gkapur

12 ms·

1.
Landscape and Perspective on Recursive Self-Improvement from a Neolab(poetiq.ai)
1 points by gkapur 1mo ago|0 comments
2.
by gkapur 1mo ago
Currently, the most efficient kernels come from vendor libraries like cuBLAS or hand-optimized libraries like FlashAttention. Compiler-generated kernels (TVM, Hidet, Inductor, etc.) can generate more efficient kernels for some specific oper
3.
by gkapur 1mo ago
Currently, the most efficient kernels come from vendor libraries like cuBLAS or hand-optimized libraries like FlashAttention. Compiler-generated kernels (TVM, Hidet, Inductor, etc.) can generate more efficient kernels for some specific oper
4.
Exploring FlashAttention-3/4 optimizations on RTX GPUs(riftstack.ai)
2 points by gkapur 2mo ago|1 comments
5.
by gkapur 2mo ago
I was curious whether any of the FA-3/4 optimizations transfer to RTX GPUs. vLLM/SGLang attention falls back to FA-2 on consumer cards (FA-3 and FA-4 are datacenter-only), so I wanted to know if there's any performance left o
6.
by gkapur 2mo ago
Peter Fenton invested in the company when it was called Infra.App originally (I think it's still online: https://infra.app/ ). It was an access management product, then it became desktop Kubernetes product, both they p
7.
by gkapur 2mo ago
The story of Reflection AI is supposedly that the company was faffing and failing at winning in the coding agent space, but was introduced to Jenson, who suggested they build an open-weight model and said he would fund it. That turned into
8.
by gkapur 2mo ago
If they have a really seamless fine-tuning experience and maybe can help you extract the data you need to FT (which is one of the big challenges in actually getting fine-tuning democratized), maybe you would use it because "Tinker&quot
9.
by gkapur 2mo ago
It could be but there are a host of companies going after open weights models: Arcee, Reflection, Llama (TBD on Meta's focus on closed-source versus open-source), etc. That said, the fine-tuning API + open weight model at least is a se
10.
Tutorial: Algebraic Foundations Powering FlashAttention(riftstack.ai)
10 points by gkapur 2mo ago|1 comments
11.
by gkapur 2mo ago
I'm writing a short series of tutorials on FlashAttention: from theory to efficient CUDA kernels. Part 1 is the theoretical foundation. It walks through a modern algebraic formalism showing that FlashAttention is an associative operati
12.
by gkapur 4mo ago
On the limitation side: Do you think this would scale to larger transformer models with more parameters per layer? How would this work with MOE models or sparse models?
13.
Surfacing a 60% performance bug in cuBLAS(kernelspace.substack.com)
10 points by gkapur 5mo ago|0 comments
14.
A Principled ML Compiler Stack in 5k Lines of Python(kernelspace.substack.com)
1 points by gkapur 5mo ago|0 comments
15.
Agentic systems redraw the Pareto frontier on ARC-AGI(poetiq.ai)
7 points by gkapur 10mo ago|1 comments
16.
by gkapur 11mo ago
Adding to the prior comments as my intuition matched yours, there’s a nice Reddit thread that gives some context into how it can be faster even if you require exact matches: https://www.reddit.com/r/LocalLLaMA/s&#x
17.
by gkapur 1y ago
Congratulations to their team. SQLGlot is a really powerful tool that a lot of companies use so a huge contribution to the OSS community so hopefully it continues to be supported and gets better and better!
18.
by gkapur 1y ago
There was also Wing cloud (fka Monada) and there’s Mojo by Modular ( https://www.modular.com/mojo .) Feels like two types of companies raised money: - Companies trying to couple the cloud with a programming language. - More r
19.
by gkapur 1y ago
If you are running things locally (I would think especially on the edge, whether on not the LLM is local or in the cloud) this would matter. Or if you are running some sort of agent orchestration where the output of LLMs is streaming it cou
20.
by gkapur 1y ago
I’m convinced I get more “deals” (temporary discounts) from Uber without Uber One/after canceling it, which offsets the benefits from Uber One. I don’t see those deals on Uber Eats so it feels like the real value of Uber One is for hea
21.
by gkapur 2y ago
Today there are so many other solutions: Stytch, Descope, PropelAuth (For B2B companies), and others. VCs went a bit ham on this category when Auth0 got bought. I sense that the general thought process was: Auth0 multi-billion dollar compan
22.
by gkapur 2y ago
Basically people are constantly calculating metrics based on existing tables. Think something as simple as a moving average or the sum of two separate columns in a table. Once upon a time you would set up a cronjob and populate these every
23.
by gkapur 2y ago
Thanks for the transparency and thoughts!
24.
by gkapur 2y ago
What’s interesting is how much it contrasts with TechCrunch’s story: ‘Most of Command AI’s 30-person, San Francisco-based team will be joining Amplitude. Command AI’s co-founder and CEO James Evans wouldn’t reveal the terms of the deal, but
25.
by gkapur 2y ago
Interestingly, according to Axios, the price was pretty limited: "Amplitude (Nasdaq: AMPL) acquired CommandAI, an SF-based software user experience startup, for $20m (net of cash). CommandAI (fka CommandBar) had raised around $23m from
26.
by gkapur 2y ago
Not an expert in the space at all and it does seem like people are exploring new file and table formats so that is really cool! How does this compare to Lance ( https://lancedb.github.io/lance/ )? What do you think the k
27.
by gkapur 2y ago
Been watching this episode unfold on Twitter and has read about Matt’s domain hijacking of thesis, etc. It seems to me like Matt is the type of person who likes to hide behind character and other ad hominem attacks rather than addressing ac
28.
by gkapur 2y ago
Not really. This is more like SQLite or DuckDB for vector databases (on disk.) Chroma is more like redis for vector databases (in memory.) We have seen similar products in the olap space, as well, ie. Clickhouse local.
29.
by gkapur 2y ago
This whole thing comes off as tone-deaf and deceptive even to me (who is all for COSS monetizing.) Warpstream was sponsoring Benthos, it sounds like they didn't get a great heads up of this happening, which makes the project owner sou
30.
by gkapur 3y ago
I empathize for the co-founder who was CTO and became CEO. I imagine some of the challenges come from the fact that there was a big chunk of equity owned by the original founding CEO. As the remaining co-founder, I can imagine feeling like

More ›