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 · Major Project
Abstractive Summarisation with Transformers
Fine-tunes BART or T5 to write summaries in its own words
3-4 weeks build
Overview
Fine-tunes a pretrained encoder-decoder model on the public CNN/DailyMail or XSum dataset so it generates new sentences rather than copying old ones. The kit covers tokenisation limits, training on a single GPU, beam search settings and a ROUGE evaluation that compares the fine-tuned model with the untuned checkpoint and an extractive baseline.
What makes it stand up
- 01Fine-tuning pipeline for both BART and T5 checkpoints
- 02Mixed-precision training sized for a free Colab GPU
- 03Decoding controls: beam width, length penalty and n-gram blocking
- 04ROUGE scores against untuned and extractive baselines in the report
- 05Hallucination review sheet with documented failure examples
Module breakdown
- 01
Data
Dataset loading, truncation strategy and train-validation-test splits.
- 02
Training
Seq2seq trainer with checkpointing and early stopping.
- 03
Decoding
Configurable beam search and sampling for generation.
- 04
Evaluation
ROUGE metrics plus a manual factual-consistency check.
- 05
Serving
FastAPI endpoint and a React page for live summaries.
After this, you will be able to
- Explain the encoder-decoder attention flow in a transformer
- Defend your ROUGE comparison and its known limitations
- Demo a live summary of an unseen news article
Technology stack
- Python
- PyTorch
- Transformers
- Hugging Face
- FastAPI
- React
- Level
- Major Project
- Domain
- Artificial Intelligence
- Build time
- 3-4 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