Kangheng Liu

kanghengliu's activity

• All
• Models
• Datasets
• Spaces
• Buckets
• Papers
• Collections
• Community
• Posts
• Upvotes
• Likes
• Articles

updated a collection 1 day ago

CORE-bench v2

Collection

Benchmark for AI agents on scientific reproducibility — mainline (39) and OOD (19) splits derived from Code Ocean capsules. • 2 items • Updated 1 day ago

updated 2 datasets 2 days ago

agent-evals/core-bench-v2-mainline

Viewer • Updated 1 day ago • 39 • 41

agent-evals/core-bench-v2-ood

Viewer • Updated 2 days ago • 19 • 28

published 2 datasets 2 days ago

agent-evals/core-bench-v2-ood

Viewer • Updated 2 days ago • 19 • 28

agent-evals/core-bench-v2-mainline

Viewer • Updated 1 day ago • 39 • 41

upvoted a paper 2 months ago

Towards a Science of AI Agent Reliability

Paper • 2602.16666 • Published Feb 18 • 16

updated a dataset 3 months ago

agent-evals/hal_traces

Updated Feb 1 • 2.72k • 5

updated 2 datasets 5 months ago

agent-evals/hal_traces

Updated Feb 1 • 2.72k • 5

agent-evals/hal_traces

Updated Feb 1 • 2.72k • 5