{"data":[{"project":"default","id":"mcp.select-hand-ae4d07b","benchmark":"mcp.select","split":"hand","system":"hanzo-mcp/ae4d07b","version":"one Kai choice over the names","baseline":true,"study":"kai-tools","embedder":"","reader":"","k":null,"temperature":null,"max_tokens":null,"prompt":"","prompt_digest":"","dataset_digest":"","store_digest":"","facts_digest":"","commit":"ae4d07b","questions":8,"answered":8,"when":"","ended":"2026-10-15T01:23:34Z","by":"","completion":"complete","key":"0f45d43eeedb5d6d3972051d9975a3c374f75247458c81eb2ea500fb44a5eef6","measures":[{"category":"all","metric":"top1","value":0.25,"lo":null,"hi":null,"n":2,"of":8},{"category":"all","metric":"mrr","value":0.4908,"lo":null,"hi":null,"n":null,"of":null},{"category":"all","metric":"abstained","value":2,"lo":null,"hi":null,"n":null,"of":null},{"category":"all","metric":"p50_ms","value":389.11,"lo":null,"hi":null,"n":null,"of":null}],"notes":"eight hand-labelled tasks over eight tools; p50_ms is the handler's wall time against kai-1.2 on dgx's CPU while other training reads ran, not a serving latency","revision":"original","canonical":true,"visibility":"public","trainable":false,"publishable":true},{"project":"default","id":"mcp.select-hand-measured","benchmark":"mcp.select","split":"hand","system":"hanzo-mcp/measured","version":"a tool the task names, then BM25","baseline":false,"study":"kai-tools","embedder":"","reader":"","k":null,"temperature":null,"max_tokens":null,"prompt":"","prompt_digest":"","dataset_digest":"","store_digest":"","facts_digest":"","commit":"863cb8a7a47f8110dfd24c779b989442d0d01c0c","questions":8,"answered":8,"when":"","ended":"2026-10-15T01:23:34Z","by":"","completion":"complete","key":"c7011c640ed326185ddc98cd216bb6af62a1b1f314f066f2e0aa6e0db48cb782","measures":[{"category":"all","metric":"top1","value":0.5,"lo":null,"hi":null,"n":4,"of":8},{"category":"all","metric":"mrr","value":0.5897,"lo":null,"hi":null,"n":null,"of":null},{"category":"all","metric":"abstained","value":2,"lo":null,"hi":null,"n":null,"of":null},{"category":"all","metric":"p50_ms","value":0.1,"lo":null,"hi":null,"n":null,"of":null}],"notes":"eight hand-labelled tasks over eight tools","revision":"original","canonical":true,"visibility":"public","trainable":false,"publishable":true},{"project":"default","id":"mcp.select-t1dev-ae4d07b","benchmark":"mcp.select","split":"t1dev","system":"hanzo-mcp/ae4d07b","version":"one Kai choice over the names","baseline":true,"study":"kai-tools","embedder":"","reader":"","k":null,"temperature":null,"max_tokens":null,"prompt":"","prompt_digest":"","dataset_digest":"","store_digest":"","facts_digest":"","commit":"ae4d07b","questions":300,"answered":300,"when":"","ended":"2026-10-15T01:23:34Z","by":"","completion":"complete","key":"2f779f80bd03b757d3331274b84adae11dc8f60eeab6c4fa3bc47f808cd7060b","measures":[{"category":"all","metric":"top1","value":0.6567,"lo":null,"hi":null,"n":197,"of":300},{"category":"all","metric":"mrr","value":0.8052,"lo":null,"hi":null,"n":null,"of":null},{"category":"all","metric":"abstained","value":43,"lo":null,"hi":null,"n":null,"of":null},{"category":"all","metric":"p50_ms","value":479.13,"lo":null,"hi":null,"n":null,"of":null}],"notes":"supplied tool lists from Kai's T1 tools6 dev split (manifest 48be841f); p50_ms is the handler's wall time against kai-1.2 on dgx's CPU while other training reads ran, not a serving latency","revision":"original","canonical":true,"visibility":"public","trainable":false,"publishable":true},{"project":"default","id":"mcp.select-t1dev-measured","benchmark":"mcp.select","split":"t1dev","system":"hanzo-mcp/measured","version":"a tool the task names, then BM25","baseline":false,"study":"kai-tools","embedder":"","reader":"","k":null,"temperature":null,"max_tokens":null,"prompt":"","prompt_digest":"","dataset_digest":"","store_digest":"","facts_digest":"","commit":"863cb8a7a47f8110dfd24c779b989442d0d01c0c","questions":300,"answered":300,"when":"","ended":"2026-10-15T01:23:34Z","by":"","completion":"complete","key":"89acd7d6ceb57dd12fa368f9ba98f8ab3930a7459e9dbea1b0c27e19a901e89e","measures":[{"category":"all","metric":"top1","value":0.7333,"lo":null,"hi":null,"n":220,"of":300},{"category":"all","metric":"mrr","value":0.8408,"lo":null,"hi":null,"n":null,"of":null},{"category":"all","metric":"abstained","value":1,"lo":null,"hi":null,"n":null,"of":null},{"category":"all","metric":"p50_ms","value":0.16,"lo":null,"hi":null,"n":null,"of":null}],"notes":"supplied tool lists from Kai's T1 tools6 dev split (manifest 48be841f)","revision":"original","canonical":true,"visibility":"public","trainable":false,"publishable":true}],"total":4,"completion":{"complete":4,"partial":0,"inconsistent":0,"unknown":0},"gaps":[{"field":"by","runs":4},{"field":"dataset_digest","runs":4},{"field":"embedder","runs":4},{"field":"facts_digest","runs":4},{"field":"k","runs":4},{"field":"max_tokens","runs":4},{"field":"prompt","runs":4},{"field":"prompt_digest","runs":4},{"field":"reader","runs":4},{"field":"store_digest","runs":4},{"field":"temperature","runs":4},{"field":"when","runs":4}]}