PopularMajor Project
Artificial Intelligence
Face Recognition Attendance System
Camera-based attendance with liveness detection and analytics
- Python
- OpenCV
- FaceNet
- Flask
- +2
₹4,900₹5,999
4.8/301 reviews
Artificial Intelligence · Capstone
Fine-Tune an LLM on Your Own Data
LoRA fine-tuning of a small open model with before-and-after evaluation
4-5 weeks build
Overview
Adapts a small open-weight language model to a task of your choosing using LoRA, so training fits on a single consumer or Colab GPU. The kit covers building an instruction dataset, 4-bit quantised training, and a held-out evaluation that compares the base and fine-tuned models on the same prompts with both automatic metrics and a rubric.
What makes it stand up
- 01Instruction dataset builder with cleaning and deduplication
- 02LoRA adapters with 4-bit quantisation to fit modest hardware
- 03Before-and-after evaluation on a held-out prompt set
- 04Rubric-based human review alongside automatic metrics
- 05Adapter merging and a Gradio chat demo
Module breakdown
- 01
Dataset
Instruction-response pairs, formatting and splits.
- 02
Training
PEFT LoRA configuration, quantisation and checkpoints.
- 03
Tuning
Rank, learning-rate and epoch experiments logged.
- 04
Evaluation
Base versus fine-tuned comparison on held-out prompts.
- 05
Demo
Merged model served in a Gradio chat interface.
After this, you will be able to
- Explain how LoRA updates so few parameters
- Defend your evaluation design and its limits
- Demo base and fine-tuned answers side by side
Technology stack
- Python
- PyTorch
- Transformers
- PEFT
- Hugging Face
- Gradio
- Level
- Capstone
- Domain
- Artificial Intelligence
- Build time
- 4-5 weeks
- Difficulty
You receive
- Complete, commented source code
- 60-90 page project report (IEEE format)
- Editable presentation deck
- Architecture, ER and UML diagrams
- Local setup and deployment guide
- Sample dataset or seed data
- Viva question bank with answers
- One 45-minute walkthrough call
- 30 days of doubt-clearing support