Loading activity…
@shannonzshen
PhD @MIT_CSAIL / Prev @AnthropicAI Fellow @metaai @allen_ai Check http://swecraft.world http://szj.io/ces
AI
The team says Gemini-3.8-Flash achieved a 23 percent speedup across three worlds, while GPT-6-Astra scored worst among tested frontier models.
XAI
An organizer says the event with E14 Fund is set to start at 8:30 p.m.
AI
A poster says they provide day-one support for CLM + Pi-agent and shares a command to try it: `pi install npm:@lolipopshock/pi-clm`.
AI
A post about the project claims a 65% score gain with the same compute on a 24-hour multi-repository agent-swarm task.
AI
@humansand calls Persimmon the first large-scale model designed to realistically simulate how people talk and interact.
AI
The user alleges that “explicit feedback”—including thumbs-up/down ratings and choosing a preferred response in OpenAI’s interface—can permit chat collection even after a data-collection opt-out.
AI
Questions incentive design after observing both hacking and collaboration among agents.
AI
Shannon Shen shares early-stage project simulating GitHub with agents to study collaboration.
AI
The technique modifies randomness during token selection to create detectable signals.
