metadata
library_name: peft
tags:
- trl
- sft
- generated_from_trainer
base_model: NousResearch/Llama-2-7b-hf
datasets:
- generator
model-index:
- name: llama2_instruct_generation
results: []
llama2_instruct_generation
This model is a fine-tuned version of NousResearch/Llama-2-7b-hf on the generator dataset. It achieves the following results on the evaluation set:
- Loss: 1.6709
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: 4
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: constant
- lr_scheduler_warmup_steps: 0.03
- training_steps: 500
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.9464 | 0.0027 | 20 | 1.8079 |
| 1.8933 | 0.0054 | 40 | 1.7789 |
| 1.8882 | 0.0081 | 60 | 1.7644 |
| 1.7861 | 0.0108 | 80 | 1.7545 |
| 1.7948 | 0.0135 | 100 | 1.7400 |
| 1.8786 | 0.0163 | 120 | 1.7138 |
| 1.8041 | 0.0190 | 140 | 1.7061 |
| 1.713 | 0.0217 | 160 | 1.6969 |
| 1.8453 | 0.0244 | 180 | 1.6944 |
| 1.8526 | 0.0271 | 200 | 1.6914 |
| 1.8792 | 0.0298 | 220 | 1.6907 |
| 1.6989 | 0.0325 | 240 | 1.6874 |
| 1.8333 | 0.0352 | 260 | 1.6868 |
| 1.8608 | 0.0379 | 280 | 1.6847 |
| 1.9045 | 0.0406 | 300 | 1.6830 |
| 1.8614 | 0.0433 | 320 | 1.6822 |
| 1.7854 | 0.0461 | 340 | 1.6801 |
| 1.8538 | 0.0488 | 360 | 1.6783 |
| 1.7165 | 0.0515 | 380 | 1.6766 |
| 1.8302 | 0.0542 | 400 | 1.6782 |
| 1.7844 | 0.0569 | 420 | 1.6761 |
| 1.747 | 0.0596 | 440 | 1.6742 |
| 1.7928 | 0.0623 | 460 | 1.6735 |
| 1.7661 | 0.0650 | 480 | 1.6732 |
| 1.7684 | 0.0677 | 500 | 1.6709 |
Framework versions
- PEFT 0.11.1
- Transformers 4.41.2
- Pytorch 2.3.0+cu121
- Datasets 2.19.2
- Tokenizers 0.19.1