ttPG - Torch and Transformers Playground

A comprehensive learning repository containing Jupyter notebooks and Python scripts for mastering PyTorch and Hugging Face Transformers. This repository serves as a hands-on learning resource covering fundamental concepts to advanced techniques in deep learning and natural language processing.

📚 Repository Structure

🔥 PyTorch Basics (pytorch_basics/)

Fundamental PyTorch concepts and operations:

• tensors.ipynb - Introduction to PyTorch tensors, data types, and basic operations
• derivatives.ipynb - Understanding automatic differentiation and gradients
• dataset-dataloader.ipynb - Working with datasets and data loaders for efficient data handling

🤗 Transformers Basics (transformers_basics/)

Hugging Face Transformers library fundamentals:

• pipeline_module.ipynb - Using pre-built pipelines for common NLP tasks
• tokenizer.ipynb - Text tokenization and preprocessing
• models.ipynb - Working with pre-trained models
• datasets.ipynb - Loading and processing datasets
• trainer_finetuning.ipynb - Fine-tuning models using the Trainer API
• behind_pipeline.ipynb - Understanding the internals of transformer pipelines

🎯 Sentiment Classification (sentiment_classification/)

Complete sentiment analysis project:

• sample_notebook/notebook.ipynb - Sample implementation and experimentation
• scripts/train.py - Training script for sentiment classification using DistilBERT
• scripts/eval.py - Evaluation script for model performance assessment

⚡ Model Quantization (model_quantization/)

Advanced techniques for model optimization:

• llm_quantizaton_inference.ipynb - Large Language Model quantization techniques for efficient inference

📊 Text Feature Exploration

• text_feature_exploration.ipynb - Text feature extraction and visualization techniques
• word_embeddings_gensim_word2vec.ipynb - Word2Vec implementation using Gensim

🚀 Getting Started

Prerequisites

• Python 3.8+
• CUDA-compatible GPU (recommended for some notebooks)

Installation

1. Clone the repository:

git clone https://github.com/yourusername/ttPG.git
cd ttPG

2. Install dependencies:

pip install -r requirements.txt

Dependencies

• torch - PyTorch deep learning framework
• torchtext - Text processing utilities
• transformers - Hugging Face Transformers library
• accelerate - Training acceleration utilities
• bitsandbytes - Quantization and optimization
• einops - Tensor operations
• sentence_transformers - Sentence embeddings

🤝 References

Official Learning Resources

• PyTorch Learn the Basics - Official PyTorch tutorial covering tensors, datasets, neural networks, and complete ML workflows
• Hugging Face LLM Course - Comprehensive course on Large Language Models and NLP using the Hugging Face ecosystem

Additional Resources

• PyTorch Documentation - Complete PyTorch API reference
• Hugging Face Transformers Documentation - Official Transformers library documentation
• Hugging Face Hub - Discover and share models, datasets, and demos

📄 License

This project is open source and available under the MIT License.

🙏 Acknowledgments

• PyTorch team for the excellent deep learning framework
• Hugging Face for the comprehensive transformers library
• The open-source community for continuous inspiration and support