LLM-GANs-Projects

A structured collection of notebooks and scripts covering deep learning, generative models, LLM applications, agent frameworks, and vector search.

GANs (GANs/)

Art-Portrate Generator

• Art portrait generation workflow from a Kaggle-sourced dataset (Art_Portraits_Generator.ipynb)
• Subject coverage: image preprocessing, training flow, generated portrait outputs

Face Generation

• Progressive GAN face generation notebook ((Progan)_Face_generation_with_GanAI_.ipynb)
• Model/theme referenced: Google ProGAN (google/progan-128)

MNIST generator

• Convolutional GAN for MNIST digits (GAN_Convolution.ipynb)
• Baseline dense GAN workflow (generated_images/GANs_normal_network.ipynb)
• Subject coverage: adversarial training loop, discriminator/generator dynamics, sample generation

models

• Stable Diffusion XL image generation
  • Script + notebook: stable-diffusion-xl-base-1.0
  • Files: models/Stable-Diffusion/main.py, stable_diffusion.ipynb
• Z-Image-Turbo generation pipeline
  • Model: Tongyi-MAI/Z-Image-Turbo
  • Files: models/Z-Image-Turbo/image-generator.py
• FLUX image generation notebooks
  • Models: black-forest-labs/flux-1.1-pro, black-forest-labs/flux-kontext-pro
• Imagen notebook
  • Model reference: google/imagen-4

LLMs (LLMs/)

Code Generator

• Code synthesis notebook using Salesforce CodeGen
• Model: Salesforce/codegen-2B-multi

Customer Service Chatbot

• Customer support QA chatbot notebook
• Model: deepset/roberta-base-squad2

Excel Data Analyzer

• Spreadsheet question-answering and analysis notebook
• Model: gemini-2.5-flash

Legal-Document-Analyzer

• Legal document summarization and analysis notebooks
• Models used/referenced:
  • facebook/bart-large-cnn
  • nlpaueb/legal-bert-base-uncased

Short Story Generator

• Creative text generation notebook
• Model: Google Flan-T5 (flan-t5 family)

Lang-Graph (Lang-Graph/)

codebasics_tutorial

• Chatbot graph basics (chatbot.ipynb)
• Conditional graph routing (conditional_graph.ipynb)
• Financial analysis graph (financial_graph.ipynb)
• Memory in graph workflows (memory_in_langgraph.ipynb)
• Tool-calling stock price flow (tool_call_stock_price.ipynb)
• Models used in notebooks:
  • google_genai:gemini-2.0-flash
  • gpt-3.5-turbo

LangChain (LangChain/)

Travel Scheduler

• Travel planning/scheduling notebook (Travel_Scheduler_LLM.ipynb)
• Model theme referenced in notebook: IBM Granite 3.2

udemy_course

• Practical multi-project workspace for course experiments and refactors
• Includes:
  • hello_world/: prompt + context foundations
  • search-agent/: Tavily-powered search agent
  • ReAct/: tool-calling ReAct patterns (LangChain + raw implementations)
  • RAGs/: retrieval projects (Medium_analyzer, Tavily, langchain Documentation Helper(agentic-rag))
  • langgraph/: ReAct-style LangGraph agent with crawl/search tools and loop safeguards
• Common stack:
  • OpenAI models (gpt-5, gpt-4o-mini, gpt-3.5-turbo)
  • Tavily search/crawl
  • LangChain + LangGraph
  • uv-managed Python environments

MCP (MCP/)

Authentiation

• FastMCP-based HTTP server scaffold for authenticated tool access
• Implemented tools:
  • retrieve_users_notes
  • add_note
• Implemented subjects:
  • MCP server structure
  • Bearer-auth/OAuth-style scaffolding imports
  • CORS middleware setup

Neural Networks (Neural Networks/)

Tensorflow

• Neural network fundamentals notebook (NeuralNetwork.ipynb)
• From-scratch workflow (from_scratch.ipynb)
• Keras CNN notebook (keras_CNN.ipynb)

PyTorch

• Differential Privacy in PyTorch (DP_in_PyTorch.ipynb)

HPC-IPM

• CNN notebooks (CNN.ipynb, CNNcode.ipynb)
• Subject coverage: CNN architecture/training exercises

SymanticSearch (SymanticSearch/)

• Semantic search with OpenAI embeddings + Pinecone (Simantic_Search_OpenAI(ada3)_and_Pinecone.ipynb)
• Text embedding pipeline with pretrained models + Pinecone (Text_Embedding_with_Pretrained_model.ipynb)
• Implemented subjects:
  • Embedding creation
  • Vector upsert/retrieval flow
  • Similarity-based semantic querying

tools (tools/)

vector_database (tools/vector_database/)

chromadb

• ChromaDB crash-course notebook (tools/vector_database/chromadb/chroma_db.ipynb)
• Implemented subjects:
  • Collection creation and management
  • Document embedding storage
  • Similarity query workflow (cosine/distance concepts)

ui/ux kits (tools/ui/ux kits/)

streamlit-capital-chatbot

• Streamlit-based country-capital chatbot (tools/ui/ux kits/streamlit-capital-chatbot/app.py)
• Uses LangChain ChatOpenAI for capital lookup responses
• Implemented subjects:
  • Prompt-guided capital retrieval from country input
  • Streamlit chat UI interaction flow
  • Invalid-country fallback behavior (Unknown country.)

gradio-capital-chatbot

• Gradio web UI country-capital chatbot (tools/ui/ux kits/gradio-capital-chatbot/app.py)
• Uses LangChain ChatOpenAI to map country names to capitals
• Implemented subjects:
  • Lightweight chatbot UI in Gradio Blocks/Interface flow
  • Country query handling and assistant response rendering
  • Testable app structure with pytest support

copilotkit-capital-chatbot

• FastAPI backend for capital queries with CopilotKit-compatible routing (tools/ui/ux kits/copilotkit-capital-chatbot/app.py)
• Uses LangChain ChatOpenAI for capital resolution logic
• API behavior and endpoints:
  • GET /health
  • GET /copilotkit-status
  • POST /capital
  • POST /copilotkit (when copilotkit imports successfully)
• Implemented subjects:
  • CopilotKit endpoint registration with runtime availability checks
  • Fallback capital API flow when CopilotKit package is unavailable
  • Service health/status reporting for integration diagnostics

scraping_and_crawling (tools/scraping_and_crawling/)

Tavily crawl demo

• Simple Tavily-based web crawling workflow with baseline vs instruction-guided crawl (tools/scraping_and_crawling/crawl_demo.py)
• Shell automation:
  • setup_env.sh: venv setup + dependency install + Jupyter kernel registration
  • run.sh: one-command setup + run
• Saved outputs:
  • crawled/baseline_results.json
  • crawled/baseline_results.txt
  • crawled/guided_results.json
  • crawled/guided_results.txt
• Implemented subjects:
  • API-key based Tavily integration (TAVILY_API_KEY)
  • Instruction-guided crawling for relevance filtering
  • JSON persistence + cleaned text export for downstream use

License

This repository is licensed under the MIT License. See the LICENSE file for full text.