metadata
license: apache-2.0
base_model: alignment-handbook/zephyr-7b-sft-full
tags:
- trl
- dpo
- generated_from_trainer
model-index:
- name: zephyr-7b-dpo-full
results: []
zephyr-7b-dpo-full
This model is a fine-tuned version of alignment-handbook/zephyr-7b-sft-full on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.5055
- Rewards/chosen: -0.9459
- Rewards/rejected: -1.9033
- Rewards/accuracies: 0.7812
- Rewards/margins: 0.9573
- Logps/rejected: -452.9886
- Logps/chosen: -357.2234
- Logits/rejected: 1.2877
- Logits/chosen: 0.5168
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 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.5659 | 0.2092 | 100 | 0.5701 | -0.7175 | -1.2998 | 0.7148 | 0.5824 | -392.6456 | -334.3749 | -0.7859 | -0.9821 |
| 0.5493 | 0.4184 | 200 | 0.5264 | -0.7175 | -1.5459 | 0.7734 | 0.8284 | -417.2501 | -334.3752 | 0.2805 | -0.3142 |
| 0.493 | 0.6276 | 300 | 0.5119 | -0.9435 | -1.8186 | 0.7617 | 0.8751 | -444.5215 | -356.9808 | 1.2687 | 0.5290 |
| 0.5014 | 0.8368 | 400 | 0.5055 | -0.9459 | -1.9033 | 0.7812 | 0.9573 | -452.9886 | -357.2234 | 1.2877 | 0.5168 |
Framework versions
- Transformers 4.43.1
- Pytorch 2.1.2+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1