Loading activity…
@iscienceluvr
CEO @SophontAI | Founder @MedARC_AI | PhD at 19 (2023) | ex Research Director Stability AI | Biomed. engineer @ 14 | TEDx talk➡https://bit.ly/3tpAuan
AI
The reported MATH-500 score rose from 7.9% to 48.1% with the same sampling settings.
AI
In the preprint's experiments, looping middle layers gave better returns under training-compute limits, while more repetitions could undermine gains.
AI
X user matt_slotnick jokes about Stewart having Zitron on as an “AI expert.”
AI
OpenAI says its claimed refutation of the Exponential Time Hypothesis was formalized in Lean and made public.
AI
One user expressed surprise at the claim; another, quoting the same post, called it “so cool.”
Science
Primitive Research says 49 candidates passed initial screening after searching TESS light curves from more than 19 million stars.
AI
In an October 24 post, Shihipar said he plans a break before starting something new.
AI
Block says the model was trained only on human conversation and can listen while speaking.
XAI
A user linked to an arXiv paper they described as fixing the issue; another user praised the work.
XAI
A user recalls the Yi family of LLMs fondly; a reply wonders whether 01 AI is still big in China.
AI
MIT's course page outlines four areas: context, agent harnesses, models and increasingly complex synthetic tasks.
AI
A user says they've started waiting for a model release before building a new feature.
AI
OpenAI says thousands of agents linked mouse-brain gene-expression data to literature in 32 minutes, proposing a testable cholesterol-transport mechanism.
AI
A person training models at Sophont says the pathology work also infers biological pathway activity from routine tissue slides.
AI
The team behind Holon-S1 says it beat two other models at predicting spatial gene activity in a controlled comparison.
XAI
A user tells viewers they “suck at prompting” and shares a video he says he enjoyed making.
XAI
A post argues small teams could develop medicines, materials and energy technologies that now require entire research organizations.
AI
A user says TypeSafe AI tried to coin “System One Models,” but sees Jev called a “decision model” instead.
AI
A user asks for help choosing between what they see as two NeurIPS conferences in 2026.
AI
Quoting a post captioned 'recession indicator,' a user said it explained what was exciting about GLM 5.3 Flash fine-tune.
