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 an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.8003
- Rewards/chosen: -1.8897
- Rewards/rejected: -2.0004
- Rewards/accuracies: 0.5273
- Rewards/margins: 0.1107
- Logps/rejected: -718.4238
- Logps/chosen: -579.4417
- Logits/rejected: -5.6556
- Logits/chosen: -5.3947
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 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.3413 | 0.2558 | 100 | 0.7230 | -0.5409 | -0.5757 | 0.5156 | 0.0348 | -575.9554 | -444.5646 | -5.0451 | -4.8217 |
| 0.2653 | 0.5115 | 200 | 0.7765 | -1.4996 | -1.6149 | 0.5430 | 0.1153 | -679.8810 | -540.4390 | -5.5042 | -5.2262 |
| 0.2424 | 0.7673 | 300 | 0.8003 | -1.8897 | -2.0004 | 0.5273 | 0.1107 | -718.4238 | -579.4417 | -5.6556 | -5.3947 |
Framework versions
- Transformers 4.40.2
- Pytorch 2.1.2+cu118
- Datasets 2.19.1
- Tokenizers 0.19.1