Vincent Weisser@vincentweisser · 2026/08/26观点输出RT @samsja19: Today we are shipping prime-rl 0.9.0
this might be the prl release I am the most proud, we shipped some of my favorite feat…
AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。
Vincent Weisser@vincentweisser · 2026/08/26观点输出RT @rasdani_: nasty sandbox escape found by @xeophon
we have the best sandboxers on our team and with upcoming cutting-edge sandbox featu…
AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。
Vincent Weisser@vincentweisser · 2026/08/25观点输出RT @iScienceLuvr: Looks like they have released their technical report.
"Prime Agent: A Self-Improving RLM Harness"
https://t.co/Lvh3FLuY…
AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。
Vincent Weisser@vincentweisser · 2026/08/25观点输出RT @sethkarten: I constantly worry that somehow agents are escaping sandboxes to access privileged information during evals.
@xeophon tha…
AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。
Vincent Weisser@vincentweisser · 2026/08/25赛道影响@perplexity_ai @nvidia very cool work!!
AI：偏主题投资表达，释放对基础设施与新叙事的偏多信号。
Vincent Weisser@vincentweisser · 2026/08/25观点输出RT @PrimeIntellect: As models become more capable, reward hacks become an increasingly serious problem.
During a controlled experiment, we…
AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。
Vincent Weisser@vincentweisser · 2026/08/25观点输出@sonnet_xu We are happy to power funds with compute for their portcos w/ https://t.co/BfmlaxxJaE 🫡
AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。
Vincent Weisser@vincentweisser · 2026/08/24观点输出RT @ar0cket1: a good example of using LLM/Agentic judges in teaching subjective or other hard to define things such as taste (which is pret…
AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。
Vincent Weisser@vincentweisser · 2026/08/24观点输出RT @soleio: The future belongs to designers who build.
AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。
Vincent Weisser@vincentweisser · 2026/08/24观点输出@maria_rcks thanks!! what would it take to reach s+++? ;)
AI：延续个人公开立场，强调长期主义、执行效率与行业方向判断。
评论
暂无评论