metadata
license: apache-2.0
base_model: alignment-handbook/zephyr-7b-sft-full
tags:
- generated_from_trainer
model-index:
- name: zephyr-7b-dpo-lora
results: []
zephyr-7b-dpo-lora
This model is a fine-tuned version of alignment-handbook/zephyr-7b-sft-full on the None dataset. It achieves the following results on the evaluation set:
- Loss: -0.2038
- Rewards/chosen: -1.1628
- Rewards/rejected: -2.4457
- Rewards/accuracies: 0.6840
- Rewards/margins: 1.2829
- Logps/rejected: -252.9479
- Logps/chosen: -282.7848
- Logits/rejected: -2.9400
- Logits/chosen: -2.9655
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: 4
- seed: 42
- distributed_type: multi-GPU
- num_devices: 4
- gradient_accumulation_steps: 2
- total_train_batch_size: 64
- total_eval_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 3
Training results
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.6137 | 1.0 | 968 | 0.6277 | -0.0287 | -0.4191 | 0.7040 | 0.3905 | -232.6823 | -271.4433 | -2.9989 | -3.0154 |
| 0.0705 | 2.0 | 1937 | 0.0570 | -0.6708 | -1.6676 | 0.6960 | 0.9968 | -245.1669 | -277.8647 | -2.9609 | -2.9830 |
| -0.2602 | 3.0 | 2904 | -0.2038 | -1.1628 | -2.4457 | 0.6840 | 1.2829 | -252.9479 | -282.7848 | -2.9400 | -2.9655 |
Framework versions
- Transformers 4.35.0
- Pytorch 2.1.1+cu121
- Datasets 2.14.6
- Tokenizers 0.14.1