Hugging Face's logo

langgptai
/
llama3-8b_sa_v0.1

PEFT
Safetensors
llama-factory
lora
Generated from Trainer
Model card Files Community
llama3-8b_sa_v0.1 / README.md
sci-m-wang's picture
sci-m-wang
Upload 14 files
f232a59 verified
preview code
|
raw
history blame
1.83 kB
metadata
license: other
library_name: peft
tags:
  - llama-factory
  - lora
  - generated_from_trainer
base_model: meta-llama/Meta-Llama-3-8B-Instruct
model-index:
  - name: LangGPT
    results: []

LangGPT

This model is a fine-tuned version of /datas/huggingface/Meta-Llama-3-8B-Instruct on the LangGPT dataset. It achieves the following results on the evaluation set:

  • Loss: 0.7564

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-05
  • train_batch_size: 12
  • eval_batch_size: 4
  • seed: 42
  • gradient_accumulation_steps: 8
  • total_train_batch_size: 96
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_steps: 20
  • num_epochs: 9.0
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss
0.8441 1.25 100 0.8462
0.7829 2.5 200 0.7942
0.747 3.75 300 0.7737
0.7507 5.0 400 0.7637
0.7327 6.25 500 0.7590
0.7331 7.5 600 0.7569
0.7396 8.75 700 0.7564

Framework versions

  • PEFT 0.10.0
  • Transformers 4.40.1
  • Pytorch 2.2.0+cu121
  • Datasets 2.16.1
  • Tokenizers 0.19.1