BODOAI_GGUF

Meta-Llama-3-8B Fine-Tuned on Bodo Language

This repository contains the GGUF file for the Meta-Llama-3-8B model fine-tuned using Unsloth to understand and generate responses in the Bodo language, a prominent tribal language of Assam, India. The fine-tuning process utilized datasets of Bodo words, grammar rules, and conversational phrases to enable the model to effectively handle Bodo text.

Model Details

• Base Model: Meta-Llama-3-8B
• Fine-Tuning Framework: Unsloth
• Language: Bodo
• File Format: GGUF (GPU-Optimized Format)

Features

• Understands and processes Bodo grammar and vocabulary.
• Responds fluently in Bodo language.
• Trained on conversational and linguistic datasets to ensure contextual understanding.
• Compatible with popular transformer libraries and inference frameworks.

Fine-Tuning Dataset

The datasets used for fine-tuning include:

• Bodo Vocabulary: A curated list of commonly used words and phrases in Bodo.
• Bodo Grammar Rules: Comprehensive rules to help the model construct grammatically correct sentences.
• Conversational Data: Sample dialogues in Bodo for real-world conversational understanding.
• Literary Excerpts: Texts from Bodo literature to enrich linguistic nuances.

Installation

Prerequisites

Ensure you have the following installed:

• Python >= 3.8
• Transformers
• GGUF Support Library

Setup

1. Clone this repository:

   git clone https://huggingface.co/AyushSisodiya/BODOAI_GGUF
   cd <repo_name>
   

2. Install the required dependencies:

   pip install -r requirements.txt
   

3. Load the model in your project:

   from transformers import AutoModelForCausalLM, AutoTokenizer
   
   model_path = "AyushSisodiya/BODOAI_GGUF"
   tokenizer = AutoTokenizer.from_pretrained(model_path)
   model = AutoModelForCausalLM.from_pretrained(model_path)
   

Usage

To use the model for generating text in Bodo, follow these steps:

from transformers import AutoModelForCausalLM, AutoTokenizer

# Load the model and tokenizer
model_path = "AyushSisodiya/BODOAI_GGUF"
tokenizer = AutoTokenizer.from_pretrained(model_path)
model = AutoModelForCausalLM.from_pretrained(model_path)

# Encode a Bodo prompt
prompt = "Meaning of Hello in BODO"  # Example prompt in Bodo
inputs = tokenizer(prompt, return_tensors="pt")

# Generate a response
outputs = model.generate(**inputs, max_length=50)
response = tokenizer.decode(outputs[0], skip_special_tokens=True)
print(response)

Model Card

License

This model is distributed under the same license as the original Meta-Llama-3-8B model. Ensure compliance with its terms and conditions.

Acknowledgments

• Meta: For providing the base Llama-3 model.
• Unsloth: For the fine-tuning framework.
• Bodo Linguistic Community: For the datasets and linguistic insights.