Buckets:

alphasyncllc/EdgeBench-bucket / lean_analysis_proofs.json
toddalanrose's picture
download
raw
1.59 kB
{
"task_id": "lean_analysis_proofs",
"name": "Lean Analysis Proofs",
"category": "Formal Math & Theorem Proving",
"base_image": "lean_4_28_0_main",
"platform": "linux/amd64",
"internet": false,
"cwd": "/home/workspace/",
"submit_paths": [
"se-bmk-intern/analysis"
],
"submit_exclude": [
"se-bmk-intern/analysis/.lake"
],
"work": {
"image_tag": "7a7b0144d6c2",
"specs_dir": null,
"agent_query": "You are a lean4 expert, now your task is to finish **all** sorries in this folder. Pay no attention to anything outside this folder. There are many sorries but do not be afraid as we value minor progress and the time is abundant. Try to do some easy problems and accumulate scores each round. Before you submit, run a `lake build` to verify the changes locally - if the build failed, no score will be given. Also, do not modify the signature of existing lemma/theorem/def as these changes will not pass the judge evaluation."
},
"judge": {
"image_tag": "5d6061092ab6",
"eval_cmd": "cd /home/workspace/se-bmk-intern/analysis && cp -r /home/workspace/baseline/.lake . && cp /home/workspace/baseline/lake-manifest.json . && lake build && cp /home/workspace/judge/se-bmk-intern/analysis/ListDeclAxiom.lean . && cd /home/workspace/judge/se-bmk-intern && python3 eval_lean.py --baseline-repo /home/workspace/judge/se-bmk-intern/analysis --current-repo /home/workspace/se-bmk-intern/analysis",
"eval_timeout": 2400,
"parser": "structured_json",
"rescale": {
"kind": "linear",
"lower": 0.0,
"upper": 1.0
}
}
}

Xet Storage Details

Size:
1.59 kB
·
Xet hash:
d4326ae17b7cc40cc81239d66c85c9331cf9dbea0c02dd116fdaf978b50447f5

Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.