Skip to content

SreevaatsavB/Finetuning_Repo

Repository files navigation

Model Fine-tuning Repository

This repository contains a collection of Jupyter notebooks and examples for fine-tuning various AI models for different tasks. Each directory focuses on a specific model type or application area.

Repository Structure

Finetuning_Repo/
├── Doc_Retrieval_finetuning/      # Fine-tuning sentence transformers for document retrieval
├── FLUX_image_generation/         # Fine-tuning FLUX models for image generation
├── LLM_finetuning/                # Fine-tuning Language Models
├── OCR_finetuning/                # OCR model fine-tuning
│   ├── Image_trascription/        # Fine-tuning for image-to-text transcription (TrOCR)
│   └── Text_detection/            # Fine-tuning for text detection in images (YOLO)
├── RAG_Reranker_finetuning/       # Fine-tuning rerankers for RAG systems
└── VLM_finetuning/                # Fine-tuning Vision Language Models

Modules Overview

Doc_Retrieval_finetuning

Contains notebooks for fine-tuning sentence transformer models for document retrieval tasks. Improves embedding quality for specific domains.

FLUX_image_generation

Notebooks and scripts for fine-tuning FLUX models for conditional image generation tasks. Enables creating customized image outputs based on specific prompts or conditions.

LLM_finetuning

Resources for fine-tuning Large Language Models for various text generation tasks.

OCR_finetuning

Contains two sub-modules:

  • Image_trascription: Fine-tune TrOCR models for converting image text to typed text
  • Text_detection: Fine-tune YOLO models for detecting text regions in images

RAG_Reranker_finetuning

Notebooks for fine-tuning reranker models used in Retrieval-Augmented Generation (RAG) pipelines.

VLM_finetuning

Resources for fine-tuning Vision-Language Models for multimodal tasks.

Getting Started

Each directory contains specific notebooks with detailed instructions. Navigate to the directory of interest and refer to its README and Jupyter notebooks for task-specific guidance.

Requirements

Each notebook specifies its own requirements, but common dependencies include:

  • Python 3.x
  • PyTorch
  • Transformers
  • Datasets
  • Specific model libraries (like sentence-transformers, YOLO, etc.)

For detailed requirements, see the specific notebook you wish to run.

About

A repo with fine-tuning scripts for training your own custom DL models on custom datasets.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published