metadata
tags:
- trl
- dpo
- generated_from_trainer
model-index:
- name: zephyr-7b-dpo-full
results: []
zephyr-7b-dpo-full
This model was trained from scratch on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.1592
- Rewards/chosen: 8.1420
- Rewards/rejected: -16.5137
- Rewards/accuracies: 0.9219
- Rewards/margins: 24.6558
- Logps/rejected: -126.2682
- Logps/chosen: -121.4882
- Logits/rejected: -1.8539
- Logits/chosen: -1.8512
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-07
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 8
- gradient_accumulation_steps: 2
- total_train_batch_size: 128
- total_eval_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 1
Training results
Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
---|---|---|---|---|---|---|---|---|---|---|---|
0.1483 | 0.21 | 100 | 0.1321 | 6.9750 | -14.0592 | 0.9336 | 21.0342 | -121.3591 | -123.8222 | -1.7705 | -1.7813 |
0.2084 | 0.42 | 200 | 0.1510 | 8.5839 | -14.8837 | 0.9375 | 23.4677 | -123.0082 | -120.6044 | -1.8622 | -1.8630 |
0.2072 | 0.63 | 300 | 0.1658 | 7.7715 | -16.5062 | 0.9219 | 24.2778 | -126.2532 | -122.2292 | -1.8451 | -1.8438 |
0.1918 | 0.84 | 400 | 0.1592 | 8.1420 | -16.5137 | 0.9219 | 24.6558 | -126.2682 | -121.4882 | -1.8539 | -1.8512 |
Framework versions
- Transformers 4.38.2
- Pytorch 2.1.2+cu118
- Datasets 2.16.1
- Tokenizers 0.15.2