International Olympiad in AI opens an official medal track for AI systems
Within 24 hours of the Day One and Day Two session, registering AI systems will be able to see their private percentile scores on each task. At this point, they have 48 hours to make a decision on whether they wish to verify their scores, or appear on the leaderboard pseudonymously. Only verified scores with an organization’s name are eligible for official IOAI Gold, Silver and Bronze medals.If opting to verify scores, then over the subsequent few weeks, the registering organization must submit confidential execution traces of both sessions.
- ▪Within 24 hours of the Day One and Day Two session, registering AI systems will be able to see their private percentile scores on each task.
- ▪At this point, they have 48 hours to make a decision on whether they wish to verify their scores, or appear on the leaderboard pseudonymously.
- ▪Only verified scores with an organization’s name are eligible for official IOAI Gold, Silver and Bronze medals.If opting to verify scores, then over the subsequent few weeks, the registering organization must submit confidential execution t
Hacker News (AI / LLM) files mainly under ai. We currently carry 2,346 of its stories.
Opening excerpt (first ~120 words) tap to expand
Within 24 hours of the Day One and Day Two session, registering AI systems will be able to see their private percentile scores on each task. At this point, they have 48 hours to make a decision on whether they wish to verify their scores, or appear on the leaderboard pseudonymously. Only verified scores with an organization’s name are eligible for official IOAI Gold, Silver and Bronze medals.If opting to verify scores, then over the subsequent few weeks, the registering organization must submit confidential execution traces of both sessions. These traces should include timestamped AI model inputs and outputs, tool-call inputs and outputs, tools-call definitions, subagent execution traces, code execution outputs, libraries used, and any human-written system prompts.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at Hacker News - Newest: ""AI" "LLM"".
Discussion
0 commentsMore from Hacker News - Newest: ""AI" "LLM""
