Topic - Edition
Seven tech giants have agreed to invest $26 billion alongside Anthropic’s IPO. Their interests extend from Claude’s share price to the chips, cloud services, and memory powering its growth.
Jev needs no training and at most three examples per intent; the open rebuild needs labels. Both still fall into the none-of-the-above trap. The post Scikit-Learn Matched or Beat Jev's Calibration on BANKING77 appeared…
My new book in The Hundred-Page Books series is now on Amazon and Leanpub.
NVIDIA’s GR00T 2 bets that human video can make robot training scale. Its reported adaptation results are promising, but testing gaps and pending model releases leave important questions unanswered.
Australia’s response to the Medicare agent breach includes a proposed safety regulator, mandatory incident reporting and possible criminal-law reform. The legislation is still ahead, but the accountability questions are…
A software tester ran six testing skills on Haiku, Sonnet, and Opus; pushy descriptions added only 5 points, and near-miss requests never fired a skill. The post A Session-Start Hook Raised Claude Code Skill Firing From…
Researchers are getting better at inspecting the machinery behind AI’s answers. A new survey examines the harder question: when does that understanding actually help prevent failures?
A Gemini Flash-Lite step pulls skills, seniority, and Nigeria eligibility from 18 RemoteOK postings, with Python checks and dbt tests guarding every answer. The post How to Validate LLM Output in a DuckDB and dbt…
A guard on negative enrollment days looked correct while the function it called overbilled a customer 11 days, and an invariant test would have caught it. The post Copilot Caught 8 of 8 Diff Bugs With Rules, 0 of 2…
OpenAI's Control Room can stop internal AI research output within 35 seconds of a stop instruction. Here's why OpenAI built it after research agents bypassed security restrictions.
arXiv:2611.09662v1 Announce Type: new Abstract: How does using less conversational history change reader-token use and benchmark scores? We examine this question on 1,100 BEAM questions. We held the retrieval ranking…
arXiv:2610.38402v2 Announce Type: replace Abstract: Predicting medical item difficulty from item text alone remains challenging. Prior work has largely relied on linguistic features or text representations derived from…
arXiv:2611.09061v1 Announce Type: new Abstract: Benchmark gains from reasoning-enhanced large language models encourage the assumption that these models should broadly perform better on complex generative tasks, but…
arXiv:2611.09107v1 Announce Type: new Abstract: Vietnamese punctuation restoration is often evaluated on relatively small benchmarks, limiting conclusions about how architectural choices scale with data volume. We…
arXiv:2611.08779v1 Announce Type: cross Abstract: Persistent-memory systems choose not only which records are retrieved but which prior information can alter current inference. We propose Jacobian-causal memory control…
arXiv:2611.08971v1 Announce Type: cross Abstract: Multi-agent LLM systems increasingly aggregate judgments across model instances, but reasoning about other agents can itself introduce failure. We study a setting in…
arXiv:2605.24371v2 Announce Type: replace-cross Abstract: CT report generation (CTRG) requires models to summarize three-dimensional anatomical context and pathological findings from hundreds of axial slices. Existing…
arXiv:2611.09868v1 Announce Type: new Abstract: Large reasoning models (LRMs) improve task performance by generating longer intermediate reasoning traces, but this substantially increases inference costs and makes…
arXiv:2611.08538v1 Announce Type: cross Abstract: Subliminal learning is a phenomenon in large language models (LLMs) where a teacher passes a hidden preference to a student through training data unrelated to that…
159 items available
