LLM-gym

Understanding Math Behind large Language Models

• Minimal Introduction to Large Language Models : Understand the High Level Concepts of Transformers Visually.
• Understanding LLM Inference : Understand operations behind Kv-Cache , Flops Counting!
• Quantization Principal : Understand types of quantization and how they can reduce memory overhead required for inferencing and training!
• Calculate GPU Memory Required for Serving LLMS : Reasons that make up memory for a LLM model!
• LLM Underhood : Under-the-hood mechanics of Transformers technique.
• Ai Compute Evolution : How compute necessaties evolved over decades.

RAG Papers/Implementations

• Speculative RAG : Use Generalist RAG to rank multiple Answer Drafts from Specialist RAG!
• Corrective RAG : Lookahead , decide whether correct or not , then rewrite the query if not correct!
• Hyde Query Rewritting : Generate a hypothetical Document from the Query and search against that in the vector space!
• Late Chunking Blog: Late chunking the sweet spot between naive chunking retrieval and late interaction based retrieval!
• Contextual Rag : While Indexing Ask LLM to contextualize the chunk and then index it!
• Retrieval via Late Interaction Models : Colipali and How it used Bag of Words for Document Similarity
• Query Routing : A Set-Fit Based approach for query routing.

Some Essential Deployment Tools

• Converting to GGUF Tools : Converting LLM to GGUF Format.

About Vector Databases and Search Algorithms

• HNSW Algorithm : It shows the intution behind how the HNSW algorithm works.

AI Safety and Harms

• Incident Database: Incidents and Reports of AI Harms.

Agents & Systems

• Building Effective Agentic Flows: Building effective agents
• LLM Systems: ML & LLM Systems
• Building Agentic Sysmtes: Chip's guide to building Agentic Systems
• Agent Frameworks Comparision: Comparision against different Agents

Benchmarks

• Benchmarks: Evaluation Datasets