metadata
license: other
library_name: peft
tags:
- generated_from_trainer
base_model: TheBloke/Wizard-Vicuna-7B-Uncensored-GPTQ
model-index:
- name: amadeus_v1
results: []
amadeus_v1
This model is a fine-tuned version of TheBloke/Wizard-Vicuna-7B-Uncensored-GPTQ on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 4.3259
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: 0.002
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 2
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 3.2904 | 1.0 | 75 | 3.2025 |
| 3.0004 | 2.0 | 150 | 3.1939 |
| 2.6935 | 3.0 | 225 | 3.2721 |
| 2.4059 | 4.0 | 300 | 3.3634 |
| 2.1472 | 5.0 | 375 | 3.5849 |
| 1.9307 | 6.0 | 450 | 3.7793 |
| 1.7398 | 7.0 | 525 | 3.8648 |
| 1.625 | 8.0 | 600 | 4.0542 |
| 1.5137 | 9.0 | 675 | 4.2189 |
| 1.4398 | 10.0 | 750 | 4.3259 |
Framework versions
- PEFT 0.10.0
- Transformers 4.38.2
- Pytorch 2.1.0+cu121
- Datasets 2.18.0
- Tokenizers 0.15.2