Shrinivasan SankarClaude 3.5 Sonnet vs GPT-4o — An honest reviewAnthropic the company behind Claude series of models has released Claude 3.5 Sonnet. It comes at a time when we all have accepted GPT-4o…3d ago3d ago
Shrinivasan SankarComparing Kolmogorov-Arnold Network(KAN ) and Multi-Layer Perceptrons (MLPs)We have taken the classic Multi-Layer Perceptrons (MLPs) for granted and built so many architectures around it. MLPs are part and parcel of…Jun 25Jun 25
Shrinivasan SankarXLSTM — Extended Long Short-Term Memory NetworksLSTMs or Long Short-Term Memory Networks have been around for a long time. They have been applied for quite a few sequence-related tasks…May 20May 20
Shrinivasan SankarinGoPenAIMake your LLM Fully Utilize the ContextA simple data-driven approach from Microsoft to increasing the context length of LLMsMay 9May 9
Shrinivasan SankarinLevel Up CodingChat with your emails with this RAG pipeline (LangChain + ChromaDB)Implement and run a simple application on your laptop to make LLMs chat with your emails in < 50 lines of code.Mar 28Mar 28
Shrinivasan SankarinLevel Up CodingNaive Quantization Methods for LLMs — a hands-onImplementing Absolute max and zero point quantization helps learn advanced methods like GPTQ.Mar 15Mar 15
Shrinivasan SankarinLevel Up CodingParameter Efficient Fine-tuning of the Gemma model on a single GPUA guide to fine-tuning the latest Gemma 2B model from Google with your in-house dataset on a single GPU.Mar 5Mar 5
Shrinivasan SankarFine-tuning an LLM — The six-step lifecycleFine-tuning is an art and a methodical process similar to software engineering. It is extremely simplified and portrayed as a cakewalk in…Feb 23Feb 23
Shrinivasan SankarRetrieval Augmented Generation(RAG) — A quick and comprehensive introductionIntroductionFeb 13Feb 13
Shrinivasan SankarLumiere — The most promising Text-to-Video model yet from GoogleU-Net architecture modified to a space-time architecture coupled with MultiDiffusion is LumiereFeb 1Feb 1