Dipkumar Patel — Blog

I wanted an RL task I could grade with a plain Python function [1]Lambert et al. (2024) · Tülu 3 (RLVR) · arXiv:2411.15124: no judge model, no human labels.

What Are Managed Agents?

IndiGo is the budget airline six of every ten Indian flyers already use.

TLDR: Checklist Make prompt caching actually work: stabilize the prefix and verify hit rates.

Modern agents can write code, call APIs, draft a memo, and pass a benchmark.

Karpathy recently released nanochat repo which cotains code for training the best ChatGPT under $100.

Creativity is a luxury.

OpenAI GPT-5 Router is like Apple removing headphone jack.

Why Your Retriever is Failing and How Context Can Save It Imagine asking "I want to buy apple" – do you mean Apple Inc. stock, the latest iPhone, or simply fruit?

Improving Retrieval in RAG (via Recall, Precision, and NDCG) Introduction Retrieval-Augmented Generation (RAG) is the superhero sidekick that grounds your Large Language Model (LLM) in cold, hard facts.

Amazon Bedrock is a fully managed service that makes high-performing foundation models (FMs) from leading AI startups and Amazon available for your use through a unified API.

Essential Fields Be it relational or not, every table should have these 5 fields: created_at (default now()) updated_at (default now()) deleted_at (default null) created_by (not null) updated_by (not null) Just to be clear, every table should have these 5…

The common optimization trick for speeding up transformer inference is KV caching 1 2.

Biweekly-66 (27th Nov, 2021) 2085.

run local server hugo server -D Create New Post hugo new content/posts/{post-name}.md Hugo build/export the site hugo -d ..