Himel09/llama-customer-support-ai-lora

🏦 Banking Chatbot – LoRA Fine-Tuned Model

This model is a LoRA fine-tuned version of a Llama-based model, designed to detect user intent for banking tasks and generate agent-style responses. It is optimized for tasks such as:

• —✔ Checking balance
• —✔ Last transaction lookup
• —✔ Transaction history
• —✔ Loan eligibility queries
• —✔ Banking account actions
• —✔ Customer service style replies

🔧 Model Details

• —Base Model: Llama-3 / Unsloth version (4-bit quantized during training)
• —Fine-Tuning Method: LoRA (Parameter-Efficient Training)
• —Frameworks Used:
• —unsloth
• —transformers
• —trl (SFTTrainer)
• —Dataset: Custom banking conversational dataset with structured responses
• —Purpose: Intent detection + Agent response generation

📘 Training Format

The model is trained on structured dialogue patterns such as:

Each record includes:

• —User query
• —Intent label
• —Agent response

This format ensures stable predictable output for banking applications.