metadata
license: apache-2.0
library_name: peft
tags:
- alignment-handbook
- generated_from_trainer
- trl
- dpo
- generated_from_trainer
datasets:
- EllieS/pubmedqa_dpo_selfgen_data
base_model: alignment-handbook/zephyr-7b-sft-full
model-index:
- name: zephyr-7b-dpo-lora-pubmedqa-selfgen-old
results: []
zephyr-7b-dpo-lora-pubmedqa-selfgen-old
This model is a fine-tuned version of EllieS/zephyr-7b-sft-qlora on the EllieS/pubmedqa_dpo_selfgen_data dataset. It achieves the following results on the evaluation set:
- Loss: 0.0027
- Rewards/chosen: -1.9349
- Rewards/rejected: -10.4879
- Rewards/accuracies: 1.0
- Rewards/margins: 8.5530
- Logps/rejected: -1094.3392
- Logps/chosen: -260.5620
- Logits/rejected: -2.3781
- Logits/chosen: -2.6042
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-06
- train_batch_size: 2
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 2
- gradient_accumulation_steps: 2
- total_train_batch_size: 8
- total_eval_batch_size: 16
- 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 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.023 | 0.42 | 7000 | 0.0164 | -1.0835 | -8.5927 | 1.0 | 7.5092 | -904.8198 | -175.4290 | -2.8442 | -2.8776 |
| 0.0052 | 0.83 | 14000 | 0.0026 | -1.9746 | -10.4931 | 1.0 | 8.5185 | -1094.8602 | -264.5324 | -2.3838 | -2.6053 |
Framework versions
- PEFT 0.7.1
- Transformers 4.36.2
- Pytorch 2.1.2+cu121
- Datasets 2.14.6
- Tokenizers 0.15.2