shlee0/llama-3.2.1-finetuning-assign

Model Details

• Model ID: llama-3.2.1-finetuning-assign

Model Description

• Developed by: Meta AI
• Funded by: Internal Project
• Shared by: Meta Open Source
• Model type: Instruction-Following Large Language Model
• Language(s) (NLP): Korean
• License: Apache 2.0
• Finetuned from model: Llama-3.2-1B-Instruct

Uses

Direct Use

• Nature Language Process

Downstream Use

• order automation

Training Details

Training Data

• Source: order agent assignment data
• Size: 723KB

Training Hyperparameters

• Batch size: 2
• Learning rate: 1e-4
• Epochs: 1

Evaluation

Metrics

• BLEU: 87.28

Environmental Impact

• Hardware Type: Google Colab T4
• Hours used: 1

Citation

• Hugging Face: Voice Cloning
• SacreBLEU: https://github.com/mjpost/sacrebleu
• PEFT LoRA: https://github.com/huggingface/peft