README.md

metadata

license: apache-2.0
base_model: google/flan-t5-small
tags:
  - resume
  - chatbot
  - flan-t5
  - text2text-generation
  - question-answering
  - fine-tuning
library_name: transformers
pipeline_tag: text2text-generation

Vaishnavi ResumeBot - FLAN-T5

This repository contains a personal ResumeBot built by fine-tuning google/flan-t5-small on resume-based question-answer pairs.

About

The bot is designed to answer questions about Vaishnavi V M's academic background, technical skills, projects, and contact information.

Base Model

google/flan-t5-small

Main Topics

• MSc Data Science at Dayananda Sagar University
• BSc Mathematics and Computer Science
• Python, Machine Learning, Deep Learning, NLP, SQL
• Medical X-ray Abnormality Localization using Attention Mechanisms
• Driver Drowsiness Detection
• Phishing Website Detection
• Language Identification and Accent Detection
• Social Media Addiction Risk Prediction

Repository Files

• Vaishnavi_ResumeBot_FLAN_T5.ipynb - complete Colab notebook
• resume_qa_dataset.csv - training question-answer pairs
• app.py - Gradio ResumeBot interface
• inference.py - command-line prediction example
• push_to_hub.py - upload your trained model to Hugging Face
• requirements.txt - required libraries
• .gitignore - files to ignore in Git
• MODEL_FILES_README.txt - explains how trained model files are created

How to Train

Open Vaishnavi_ResumeBot_FLAN_T5.ipynb in Google Colab.

Use:

Runtime -> Change runtime type -> T4 GPU

Then run the notebook from top to bottom.

After training, save the model using:

model.save_pretrained("vaishnavi-resumebot-flan-t5")
tokenizer.save_pretrained("vaishnavi-resumebot-flan-t5")

This produces files such as:

• config.json
• generation_config.json
• model.safetensors
• tokenizer files
• spiece.model

Run the App Locally

pip install -r requirements.txt
python app.py

Example Questions

• What is your educational background?
• What are your technical skills?
• Tell me about your medical X-ray project.
• Which algorithms did you use for phishing detection?
• What is your language identification project?
• What is your email address?

Important

The final fine-tuned model.safetensors is not included in the starter repository because it must be generated by actually training the model.