HackerTrans
TopNewTrendsCommentsPastAskShowJobs

nsomani

no profile record

Submissions

[untitled]

1 points·by nsomani·2 miesiące temu·0 comments

Autoformalization and the Future of Math Research

neelsomaniblog.com
2 points·by nsomani·5 miesięcy temu·0 comments

Show HN: GPT-Erdos – the results of GPT 5.2 Pro on the Erdos problems

ocf.berkeley.edu
1 points·by nsomani·6 miesięcy temu·0 comments

Show HN: A website that auctions itself daily

thedailyauction.com
43 points·by nsomani·6 miesięcy temu·19 comments

Show HN: Symbolic Circuit Distillation: prove program to LLM circuit equivalence

github.com
16 points·by nsomani·6 miesięcy temu·2 comments

The endgame for mechanistic interpretability is formal methods

neelsomaniblog.com
2 points·by nsomani·6 miesięcy temu·0 comments

Show HN: Do Models Represent Epistemic Stance?

github.com
3 points·by nsomani·7 miesięcy temu·0 comments

Intro to Routing: Mixture-of-Experts and Expert Choice

neelsomaniblog.com
1 points·by nsomani·8 miesięcy temu·0 comments

Show HN: KV Marketplace – share LLM attention caches across GPUs like memcached

github.com
2 points·by nsomani·8 miesięcy temu·1 comments

A Minimal Route to Transformer Attention

neelsomaniblog.com
2 points·by nsomani·9 miesięcy temu·0 comments

Show HN: Cuq – Formal Verification of Rust GPU Kernels

github.com
94 points·by nsomani·9 miesięcy temu·63 comments

Killing the GIL: How to Use Python 3.14's Free-Threading Upgrade

neelsomaniblog.com
2 points·by nsomani·9 miesięcy temu·0 comments

Is GRPO Broken?

neelsomaniblog.com
2 points·by nsomani·9 miesięcy temu·0 comments

Intro to RL: Off-Policy Methods

neelsomaniblog.com
2 points·by nsomani·9 miesięcy temu·0 comments

What You Didn't Learn in Berkeley CS 188: Intro to RL

neelsomaniblog.com
2 points·by nsomani·9 miesięcy temu·0 comments

comments

nsomani
·6 miesięcy temu·discuss
Appreciate that. I didn't realize how derivative this was when I built it! I was just trying to build something cool with the Codex SDK.
nsomani
·6 miesięcy temu·discuss
They cannot edit the iFrame itself. The user is allowed to edit the contents within the iFrame.
nsomani
·6 miesięcy temu·discuss
The Codex agent is only given tools to edit the single HTML file that displays on the homepage. The page is on a separate domain, so there's no cookie sharing, and the iFrame is in a sandbox. That said, the biggest risk is social engineering attacks.
nsomani
·6 miesięcy temu·discuss
There are two examples provided - quote matching and bracket closing.
nsomani
·8 miesięcy temu·discuss
Hi all - this is a small research prototype I built to explore cross-GPU reuse of transformer attention states.

When inference engines like vLLM implement prefix/KV caching, it's local to each replica. LMCache recently generalized this idea to multi-tier storage.

KV Marketplace focuses narrowly on the GPU-to-GPU fast path: peer-to-peer prefix reuse over RDMA or NVLink. Each process exports completed prefix KV tensors (key/value attention states) into a registry keyed by a hash of the input tokens and model version. Other processes with the same prefix can import those tensors directly from a peer GPU, bypassing host memory and avoiding redundant prefill compute.

Under optimistic conditions (perfect prefix importing), the prototype shows about a 15% reduction in latency and throughput gains without heavy tuning. The code is intentionally minimal (no distributed registry, eviction, or CPU/disk tiers yet) but it's a prototype of "memcached for attention."

I thought others exploring distributed LLM inference, caching, or RDMA transports might find the repo useful or interesting.
nsomani
·9 miesięcy temu·discuss
Oh really? I can't find anything about the memory model online. I'm not sure what's the best way to do this, but if there's a way for us to get in contact, I'd be interested in adjusting the project so it's developed in the most ergonomic way possible. I'm chatting with a couple of universities and I might issue a research grant for this project to be further fleshed out, so would be keen to hear your insights prior to kicking this off. My email is neel[at]berkeley.edu.
nsomani
·9 miesięcy temu·discuss
I think I've just spent too much time reading the word "CUDA" that I read "cu" as "koo", lol.
nsomani
·9 miesięcy temu·discuss
That instinct is right. cuTile would be easier to parse but harder to reason about formally.
nsomani
·9 miesięcy temu·discuss
Oh wow, honestly this caught me off guard - I've been pronouncing it "kook" in my head the whole time.
nsomani
·9 miesięcy temu·discuss
Hi all, this is a small research prototype I built that connects Rust's MIR (Mid-level IR) to Coq, the proof assistant used for formal verification.

cuq takes the MIR dump of a Rust CUDA kernel and translates it into a minimal Coq semantics that emits memory events, which are then lined up with the PTX memory model formalized by Lustig et al., ASPLOS 2019.

Right now it supports:

* a simple saxpy kernel (no atomics)

* an atomic flag kernel using acquire/release semantics

* a "negative" kernel that fails type/order checking

The goal isn't a full verified compiler yet. It's a first step toward formally checking the safety of GPU kernels written in Rust (e.g. correct use of atomics, barriers, and memory scopes).

Happy to hear thoughts from folks working in Rust verification, GPU compilers, or Coq tooling.