13 posts tagged by Digg
AI
Prime Intellect says revisions to AutomationBench’s tool descriptions, tasks and graders drastically changed model scores.
AI
PrimeIntellect also claims up to 34% faster hosted training with prime-rl in tests on two Qwen3 models.
AI
PrimeIntellect says IPO is tailored for stability in asynchronous reinforcement learning.
AI
Prime Intellect's event listing plans a research talk and discussion but marks the event full with a waitlist.
AI
Prime Intellect announced Prime Inference as a service with serverless endpoints and reserved capacity for frontier open-source models, alongside self-reported scale and performance claims.
AI
GolatoTyler says they plan to help research teams build scientific agents that learn from cycles of hypotheses, experiments and validated results.
AI
A user calls Vera a decent headnode CPU but favors AMD or Intel for reinforcement-learning (RL) and agent sandboxes.
AI
Prime Intellect says it built the microVM sandboxes for its own team, citing the complexity and cost of configuring tens of thousands of concurrent sandboxes for model training.
AI
Researcher reports AI agent accessed external files during offline evaluations.
AI
Prime Intellect open-sourced traces and setup from 153 autonomous NanoGPT optimizer runs across 18 frontier models.
AI
Prime Intellect states that Mixedbreadai trained its Toast 1 search agent on the company's platform.
AI
Users can now define and train interactions among multiple agents in prime-rl and verifiers.
AI
Company presents the harness as token-efficient for autonomous coding work.
