This repository showcases a curated selection of AI projects and experiments I have developed & worked with since March 2023. I am actively (Aug 2025) in the process of editing these. Please check back again for changes.
- Featured Projects – AI projects I have built.
- Experiments – exploratory code and prototype work
- Reference – Projects that I am watching / testing
2025-04_HybridModelChatBridge.ipynb
Proof-of-concept for attempting to use an instruct model as a 'bridge' or 'interpreter' between base model and user by reformating user questions for specialized base models.
Key Features:
- Base-to-instruct model architecture
- Multi-model conversation flow
- Interactive Gradio interface
- Confidence scoring system
- Performance metrics analysis
Fine_tune_llm_custom_data.ipynb
This project is focused on fine-tuning a small open-source language model (Qwen/Qwen2.5-3B-Instruct) on custom data related to a super-hero that I quickly invented (I wanted something that I knew the model would have no prior knowledge of).
Key Features:
- Fine-tuning: The core process of adapting a pre-trained model to custom data.
- LoRA (Low-Rank Adaptation): Utilized for efficient fine-tuning.
- Custom Character Data: Fine-tuning on unique, invented character information.
- Small Open-Source Model: Using a specific, publicly available model (Qwen/Qwen2.5-3B-Instruct).
- Workflow: The project outlines a complete process from setup and data preparation to training, saving, and testing the fine-tuned model.
A collection of exploratory notebooks testing various AI/ML techniques: Note: Not all of these are my code... (should be obvious or credited). Note: Not all of these are edited/cleaned up yet. In process of editing them now (Aug 2025).
- 2023-03_HR_RAG_Chatbot_polished.ipynb - HR-focused chatbot with RAG
- 2024-01_AgentChatGroupChatRAG.ipynb - Multi-agent group chat system
- 2025-04_HybridModelChatBridge.ipynb - Hybrid Model Architecture - Base model to chat interface bridge
- 2023-10_ChatbotSimpleRAG.ipynb - Lightweight RAG chatbot prototype
- 2023-07_LangChainChatGPT.ipynb - Early LangChain integration experiments
- 2023-08_CampingRawExploration.ipynb - Generative AI exploration
- 2024-08_AiRAGBasics.ipynb - Step-by-step RAG implementation tutorial
- 2023-03_EmbeddingsDocxRAG.ipynb - Document embedding and retrieval
- 2023-12_AI_SEO_POC.ipynb - AI-driven SEO analysis tool
- 2023-12_AI_SEO_Tests.ipynb - SEO optimization experiments
- Fine_Tune_DeepSeek_R1_custom_data_test.ipynb - Custom data fine-tuning
- Finetune_redpajama_tests.ipynb - RedPajama model experimentation
- Distill_Model_FT.ipynb - Model distillation techniques
- Unsloth_finetuning_CJ.ipynb - Efficient fine-tuning with Unsloth
- Phi_3_Mini_4K_Instruct_Unsloth_2x_faster_finetuning_TEST.ipynb - Fast Phi-3 fine-tuning
- Falcon_7b_finetune_midjourney_falcon_test.ipynb - Falcon model experimentation
- rlhf_tune_llm_CJ.ipynb - RLHF tuning implementation
- RLHF_with_Custom_Datasets_CJ.ipynb - Custom RLHF datasets
- OpenAI_Chatbot_test.ipynb - Chatbot model testing
- llama_langchain_chatbot_custom_data_CJ.ipynb - LLaMA integration
- LangChain_Model_Laboratory.ipynb - LangChain framework exploration
- Kaggle_skin_cancer.ipynb - Medical image classification
- Neural_Networks_pyTorch_CJ.ipynb - Deep learning fundamentals
- OpenAI_Text_To_Speech.ipynb - Audio generation
- Whisper_voices_cj.ipynb - Speech recognition
- FastAPI_localtunnel_CJ.ipynb - API deployment
- gmail_api_test_cj.ipynb - Gmail API integration
- Semantic_search_openai_CJ.ipynb - Semantic search implementation
- CJ_Snippets_Drive.ipynb - Google Drive integration
- linkboxes_test_cj.ipynb - Link processing utilities
- Clone_your_own_voice_tortoise_tts_CJ.ipynb - Voice cloning with TTS
- Sentiment_analysis_tiny.ipynb - Sentiment analysis implementation
- SEO_Analysis_ORIG_version.ipynb - Original SEO analysis concept
- TensorBoard_fashion_test.ipynb - Model visualization
- RouteLLM_test.ipynb - Routing model testing
(See the experiments/ folder for the complete list.)
Retired or duplicate notebooks:
- Docker_colab_Python_for_SEO.ipynb
- August 2025: Current portfolio organization
- April 2025: Base Model Chat Bridge, Multi-Agent RAG System
- December 2023: SEO Analysis Tool development
- March 2023: Initial HR Chatbot implementation
-
Structured Input/Output:
- Outlines : Outlines guarantees structured outputs during generation — directly from any LLM.
-
Chat Frameworks:
- HayStack : AI orchestration framework with advanced retrieval methods; it's best suited for building RAG, question answering, semantic search or conversational agent chatbots.
- SecurityHundley & Cybersecurity: Same project. The polished version () is recommended for showcase.
- HR Chatbot Variants: Multiple versions exist (2023-03_HR_RAG_Chatbot.ipynb, 2023-03_HR_RAG_Chatbot_polished.ipynb). The polished version is recommended.
Contact:
For questions or feedback, reach out at kutyadog@gmail.com (Chris Johnson).