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: 1.9448
- Rewards/chosen: 0.2004
- Rewards/rejected: 1.1728
- Rewards/accuracies: 0.3984
- Rewards/margins: -0.9724
- Logps/rejected: -516.0420
- Logps/chosen: -390.0737
- Logits/rejected: -4.6999
- Logits/chosen: -4.4838
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: 1e-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 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.1833 | 0.26 | 100 | 1.8369 | 0.4269 | 1.4521 | 0.375 | -1.0251 | -515.4835 | -389.6206 | -4.7248 | -4.5091 |
| 0.1786 | 0.51 | 200 | 2.0163 | 0.6049 | 1.7456 | 0.375 | -1.1407 | -514.8965 | -389.2646 | -4.6879 | -4.4698 |
| 0.1648 | 0.77 | 300 | 1.9448 | 0.2004 | 1.1728 | 0.3984 | -0.9724 | -516.0420 | -390.0737 | -4.6999 | -4.4838 |
Framework versions
- Transformers 4.38.2
- Pytorch 2.1.2+cu118
- Datasets 2.16.1
- Tokenizers 0.15.2