task_id string | split string | family string | level string | domain string | converted_to_xlsx list | converted_to_pdf list | example_instruction string | example_files list | n_files int64 |
|---|---|---|---|---|---|---|---|---|---|
loans-medium | train | loans | medium | finance-economics | [
"data/loans.csv",
"data/payments.csv"
] | [
"policy/credit_policy.md"
] | # Month-end loan provisions
The CFO writes: "Please prepare the August 2026 month-end loan-loss provisions as at 2026-08-31. Note that loans L-1128, L-1196 and L-1803 were written off last month, so leave them out."
- The loan book is in /root/data/loans.xlsx, installments and payments in /root/data/payments.xlsx, an... | [
"TASK.md",
"data/loans.xlsx",
"data/payments.xlsx",
"data/payments_returned.csv",
"data/writeoffs.csv",
"finance/fx_rates.csv",
"memos/memo-41_2026-06-06.md",
"memos/memo-42_2026-06-09.md",
"memos/memo-43_2026-08-09.md",
"memos/memo-44_2026-08-12.md",
"policy/credit_policy.pdf"
] | 11 |
loans-hard | train | loans | hard | finance-economics | [
"data/loans.csv",
"data/payments.csv",
"finance/fx_rates.csv"
] | [
"policy/credit_policy.md"
] | # Month-end loan provisions
The CFO writes: "Please prepare the August 2026 month-end loan-loss provisions as at 2026-08-31. Note that loans L-1006, L-2498 and L-3170 were written off last month, so leave them out."
- The loan book is in /root/data/loans.xlsx, installments and payments in /root/data/payments.xlsx, an... | [
"TASK.md",
"data/loans.xlsx",
"data/payments.xlsx",
"data/payments_returned.csv",
"data/writeoffs.csv",
"finance/fx_rates.xlsx",
"memos/memo-41_2026-06-15.md",
"memos/memo-42_2026-07-27.md",
"memos/memo-43_2026-07-28.md",
"memos/memo-44_2026-08-06.md",
"memos/memo-45_2026-08-09.md",
"memos/mem... | 16 |
expenses-easy | train | expenses | easy | office-white-collar | [
"data/claims.csv"
] | [] | # Expense audit
Finance needs an audit of the Research department's expense claims dated in September 2026 (2026-09-01 to 2026-09-30).
- Claims are in /root/data/claims.xlsx; employees and departments in /root/data/employees.csv.
- The travel and expense policy is in /root/policy, including amendments. The finance te... | [
"TASK.md",
"data/claims.xlsx",
"data/employees.csv",
"finance/README.md",
"finance/fx_corrections.md",
"finance/fx_rates.csv",
"policy/amendments/amendment-21_2026-09-02.md",
"policy/travel_policy.md"
] | 8 |
expenses-medium | train | expenses | medium | office-white-collar | [
"data/claims.csv",
"data/employees.csv"
] | [
"policy/travel_policy.md"
] | # Expense audit
Finance needs an audit of the Finance department's expense claims dated in September 2026 (2026-09-01 to 2026-09-30).
- Claims are in /root/data/claims.xlsx; employees and departments in /root/data/employees.xlsx.
- The travel and expense policy is in /root/policy, including amendments. The finance te... | [
"TASK.md",
"data/claims.xlsx",
"data/employees.xlsx",
"finance/README.md",
"finance/fx_corrections.md",
"finance/fx_rates.csv",
"policy/amendments/amendment-21_2026-08-15.md",
"policy/amendments/amendment-22_2026-09-13.md",
"policy/amendments/amendment-23_2026-09-21.md",
"policy/travel_policy.pdf"... | 10 |
downtime-medium | train | downtime | medium | industrial-physical-systems | [] | [
"handbook/maintenance_manual.md"
] | # Line downtime report
Plant engineering needs the unplanned downtime of production line A for the evening shift of 2026-09-21.
- The plant documentation is in /root/handbook, including amendments. Use it for the definition of unplanned downtime, the per-shift downtime limit that applies on that date, and any rules a... | [
"TASK.md",
"data/machines.json",
"data/sensor_log.csv",
"data/shifts.csv",
"handbook/amendments/amendment-101_2026-03-10.md",
"handbook/amendments/amendment-102_2026-03-25.md",
"handbook/amendments/amendment-103_2026-07-16.md",
"handbook/amendments/amendment-104_2026-08-16.md",
"handbook/maintenance... | 9 |
reflow-hard | train | reflow | hard | industrial-physical-systems | [
"data/runs.csv"
] | [
"docs/handbook.md"
] | # Reflow profile audit
Process engineering needs an audit of today's reflow oven runs on line 3.
- The process handbook is in /root/docs (read every document there; definitions and limits come only from the handbook).
- Thermocouple readings are in /root/data/thermocouples.csv; run details are in /root/data/runs.xlsx... | [
"TASK.md",
"data/boards.csv",
"data/runs.xlsx",
"data/thermocouples.csv",
"docs/handbook.pdf",
"docs/handbook_revision_B.md"
] | 6 |
pyrepo-bug1 | train | pyrepo | bug1 | software-engineering | [] | [] | # Fix the tabulate library
A recent change introduced a bug in the `tabulate` library at /root/repo/tabulate. For example, the test `tests.test_output::test_break_long_words` now fails.
Find the cause and fix it in the library code.
- Run the visible tests with `cd /root/repo/tabulate && python3 -m pytest -q tests`.... | [
"TASK.md",
"repo/tabulate/LICENSE",
"repo/tabulate/pytest.ini",
"repo/tabulate/tabulate/__init__.py",
"repo/tabulate/tests/common.py",
"repo/tabulate/tests/test_api.py",
"repo/tabulate/tests/test_internal.py",
"repo/tabulate/tests/test_output.py",
"repo/tabulate/tests/test_regression.py"
] | 9 |
pyrepo-implement | train | pyrepo | implement | software-engineering | [] | [] | # Fix the tabulate library
The function `simple_separated_format` in `/root/repo/tabulate/tabulate/__init__.py` is not implemented yet: it raises NotImplementedError. Implement it so it behaves as its docstring describes, consistent with the rest of the `tabulate` library.
- Run the visible tests with `cd /root/repo/... | [
"TASK.md",
"repo/tabulate/LICENSE",
"repo/tabulate/pytest.ini",
"repo/tabulate/tabulate/__init__.py",
"repo/tabulate/tests/common.py",
"repo/tabulate/tests/test_api.py",
"repo/tabulate/tests/test_internal.py",
"repo/tabulate/tests/test_output.py",
"repo/tabulate/tests/test_regression.py"
] | 9 |
repair-medium | train | repair | medium | software-engineering | [] | [] | # Fix the accrual package
Customers report wrong amounts from the invoicing service. The cause is in the Python package at /root/project/src/accrual.
- The required behaviour is in /root/project/docs/SPEC.md and /root/project/docs/CHANGELOG.md (the latest changelog entry wins where they differ).
- The visible tests i... | [
"TASK.md",
"project/README.md",
"project/docs/CHANGELOG.md",
"project/docs/SPEC.md",
"project/pytest.ini",
"project/scripts/export_invoices.py",
"project/src/accrual/__init__.py",
"project/src/accrual/dates.py",
"project/src/accrual/fees.py",
"project/src/accrual/invoices.py",
"project/src/accru... | 15 |
Eval-rules desk (Workbench v9)
An OpenEnv environment for OpenEnv Arena. It trains the survival habits that, in our local proxy of the hidden test, decided whether an episode ever reached an answer:
- one clean JSON object per turn
- small observations
- short scripts
- write the output early, then submit
The tasks are served under the hidden test's published rules.
- Source: https://github.com/moarshy/workbench (
server/evalrules_environment.py) - Image:
ghcr.io/moarshy/workbench@sha256:abc5496b962225f500fa40c699f4962cab6cea12b22d732667f891dc583700a5 - Design, calibration and pre-registration: https://github.com/moarshy/openenv-arena-notes/blob/main/versions/v9-eval-rules-desk.md
- Day-0 proxy: 23 public SkillsBench tasks, untrained Qwen3.8-27B. 0/91 episodes reached submit; 48 died on invalid replies and 35 on the budget. https://github.com/moarshy/openenv-arena-notes/blob/main/research/2026-10-18-day0-proxy-eval.md
Interface
Field for field the organizers' PostTrain adapter: {type: run_bash | submit | finalize, command, answer}.
- Output is cut to the first 8,192 characters.
- The workspace is
/root. - The first observation is the instruction only.
- Budgets: 8,192 completion tokens (counting observations), 16,384 context, 128 tool calls.
Tasks (9)
loans-medium/hard (finance), expenses-easy/medium (office), downtime-medium and reflow-hard (industrial), pyrepo-bug1/implement and repair-medium (software).
- Data arrives as real .xlsx workbooks and .pdf documents where a SkillsBench task would use them.
- Each task keeps the late caveats and traps of its Workbench family.
Reward
task_score = share of checks passed, 0 if a protected check fails (schema, inputs unchanged, regressions)
reward = 0.85*task_score + 0.15*(submitted, if task_score >= 0.5) + 0.02*valid actions (max 5) - 0.05*repeated turns (max 0.2)
The reward is clipped to [0, 1].
Licences
Vendored libraries keep their own licences: toolz and boltons are BSD-3-Clause, tabulate and inflection are MIT.
- Downloads last month
- -