339111328125,
|
| 1496 |
+
"logps/rejected": -274.9059753417969,
|
| 1497 |
+
"loss": 0.4783,
|
| 1498 |
+
"rewards/accuracies": 0.768750011920929,
|
| 1499 |
+
"rewards/chosen": -0.0002611428499221802,
|
| 1500 |
+
"rewards/margins": 0.9774513244628906,
|
| 1501 |
+
"rewards/rejected": -0.9777124524116516,
|
| 1502 |
+
"step": 900
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 0.942161737764983,
|
| 1506 |
+
"eval_logits/chosen": -0.4219192862510681,
|
| 1507 |
+
"eval_logits/rejected": -0.391748309135437,
|
| 1508 |
+
"eval_logps/chosen": -291.15997314453125,
|
| 1509 |
+
"eval_logps/rejected": -283.916259765625,
|
| 1510 |
+
"eval_loss": 0.5189567804336548,
|
| 1511 |
+
"eval_rewards/accuracies": 0.7260000109672546,
|
| 1512 |
+
"eval_rewards/chosen": -0.13985459506511688,
|
| 1513 |
+
"eval_rewards/margins": 0.8060198426246643,
|
| 1514 |
+
"eval_rewards/rejected": -0.9458745121955872,
|
| 1515 |
+
"eval_runtime": 349.4658,
|
| 1516 |
+
"eval_samples_per_second": 5.723,
|
| 1517 |
+
"eval_steps_per_second": 0.358,
|
| 1518 |
+
"step": 900
|
| 1519 |
+
},
|
| 1520 |
+
{
|
| 1521 |
+
"epoch": 0.9526302015179272,
|
| 1522 |
+
"grad_norm": 5.15625,
|
| 1523 |
+
"learning_rate": 3.378064801637687e-08,
|
| 1524 |
+
"logits/chosen": -0.5818526148796082,
|
| 1525 |
+
"logits/rejected": -0.5350463390350342,
|
| 1526 |
+
"logps/chosen": -314.6712341308594,
|
| 1527 |
+
"logps/rejected": -313.63116455078125,
|
| 1528 |
+
"loss": 0.5208,
|
| 1529 |
+
"rewards/accuracies": 0.731249988079071,
|
| 1530 |
+
"rewards/chosen": -0.047819893807172775,
|
| 1531 |
+
"rewards/margins": 0.7616057991981506,
|
| 1532 |
+
"rewards/rejected": -0.8094256520271301,
|
| 1533 |
+
"step": 910
|
| 1534 |
+
},
|
| 1535 |
+
{
|
| 1536 |
+
"epoch": 0.9630986652708715,
|
| 1537 |
+
"grad_norm": 4.6875,
|
| 1538 |
+
"learning_rate": 2.0453443778310766e-08,
|
| 1539 |
+
"logits/chosen": -0.5121539831161499,
|
| 1540 |
+
"logits/rejected": -0.48242586851119995,
|
| 1541 |
+
"logps/chosen": -329.2786865234375,
|
| 1542 |
+
"logps/rejected": -304.5957336425781,
|
| 1543 |
+
"loss": 0.5356,
|
| 1544 |
+
"rewards/accuracies": 0.768750011920929,
|
| 1545 |
+
"rewards/chosen": -0.06458105146884918,
|
| 1546 |
+
"rewards/margins": 0.869833767414093,
|
| 1547 |
+
"rewards/rejected": -0.9344147443771362,
|
| 1548 |
+
"step": 920
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 0.9735671290238157,
|
| 1552 |
+
"grad_norm": 4.78125,
|
| 1553 |
+
"learning_rate": 1.0442413283435759e-08,
|
| 1554 |
+
"logits/chosen": -0.509577751159668,
|
| 1555 |
+
"logits/rejected": -0.47219425439834595,
|
| 1556 |
+
"logps/chosen": -317.72943115234375,
|
| 1557 |
+
"logps/rejected": -278.4623718261719,
|
| 1558 |
+
"loss": 0.5143,
|
| 1559 |
+
"rewards/accuracies": 0.762499988079071,
|
| 1560 |
+
"rewards/chosen": -0.05142117291688919,
|
| 1561 |
+
"rewards/margins": 1.1160078048706055,
|
| 1562 |
+
"rewards/rejected": -1.167428970336914,
|
| 1563 |
+
"step": 930
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 0.98403559277676,
|
| 1567 |
+
"grad_norm": 4.1875,
|
| 1568 |
+
"learning_rate": 3.760945397705828e-09,
|
| 1569 |
+
"logits/chosen": -0.5354495048522949,
|
| 1570 |
+
"logits/rejected": -0.4856652617454529,
|
| 1571 |
+
"logps/chosen": -290.7127380371094,
|
| 1572 |
+
"logps/rejected": -260.788330078125,
|
| 1573 |
+
"loss": 0.487,
|
| 1574 |
+
"rewards/accuracies": 0.8125,
|
| 1575 |
+
"rewards/chosen": 0.03208203241229057,
|
| 1576 |
+
"rewards/margins": 1.0025131702423096,
|
| 1577 |
+
"rewards/rejected": -0.9704310297966003,
|
| 1578 |
+
"step": 940
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 0.9945040565297043,
|
| 1582 |
+
"grad_norm": 4.6875,
|
| 1583 |
+
"learning_rate": 4.1797599220405605e-10,
|
| 1584 |
+
"logits/chosen": -0.5512427687644958,
|
| 1585 |
+
"logits/rejected": -0.5348908305168152,
|
| 1586 |
+
"logps/chosen": -295.21319580078125,
|
| 1587 |
+
"logps/rejected": -278.82257080078125,
|
| 1588 |
+
"loss": 0.5053,
|
| 1589 |
+
"rewards/accuracies": 0.7749999761581421,
|
| 1590 |
+
"rewards/chosen": -0.0824163556098938,
|
| 1591 |
+
"rewards/margins": 0.8457223176956177,
|
| 1592 |
+
"rewards/rejected": -0.9281387329101562,
|
| 1593 |
+
"step": 950
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 0.9997382884061764,
|
| 1597 |
+
"step": 955,
|
| 1598 |
+
"total_flos": 0.0,
|
| 1599 |
+
"train_loss": 0.543355998318857,
|
| 1600 |
+
"train_runtime": 19333.2554,
|
| 1601 |
+
"train_samples_per_second": 3.162,
|
| 1602 |
+
"train_steps_per_second": 0.049
|
| 1603 |
+
}
|
| 1604 |
+
],
|
| 1605 |
+
"logging_steps": 10,
|
| 1606 |
+
"max_steps": 955,
|
| 1607 |
+
"num_input_tokens_seen": 0,
|
| 1608 |
+
"num_train_epochs": 1,
|
| 1609 |
+
"save_steps": 100000,
|
| 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": 1,
|
| 1624 |
+
"trial_name": null,
|
| 1625 |
+
"trial_params": null
|
| 1626 |
+
}
|