shreyansh26
Re:
on Thursday February 19, 2026 @10:47AM (#47075062)
What’s covered:
1. Hierarchical scans: block-local scan → write block totals → scan totals → carry-in add 2. Single-pass scans: the "domino" idea, and why naive inter-block propagation can stall / deadlock without the right coordination 3. Decoupled lookbacks: how modern single-pass scans coordinate across blocks safely 4. Warp-window lookback optimization: scanning lookback metadata in warp-sized chunks (and why it helps)
I also include H100 timings and compare against CUB for context.
Re:Understanding Multi-Head Latent Attention (From DeepSeek)
on Monday January 26, 2026 @11:27AM (#46767671)
A short deep-dive on Multi-Head Latent Attention (MLA) (from DeepSeek): intuition + math, then a walk from MHA → GQA → MQA → MLA, with PyTorch code and the fusion/absorption optimizations for KV-cache efficiency.
Deriving the gradient for the backward pass of Layer Normalization
GTC'25 Notes: CUDA Techniques to Maximize Memory Bandwidth – Part 1
FlashAttention in PyTorch
Re:FlashAttention in PyTorch
on Wednesday June 14, 2023 @12:38AM (#36321509)
A simplified implementation of FlashAttention in PyTorch for educational purpose.
Re:A hyper-fast local vector database for use with LLM Agents
on Sunday April 16, 2023 @11:00PM (#35596414)
Haha, so all one has to do is to add some fancy terms like LLMs, vectors, hyper-fast and 300+ Github stars. What a a joke lol!
Understanding FlashAttention
Re:OpenAI Is Now Everything It Promised Not to Be: Closed-Source, and ...
on Wednesday March 01, 2023 @09:15PM (#34991087)
OpenAI is actually the one which has made AI so open and accessible for all. Perhaps not open source but all the developers building cool AI products these days are mostly due to OpenAI.
Re:Ask HN: What are some good resources on Recommender Systems?
on Friday November 25, 2022 @10:50PM (#33749720)
That's helpful. Thanks!
Ask HN: What are some good resources on Recommender Systems?
I am looking for good technical resources to study recommender systems in depth. Anything from books, courses to blogs would be helpful. Would also be better if I could follow along with code and implement stuff.
Re:Including “And. And. And. And. And.” in a Google doc causes it to c...
on Thursday May 05, 2022 @04:40PM (#31278143)
Reproduced on Edge on Windows 11 as well. What a bug! Really need to know the root cause of this.
Re:Ask HN: Which book can attract anyone towards your field of study?
on Monday March 28, 2022 @05:08AM (#30828671)
Someone post something about astronomy also please!