1. Kernel optimization is essential for efficient large language model (LLM) training and inference on Ascend NPUs. Existing Ascend kernel benchmarks primarily focus on kernel generation, with less emphasis on the ability of agents to optimiz…

  2. Behavioral clusters learned over user or item interactions drift as behavior evolves: some fragment into sub-behaviors, some converge, some disappear, and new patterns emerge. Two paradigms dominate the maintenance of evolving clusterings.

  3. Evaluations of multi-agent language models often compare communicating architectures against a single greedy pass. This practice confounds coordination with sample multiplicity and inference compute.

  4. Autonomous crystal-structure determination from powder X-ray diffraction (pXRD) remains a key bottleneck for self-driving materials laboratories. We formulate the task of pXRD structure determination as a verifiable agentic search problem:…

  5. Can sustained gain-loss (valence) framing sway an LLM agent toward market manipulation? We place Gemma 3 12B into the seat of a defendant in the London silver-fixing litigation, in a market simulation built from real case transcripts.

  6. Hallucination detection for large language models is dominated by methods evaluated on 7B-70B research models with abundant sampling budgets and exact-match or ROUGE correctness labels. The small models practitioners actually ship on-devic…

  7. LLM memory systems often delete facts classified as stale. We show that this hard-filtering step can reduce downstream accuracy when resolver errors delete query-critical facts.

  8. What music does Gemini Decision 1 like? Taking a page out of Technology Connections’ book, I’m making a quick blog post as part of No Effort November.

  9. I run a business mostly solo with Claude and agents doing a lot of the work. codebase, infra, multiple business contexts, so the top layer, me, gets full with output.

  10. Use Claude daily for work and run multiple projects for each client. I have quite a few connectors active, mostly fire crawl, gmail, hubspot ect.

    +8 more
  11. I know that many folks here post their personal emails in the Who's for Hire monthly pages and other places, and want to remind you that this public site gets views from folks with ill intent (or their bots/agents) as well. I use unique em…

  12. I built those childhood lift-the-flap book using claude in a single prompt. I gave one prompt describing the book in claude code using fable 5.2 and it wrote the whole page: painting on canvas, the interactions and the sounds, all generate…

  13. redacted Removes secrets from tool output before your AI coding assistant reads it. What you see Tool output: DATABASE_URL=postgres://admin:secret@db.example.com:5432/prod AWSACCESSKEY_ID=AKIAEXAMPLEKEYID0001 SMTP_PASS="Qf7mZp2kLw9vTx4nRb2…

  14. Claude is incredible. It has become invaluable to me and my organization.

  15. Everyone is buying a Mac Mini for their agents, or better, giving them a Linux box (see https://x.com/theo/status/2090528543746965991). I found this quite tedious to setup every time, so I built a "custom" nixOS image to create a repeatabl…

  16. I used Chatgpt from when it first came out. And I guess first mover advantage and sunk cost fallacy just made me stay.

  17. I hit the pro usage limit but I subscribed in iOS. How do I upgrade to max via web?

  18. Can we just “vibe-engineer” actual analog circuits in hardware? When GPT-6 Astra and Claude Opus 5.5 were released in September 2026, it became quite clear that the models’ capabilities in spatial reasoning and engineering in general had s…

  19. JA4 is a TLS ClientHello digest, not a client ID. How one digest on BetterReads covered seven rotating Windows Chrome UAs, and how I turned that into Vercel Firewall rules without Observability Plus.

  20. They list that the support bot Fin will connect/route you to one but its telling me it cant, and that it isnt possible? Just stating its not a thing it can do.

  21. I use some prebuilt prompt that I found from web or sometimes I set memory in claude to write cv and cover letter for me. But what to do, for instance, I have done some work before but I have no background experience directly in organisati…

  22. Howdy! I've been working on an open-source tool that helps lighten the burden of reviewing what your coding agents write.