RSI-Inference Bench is a long-horizon benchmark for coding agents that improve the inference stack of large language models. It follows a model through its life in a production engine, vLLM or SGLang: supported on day 0, optimized after launch, and repaired when a regression lands. An agent works on the real engine, on real GPUs, and is graded by running what it built.
Apple's RSI-Inference Bench scores coding agents on 37 real vLLM and SGLang inference tasks via x.com
Each month, the BPP tracks millions of prices, with sources spanning e-commerce sites, credit- and debit-card transactions, retail food surveys, regulatory filings, pharmacy transactions and corporate datasets. The goal is to measure prices in a way that is as close as possible to the approach used by BLS, but adapted to alternative sources.
We construct price indexes at a deeper level of granularity than the CPI, breaking them down by criteria like product, geography and brand.
The Bloomberg Price Project rebuilds the US CPI from millions of prices at almost 700 line items via x.com
Almost half (44%) of our tested claims are now expired in today’s models. Another 34% have weakened significantly, 16% persist, and 6% have strengthened relative to the original claim. Persistence varies by topic. Five of nine reasoning claims and six of eleven sycophancy-and-deception claims are classified as expired. Three of ten claims about models as judges, reviewers, and raters persist or strengthen.
Stanford researchers retest 50 widely cited LLM findings - most fade within 15 months via x.com
Cusack succeeds in delighting his readers with Momo, too, but only if traditional definition of “delight” is expanded to include hard-earned enlightenment. The story piles on enough facts, half-facts, and non-facts to create cognitive dissonance, which only then leads to something delightful: the moment when the reader asks where the truth is. Or more precisely, how much they’re being lied to.
Evan Brown on John Cusack's disorienting graphic novel Momo via x.com
Everything posted this week — 1 link.
One might ask if at some point pure mathematics will somehow be “finished”. And to this we can definitively answer that it will not. Computational irreducibility ensures that there will always be an infinite sequence of mathematical facts—and surprises—to be discovered. And there will also be an infinite number of pockets of computational reducibility, representing new mathematical concepts and structures.
Many beautiful thoughts here in Stephen Wolfram's latest essay on AI's role in pure mathematics.
AI Newsletter — A curated weekly digest of AI links, with pull quotes and commentary.
$50/month $500/year (save 17%)
Weekly Roundup — Get a curated digest of the best links, ideas, and insights delivered to your inbox every week.
Subscribe to Newsletter — Stay up to date with email notifications of new posts.
Discussion