metadata
base_model: unsloth/Llama-3.2-1B-Instruct
tags:
- text-generation-inference
- transformers
- unsloth
- llama
- gguf
license: apache-2.0
language:
- en
🧠 Llama 3.2 - 1B Instruct | Toolcalling Test Finetuned Model (ADROIT NOT USING ANYMORE DEVELOPED FOR TESTING ONLY)
- Developed by: nawazadroit
- License: Apache-2.0
- Finetuned from base model:
unsloth/Llama-3.2-1B-Instruct
🧪 Purpose
This is a simple tool-calling test model, finetuned specifically on a custom dataset (dataset.json) related to scheme application workflows. It is designed for structured assistant-like behavior with basic tool invocation capabilities, especially in the domain of public healthcare scheme assistance (e.g., applying to MJPJAY / PM-JAY schemes in government hospitals).
🚀 Features
- ✅ Lightweight 1B parameter model
- ✅ Uses Huggingface's TRL library for reward modeling and instruction tuning
- ✅ Ideal for testing local tool-calling setups
- ✅ Compatible with GGUF format for efficient inference
📂 Dataset Info
The model was trained on a JSON dataset (dataset.json) containing multi-turn dialogues structured for:
- Verifying patient eligibility
- Applying health schemes
- Extracting document information
- Calling specific tools via structured API-like responses