This is a practice fine-tune of TinyLlama-1.1B on the Alpaca dataset using Unsloth. The goal of this project was not to achieve state-of-the-art performance but to understand the end-to-end fine-tuning pipeline using Unsloth, LoRA, and SFTTrainer on Google Colab.

What I learned

Setting up Unsloth for memory-efficient fine-tuning on Colab T4 GPU
Configuring LoRA parameters (rank, alpha, target modules)
Formatting datasets using the Alpaca instruction template
Using SFTTrainer from the trl library
Saving and pushing LoRA adapters to HuggingFace Hub

Training Details

Setting	Value
Base model	unsloth/tinyllama-bnb-4bit
Dataset

Uploaded model

Developed by: sagar-kc7
License: apache-2.0
Finetuned from model : unsloth/tinyllama-bnb-4bit

This llama model was trained 2x faster with Unsloth

sagar-kc7/tinyllama-alpaca-unsloth

TinyLlama Alpaca — Unsloth Fine-tuning Practice

Overview

What I learned

Training Details

Uploaded model

sagar-kc7/
tinyllama-alpaca-unsloth