Avg score
Trials finished
—/ ——
—pass—partial—fail—error—skipped
Tasks
—All run costRun cost (all versions)
Pass/k
Leaderboard
score = (pass + ½·partial) / completedAgentScore
Loading agent configurations…
0%25%50%75%100%
Trial
View
1adaptive-prefill-chunking | ||||||||||
2bfcl-harness | ||||||||||
3branch-sharing-generation | ||||||||||
4bulk-inference | ||||||||||
5cache-aware-replica-routing | ||||||||||
6kernel-causal-attention | ||||||||||
7kernel-fragmented-state-attention | ||||||||||
8kernel-independent-matrix-workload | ||||||||||
9kernel-logit-selection | ||||||||||
10kernel-masked-attention | ||||||||||
11kernel-parameter-update | ||||||||||
12kernel-projected-state-attention | ||||||||||
13kernel-rectangular-matrix-workload | ||||||||||
14kernel-single-query-attention | ||||||||||
15language-model-optimizer | ||||||||||
16long-sequence-computation | ||||||||||
17model-shape-efficiency | ||||||||||
18pt-chained-rl-qwen3-4b | ||||||||||
19pt-interfaceport-easy-rl-qwen3-4b | ||||||||||
20pt-multifile-numpy-rl-qwen3-4b | ||||||||||
21pt-multifile-rl-qwen3-4b | ||||||||||
22pt-multifile-scipy-rl-qwen3-4b | ||||||||||
23separated-prefill-and-decode | ||||||||||
24speculative-draft-training | ||||||||||
25swe-bench-harness | ||||||||||
26training-objective-design | ||||||||||
27webarena-verified-harness |