🏗️ Building on HF

Dipankar Sarkar PRO

dipankarsarkar

4 642 119

https://www.dipankar.cc

AI & ML interests

Building the AI-native stack. Agents as infrastructure, safety as architecture, performance as plumbing. I publish the receipts: papers, datasets, demos.

Recent Activity

reacted to lorraine2's post with 🔥 35 minutes ago

🧠 New NVIDIA paper: WorldTrace 🧠 Long-horizon video world models can store prior scenes in their KV cache and still fail to retrieve them once RoPE offsets leave the training range. We show this is an addressability problem, not just a compression problem. WorldTrace assigns compressed memory slots fixed, in-distribution positions. WT-Field improves temporal consistency by +15.5%, while WT-Landmark improves episodic recall by +19.5% on LoopBench, without retraining. 🏆 Best Paper + Oral, ICML 2026 From Frames to Stories Workshop 🔎 Project page: https://research.nvidia.com/labs/sil/projects/WorldTrace/ 📄 Paper: https://huggingface.co/papers/2608.07408 Led by Xindi Wu, with Sven Elflein, James Lucas, Olga Russakovsky, Laura Leal-Taixé, Despoina Paschalidou, Aljoša Ošep, and me.

reacted to bshepp's post with 🔥 about 2 hours ago

I pulled a dead 2013 Butterfly Labs "Jalapeno" SHA-256 mining ASIC out of a drawer and built a modern Python toolkit for it. Then I hit the wall every honest hardware project should. I "found" undocumented serial commands the mining software (cgminer) defines but never sends, including a persistent NVRAM scratchpad that survives power cycles. I wrote my name and the repo URL into the silicon; it's still there. Then I checked prior art. All of it is in Butterfly Labs' own 2012 protocol spec and their open source firmware. Rederivation, not discovery. I even almost filed a "bug" against cgminer before a last look at its code showed it was right and I'd misread it. Twice, the "gotcha" was me not reading carefully. Why it was still worth it, the firmware source can't tell you the chip still works. So I measured it, model free: 1. Four hours of continuous work, zero compute errors, fully deterministic. 2. The winning nonce count is Poisson(~1), the chip scans the whole 2^32 nonce space per job. 3. Thermally over built: it won't error even with the fan off (~41C max on a desk). The one genuinely new thing: a dead-core detector. It flags a dead engine as a cold band in the nonce histogram. It can't map the healthy engine partitions (they sum to uniform), only localize the dead ones. The honest move, go check whether it's already known, costs you a discovery and gives you the truth. Better trade every time. Code (MIT), data, and the full writeup: https://github.com/bshepp/bfl-asic Written with AI assistince.

upvoted a paper about 5 hours ago

Metric Match: A Subset Selection Approach to Evaluating LLM Judge Reliability

View all activity

Organizations

Posts 2

Post

437

Your issue tracker is in the wrong place.

It lives on a server. Your code lives in git. So every time an agent picks up work it makes an API call, burns a token, fights a rate limit, and still cannot see what the other agent just did.

Move the issues into the repo. Append-only event log in git refs. Branches when you branch, merges when you merge, CRDT so two agents never conflict. No server, no database.

The coordination signal that PR-level telemetry misses lives before the pull request. The paper, and a live demo running the real tool:

Before the Pull Request: Mining Multi-Agent Coordination (2606.19616)
https://huggingface.co/spaces/neullabs/grite

If your agents share a repo, where does their shared state actually live right now?