Unverified50% confidenceBenchmarkExact time
The paper evaluates on 26 real robot tasks and 4 policies, and frozen encoder features achieve the highest score in 89 of the 104 policy-task combinations.
1
Sources
50%
Confidence
Long-term
Relevance
11/9/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedIn real-world tests covering twelve tasks and three robot instances, the AMP1 agent succeeded zero-shot in 104 of 108 trials.72% similarUnverifiedPilot experiments had three coding agents and eight models each run eight 20-minute single-pass tasks, with label-aware F1 showing extremely high variance ranging from 0.143 to 0.892.71% similarUnverifiedThe developer ran 1932 test runs, using 100 realistic tools (including deliberately confusable tools such as pdf_compress, pdf_optimize, image_compress), with tool counts in 8 tiers from 5 to 100, running each condition twice, testing 18 tasks (including 10 ambiguous tasks).70% similarUnverifiedThe sealed task pack contains 30 tasks, divided into 20 ordinary tasks and 10 negative control tasks.69% similarUnverifiedThe Cognition evaluation suite contains 132 test tasks covering categories such as security patches and performance optimization, was built with the participation of more than 20 open-source maintainers, and each task takes over 40 hours.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1152664API
curl https://kongchang.com/api/v1/knowledge/claims/1152664MCP
get_claim(id=1152664)