sankalp (dejavucoder)

Alex L. Zhang | A Meticulous Guide to Advances in Deep Learning Efficiency over the Years A very long and thorough guide how deep learning algorithms, hardware, libraries, compilers, and more have become more efficient.

bookmarking here to read this soon
alexzhang13.github.io/blog/2024/ef...

The state of post-training in 2025 A re-record of my NeurIPS tutorial on language modeling (plus some added content).

The state of post-training in 2025: a tutorial on modern post-training
A re-record of my NeurIPS tutorial on language modeling (plus some added content on the high level state of things)
Blog + extra context: https://buff.ly/424VvLm
YouTube: https://buff.ly/40808l5
Slides: https://buff.ly/404jGa9

The Evolution of AI-assisted coding features and developer interaction patterns Yes, I agree that's a fancy title. There have been several developments over the last 7 years in the AI-assisted coding arena. We have gone from simple autoc...

new blog post

Evolution of AI-assited coding features and developer interaction patterns. I go through the history of progression of ai-assisted coding features, talk about how we interact with them and a Gears analogy control vs speed tradeoff

sankalp.bearblog.dev/evolution-of...

First slide deck for NeurIPS is done -- an overview of how I view post-training for applications.
A higher level summary on the key decisions along the way of scoping a problem, choosing a base model, optimization algorithm, etc. (+some thoughts on OpenAI's RL Finetuning).

https://buff.ly/3ZpY5IR

agent orchestrator more like agent pimp

will check this out for synthetic data creation and evals

OpenAI's o1 using "search" was a PSYOP How to understand OpenAI's o1 models as really just one wacky, wonderful, long chain of thought

New post! OpenAI's o1 using "search" was a PSYOP.
How to understand OpenAI's o1 models as really just one wacky, wonderful, long chain of thought.

A fun one trying to communicate intuitions for what large scale RL training does to LLMs. Much more to explore here in 2025!

Wow, this is such a useful resource of industry LLM applications! And filtering via search/tags is so responsive. I was thinking of compiling something like this over the holidays (ala applied-ml) but thanks to @strickvl.bsky.social I can spend the time reading instead ♥️

zenml.io/llmops-datab...

lmao

this is kinda nice

same haha. planning to spend some time on bluesky to check ai discussions and meet mutuals who are more active here

hello sir

we are planning to read this blog blog.dottxt.co

hello world