Skip to content

Actions: zjunlp/JevLoop

Actions

All workflows

Actions

Loading...
Loading

Showing runs from all workflows
131 workflow runs
131 workflow runs

Filter by Workflow

⌥ + click/return to exclude

Filter by Event

⌥ + click/return to exclude

Filter by Status

⌥ + click/return to exclude

Filter by Branch

⌥ + click/return to exclude

Filter by Actor

⌥ + click/return to exclude
The retrieval corpus is the one external set where both numbers come out CI #131: Commit 94d3103 pushed by Xubqpanda
48s main
main
48s
Report the DRACO number as a limitation, not as a rate CI #130: Commit 5eb6687 pushed by Xubqpanda
44s main
main
44s
Say why terminal-bench cannot give the number yet, and name two gaps … CI #129: Commit ed8a081 pushed by Xubqpanda
49s main
main
49s
Read the field that has the prose in it, and give the ruler a second … CI #128: Commit dce698e pushed by Xubqpanda
50s main
main
50s
Let someone else's data audit our ruler, and it finds a blind spot CI #127: Commit 09eeb82 pushed by Xubqpanda
46s main
main
46s
Adapt someone else's corpus, since their gold is the one we do not ha… CI #126: Commit c8aade9 pushed by Xubqpanda
45s main
main
45s
Pick the datasets by whether they can produce the two numbers, not by… CI #125: Commit 06535d4 pushed by Xubqpanda
49s main
main
49s
Put the datasets on the table: ours in three layers, theirs in one table CI #124: Commit 93961a1 pushed by Xubqpanda
48s main
main
48s
Write down the claim we would actually defend, and the three we would… CI #123: Commit c492259 pushed by Xubqpanda
50s main
main
50s
Weave the workflow axis into the brief instead of giving it its own s… CI #122: Commit 98bd198 pushed by Xubqpanda
51s main
main
51s
Separate the three axes, since the placement claim was being read as … CI #121: Commit 94b3659 pushed by Xubqpanda
53s main
main
53s
Write the one-page version, and correct the page that claimed 3% CI #120: Commit 0273769 pushed by Xubqpanda
45s main
main
45s
Write down what we found, what we built, and what is still unproven CI #119: Commit ee6daa4 pushed by Xubqpanda
49s main
main
49s
Freeze the operating point, then go and test it on candidates it has … CI #118: Commit b1232b9 pushed by Xubqpanda
50s main
main
50s
Separate the judgement from the policy, and the false alarms turn out… CI #117: Commit 650541b pushed by Xubqpanda
44s main
main
44s
Grow the misleading class on purpose, and the gate's ceiling shows up CI #116: Commit 5f46331 pushed by Xubqpanda
46s main
main
46s
Report the swap control, and keep Jev in the experiment CI #115: Commit 87a236d pushed by Xubqpanda
49s main
main
49s
Add the negative control and the falsification arm, and fix the three… CI #114: Commit 2a362a9 pushed by Xubqpanda
18s main
main
18s
Measure what the gate decides, since the average cannot see it CI #113: Commit f76cb6f pushed by Xubqpanda
18s main
main
18s
Write down how the five closest papers measure this, and what we shou… CI #112: Commit bd6de02 pushed by Xubqpanda
19s main
main
19s
Make the fact deterministic before asking anyone about the wording CI #111: Commit 4b99b4a pushed by Xubqpanda
16s main
main
16s
Repeat the variant matrix, and the controls come out clean CI #110: Commit a32ac93 pushed by Xubqpanda
17s main
main
17s
Widen the variant matrix, and fix the second false accusation it exposed CI #109: Commit ca00f2e pushed by Xubqpanda
17s main
main
17s
Let a model answer the delivery gate, and watch it approve a lie CI #108: Commit 0c2d275 pushed by Xubqpanda
17s main
main
17s
Build the condition that tempts a false claim, and get the first real… CI #107: Commit b85a9d9 pushed by Xubqpanda
23s main
main
23s