Spaces:
Sleeping
Sleeping
Commit History
-
SpreadsheetBench: SkillOpt mode=multi — 30-turn code repair with execution feedback, 600 s task budget, training-only gold cell feedback
Verified signaturef24ba1b
-
SpreadsheetBench: actor prompts aligned verbatim with the SkillOpt codegen harness
Verified signaturea15f83f
-
Argument descriptions carry format only; environment description is the single home of world rules
Verified signatureadaffb2
-
fix: prevent stalled SpreadsheetBench sessions
Verified signature81a00cc -
feat: expose native typed action tools
Verified signature2869fbc
-
Preserve official SpreadsheetBench grading positions
Verified signature9426326 -
Stage isolated SpreadsheetBench concurrency candidate
Verified signature3f44ff2 -
Isolate runner identities and cap BLAS threads
Verified signature1fa02d1 -
Freeze public SpreadsheetBench task identities
Verified signatureec813f6 -
Remove generated bytecode from Space
Verified signaturee250eb8 -
Deploy standalone SpreadsheetBench OpenEnv
Verified signatureb4a5305