metadata
base_model: mistralai/Mistral-7B-v0.1
library_name: peft
license: apache-2.0
tags:
- trl
- dpo
- generated_from_trainer
model-index:
- name: zephyr-7b-dpo-qlora
results: []
zephyr-7b-dpo-qlora
This model is a fine-tuned version of mistralai/Mistral-7B-v0.1 on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 1964.4768
- Rewards/chosen: -0.1407
- Rewards/rejected: -0.2359
- Rewards/accuracies: 0.7243
- Rewards/margins: 0.0952
- Logps/rejected: -25.2353
- Logps/chosen: -15.6615
- Logits/rejected: -6.0582
- Logits/chosen: -5.0283
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: 4
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- num_devices: 8
- gradient_accumulation_steps: 4
- 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 |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 2178.8832 | 0.3017 | 100 | 2139.6340 | -0.1014 | -0.1518 | 0.6781 | 0.0504 | -16.8316 | -11.7293 | -3.3193 | -3.1137 |
| 1986.1408 | 0.6033 | 200 | 1989.8781 | -0.1408 | -0.2280 | 0.7055 | 0.0872 | -24.4503 | -15.6687 | -6.1921 | -5.2910 |
| 2000.1098 | 0.9050 | 300 | 1964.4768 | -0.1407 | -0.2359 | 0.7243 | 0.0952 | -25.2353 | -15.6615 | -6.0582 | -5.0283 |
Framework versions
- PEFT 0.12.0
- Transformers 4.43.3
- Pytorch 2.4.0+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1