Loading activity…
@thom_wolf
co-founder @HuggingFace - moonshots
AI
DeepSeek's README offers offline inference for DeepSeek-V4.1-Flash but lists additional systems needed for a complete service.
AI
A user says the result builds its world well but is still weak at understanding the player.
AI
Hugging Face cofounder Thom Wolf says the figure was under $40 million at the end of 2023.
AI
DeepSeek says the model is live on its website, app and API, and optimized for DeepSeekHarness and Tencent CodeBuddy.
Technology
Hugging Face reposted the line and its video link without adding a comment.
AI
A quote post says 'it's time to invent math 2.0' and links to a video, adding '(sounds on).'
AI
finkd says the model runs well on new Mac Studios, crediting work with Apple and John Ternus.
AI
The speaker says he wrote the talk for MPhil students to discuss avoiding intelligence atrophy from AI use.
AI
A post says METR seeks rapid publication, with transparency about assessment scope, redactions and limitations.
AI
One poster argues that the EU lags in building AI and receives almost none of its gains amid scant investment.
AI
Its creators say the database spans 22,617 species, and the model can find genes without a close relative.
AI
A participant invites attendees to say hello to them and two others at the conference.
AI
Mistral says Mistral Large 4 is live via API in public preview now, with open weights due at the end of October. Reuters says the launch is part of Mistral’s push to narrow the gap with leading Chinese open-weight models.
AI
OpenAI says it drew on an independent mathematics advisory group’s advice about how to release the results.
AI
A user said Opus 4.6 repeatedly won by imagining contradictory worlds inside falling water droplets.
AI
Underdog says its Mac assistant stores personal data on-device, with iOS, Windows and Android versions planned.
AI
A user endorses Kevin Buzzard’s proposal to let machines explore math, then begin the journey toward wherever they get stuck.
AI
A post describing a Bloomberg Live clip says InterPositive shot its own dataset on a controlled stage for late-stage training.
AI
A post points to ESM-2’s continued downloads as an example of why the libraries beneath widely used tools need maintenance.
AI
A user asked whether Anthropic was training with “confessions.” The person they addressed said they didn’t think Anthropic was talking about doing that, but had no special knowledge.
