Loading activity…
@lateinteraction
prof @MIT CSAIL | ColBERT, http://DSPy.ai (@DSPyOSS), GEPA, RLMs, Pedagogical RL
AI
The reported MATH-500 score rose from 7.9% to 48.1% with the same sampling settings.
AI
A Claude user jokingly thanks it for its hard work but wishes it had said 'our laptop'.
AI
A developer credits retrieval-augmented n-gram hints for the claimed 83.2% cut in time.
XAI
One post says researchers pivoted as some problems were solved or became irrelevant and new ones took their place.
AI
One poster calls Anthropic early on reinforcement learning and tool use; a reply points to a LaMDA paper.
AI
A user links a PDF they describe as the roadmap and attributes its release to copyright litigation.
AI
A user quips that the RLMs are “having fun,” without explaining what prompted the joke.
AI
A keynote listener highlights the idea that stronger models will need longer prompts because people will demand more of them.
AI
Perplexity says embedding images and rendered pages directly preserves tables, figures and layout that text extraction drops.
AI
A post attributes the proposal to Meta's team and reports Qwen3-8B average accuracy rising from 30.76% to 65.97%.
AI
One commenter argues that LLMs can fit the ‘stochastic parrot’ analogy and still be powerful and useful, even with uneven capabilities.
AI
The links exposed credentials and attack details that could have let anyone who found them compromise Hugging Face, says investigator Jeff Ladish.
AI
The release announcement also introduces ReAnchor, a new optimizer described as specifically for calibrating outputs with confidence.
AI
One commenter joked about “like 7%” of Prime Intellect’s technical staff; a reply added “a whopping 3%” of the “oasys/dspy gang.”
AI
An early supporter calls an unnamed release “perfectly good (even great)” but argues that the surrounding praise has become excessive.
AI
Creator Diogo Almeida claims Jev is 20–200x faster and 40–400x cheaper, with free output tokens. Investor DCVC says it led TypeSafe’s $40 million seed round.
AI
One post argues that favoring math and coding over clear, delightful writing has badly hurt perceptions of AI labs.
AI
The models are good enough to be daily tools for much of what they want to do, the user says, but not superhuman at any broad skill they personally care about.
AI
Posts say 25 Fields Medal winners signed a declaration warning of a mismatch between AI companies and mathematics. Supporters stress understanding and scientific communities; critics say AI proofs can still yield insights.
AI
The host’s September 9 announcement said the discussion would examine what would count as a verified solution and what such a result could mean for artificial general intelligence.
