metadata
library_name: peft
tags:
- trl
- sft
- generated_from_trainer
base_model: ytcheng/llama-3-8b-hf-sm-lora-merged
datasets:
- generator
model-index:
- name: llama-3-8b-hf-ft-chat-lora
results: []
llama-3-8b-hf-ft-chat-lora
This model is a fine-tuned version of ytcheng/llama-3-8b-hf-sm-lora-merged on the generator dataset. It achieves the following results on the evaluation set:
- Loss: 4.8363
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: 0.0001
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- distributed_type: multi-GPU
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.3
- num_epochs: 4
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 2.3345 | 0.9963 | 133 | 2.4116 |
| 1.9916 | 2.0 | 267 | 3.9850 |
| 1.8552 | 2.9963 | 400 | 4.8651 |
| 1.8627 | 3.9850 | 532 | 4.8363 |
Framework versions
- PEFT 0.10.0
- Transformers 4.40.0
- Pytorch 2.2.2+cu121
- Datasets 2.18.0
- Tokenizers 0.19.1