metadata
license: apache-2.0
base_model: TheBloke/OpenHermes-2-Mistral-7B-GPTQ
tags:
- trl
- dpo
- generated_from_trainer
model-index:
- name: openhermes-mistral-dpo-gptq
results: []
openhermes-mistral-dpo-gptq
This model is a fine-tuned version of TheBloke/OpenHermes-2-Mistral-7B-GPTQ on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.6643
- Rewards/chosen: 0.0014
- Rewards/rejected: -0.0701
- Rewards/accuracies: 0.8125
- Rewards/margins: 0.0714
- Logps/rejected: -216.5143
- Logps/chosen: -215.7596
- Logits/rejected: -2.5311
- Logits/chosen: -2.5242
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: 0.0002
- train_batch_size: 1
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 2
- training_steps: 50
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.6845 | 0.01 | 10 | 0.6884 | 0.0077 | -0.0181 | 0.6875 | 0.0258 | -215.9950 | -215.6967 | -2.5278 | -2.5201 |
| 0.7249 | 0.01 | 20 | 0.6905 | -0.0073 | -0.0256 | 0.75 | 0.0183 | -216.0695 | -215.8464 | -2.5330 | -2.5253 |
| 0.6441 | 0.01 | 30 | 0.6808 | 0.0070 | -0.0255 | 0.75 | 0.0325 | -216.0689 | -215.7035 | -2.5350 | -2.5269 |
| 0.6393 | 0.02 | 40 | 0.6657 | -0.0032 | -0.0731 | 0.875 | 0.0699 | -216.5449 | -215.8051 | -2.5327 | -2.5248 |
| 0.6818 | 0.03 | 50 | 0.6643 | 0.0014 | -0.0701 | 0.8125 | 0.0714 | -216.5143 | -215.7596 | -2.5311 | -2.5242 |
Framework versions
- Transformers 4.35.2
- Pytorch 2.0.1+cu117
- Datasets 2.16.1
- Tokenizers 0.15.0