Customer Support Llama 3.2 3B LoRA
This model is a fine-tuned LoRA adapter based on unsloth/Llama-3.2-3B-Instruct, optimized to behave like a trained, professional customer support assistant.
Model Description
- Developed by: [Your Hugging Face Username]
- Model Type: LoRA Adapter (Fine-tuned)
- Base Model: unsloth/Llama-3.2-3B-Instruct
- Language: English
Intended Uses & Limitations
This model is designed to answer customer support queries across various retail e-commerce categories, including:
- Order cancellations and status updates
- Refund policy inquiries
- Changing shipping addresses
- Account deletion requests
Training Dataset
The model was fine-tuned on the bitext/Bitext-customer-support-llm-chatbot-training-dataset, which includes 26,872 real-world customer support instruction-response pairs spanning 27 intents.
Training Configuration
The adapter was trained using the following parameters:
- r (rank): 16
- lora_alpha: 16
- Training Steps: 60
- Learning Rate: 2e-4
- Optimizer: adamw_8bit
- Hardware: Google Colab T4 GPU
