Datasets:
task_id stringlengths 7 13 | domain stringclasses 8
values | prompt stringlengths 365 3.39k |
|---|---|---|
finance-13 | finance | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
Value American options using a Cox-Ross-Rubinstein binomial tree. market.json contains spot, rate (... |
industrial-27 | industrial | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
The input file log.csv contains columns for time, value, and unit, where unit is either C or F, and... |
math-27 | math | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
A ticket from the ops desk is stored in ticket.txt, located alongside the referenced exports. Follo... |
media-12 | media | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
stream.mcf is a corrupted media container. Its structure consists of the magic bytes 'MCF1' followe... |
office-15 | office | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
Perform reconciliation of the intercompany ledgers for a corporate group with a cutoff date of 2026... |
science-24 | science | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
ticket.txt contains a request from a colleague, and the limits table, the flow log, and the lab exp... |
security-29 | security | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
Read the verification spec for a webhook receiver located at SPEC.md in this directory. Then classi... |
software-5 | software | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
Discard the temporary `old_report.py` script. Create a new command-line utility named `logq.py`, in... |
finance-21 | finance | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
Determine the contents of credit-card statements. The file transactions.csv contains the columns da... |
industrial-32 | industrial | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
File states.csv contains columns time (HH:MM) and state (RUN, STOP or FAULT). A state continues unt... |
math-31 | math | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
ticket.txt outlines a check-digit audit task. Create scan_report.csv containing the header scan_id,... |
media-13 | media | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
The file stream.mcf is a corrupted media container. Its structure begins with the magic bytes 'MCF1... |
office-33 | office | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
The file timesheet.csv contains columns employee,rate,date,start,end,break_min. Rate is hourly, and... |
science-37 | science | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
process prep.csv, which has columns id,stock (number and unit),take_uL,total_uL. The take_uL amount... |
security-32 | security | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
The request is contained in ticket.txt, while the inputs are policy.pdf and review_export.xlsx (usi... |
software-30 | software | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
README.md contains a ticket describing bugs in `migrate.py`, which loads a legacy CRM export into a... |
finance-24 | finance | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
The file TICKET.md located in the current directory contains the month-end close request from the c... |
industrial-33 | industrial | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
The file states.csv contains a time column (format HH:MM, rows NOT in time order) and a state colum... |
math-43 | math | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
The file graph.txt contains one edge per line: a line in the format 'U-V w' denotes a bidirectional... |
media-28 | media | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
The file request.txt contains a note from the podcast producer. The CMS export, site data, and uplo... |
office-34 | office | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
The file claim.csv contains columns for date, category, and amount, with all values in EUR and cate... |
science-43 | science | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
temps.csv contains one line per day with fields date,tmax_f (daily maximum temperature in Fahrenhei... |
security-33 | security | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
ticket.txt contains the request; the inputs are policy.pdf and review_export.xlsx (use openpyxl: in... |
software-32 | software | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
Read the README.md file, which describes a flaky scheduled job and directs you to its logs, then ge... |
finance-25 | finance | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
In the current directory, TICKET.md contains the month-end close request from the controller. Read ... |
industrial-38 | industrial | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
Each line in meter.log contains a single counter reading formatted as `YYYY-MM-DD HH:MM ZONE,counte... |
math-45 | math | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
conv.txt contains lines in the format 'id TYPE value'. N converts an integer to a spreadsheet colum... |
media-31 | media | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
The request is found in ticket.txt. Use run_of_show.pdf (parsed via pypdf) and clips.csv as inputs.... |
office-42 | office | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
meets.csv contains columns for id,local_start,minutes, representing local time in the PA zone forma... |
science-46 | science | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
The file meas.csv contains columns id,value,unc,unit, where unit is either g or mg. Precisely conve... |
security-35 | security | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
ticket.txt contains the request; scan_report.pdf and assets.xlsx (using openpyxl: inspect every she... |
software-33 | software | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
The file README.md contains a ticket describing a flaky scheduled job and references its logs; afte... |
finance-32 | finance | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
The file README.txt located in the current directory contains a request from treasury. Read this fi... |
math-46 | math | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
The file ranges.txt contains a single interval on each line, for example `[1.25, 4.5)`. A square br... |
media-39 | media | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
The file tc.txt contains 10 lines in the format `LABEL HH:MM:SS:FF`, representing timecodes at 24 f... |
office-43 | office | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
The file meets.csv contains columns id,local_start,minutes, where local time is specified in zone P... |
science-47 | science | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
The file meas.csv contains columns id,value,unc,unit, where unit is g, mg or kg. Precisely convert ... |
software-34 | software | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
Read the README.md file located in this directory, which contains a feature request for the `spend.... |
finance-37 | finance | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
Check the README.txt file located in the current directory, which contains a request from group acc... |
math-47 | math | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
The file ranges.txt contains one interval per line, formatted like `[1.25, 4.5)`. Square brackets d... |
media-44 | media | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
The file clips.csv contains the columns clip, seconds, video_kbps, and audio_kbps, where 1 kbps equ... |
office-46 | office | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
The file teams.csv contains the columns dept and staff. Distribute 80 new laptops. Calculate each q... |
software-49 | software | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
The file versions.txt contains lines formatted as `range <comparators>`, followed by one version pe... |
finance-40 | finance | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
The file loans.csv contains the fields id,principal,rate,start,end,basis, where principal is in dol... |
math-50 | math | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
The file prog.txt begins with the line `start N`, followed by one operation per line, processed seq... |
media-45 | media | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
clips.csv contains the columns clip,seconds,video_kbps,audio_kbps (where 1 kbps equals 1000 bits pe... |
software-54 | software | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
The ignore.txt file contains gitignore-style patterns, followed by a line consisting of ---, and th... |
finance-41 | finance | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
loans.csv contains the columns id, principal, rate, start, end, and basis, where principal is expre... |
software-55 | software | Operate from the current directory by executing shell commands; the environment provides python3 with access to numpy, pandas and Pillow. Once completed, respond with {"answer": "done"} to trigger scoring of your files.
ignore.txt contains gitignore-style patterns, followed by a --- line, then one path per line (a tra... |
Mixed Skills v9: eight-domain verified workspace tasks
One OpenEnv server, 406 deterministic procedurally generated tasks across the eight evaluation domains: software engineering, industrial and physical systems, natural science, office work, finance and economics, math and formal reasoning, cybersecurity, and media production. Each task gives the agent a scratch workspace with input files and a written specification. The agent works with shell commands and is scored by a hidden verifier on the files it leaves behind.
tasks.jsonl lists every task id with its domain and the exact first-turn prompt. The image is
docker.io/chrismun/mixed-env:v9 (linux/amd64); it contains 406 task ids and the submission names the 49 used for training.
Interaction
Each turn the model replies with exactly one JSON object:
{"command": "<shell command>"}runs in the workspace (python3 with numpy, pandas, Pillow and, for file-format tasks, openpyxl, XlsxWriter and pypdf). At most 12 commands per episode; each command's output is truncated to its last 1500 characters, so strong solutions inspect withheadand do the work in a script.{"answer": "done"}ends the episode and scores the workspace.
Observations carry the command output, the commands left and. There is no reminder text. Replies that are not a single JSON object end the episode, which is then scored on the workspace as it stands.
Reward
Terminal and in [0, 1]. Every verifier is a weighted set of independent sub-checks recomputed from the task inputs, so partial progress earns partial reward and only a fully correct deliverable scores 1.0. References live only inside the image, agent commands run as an unprivileged user with memory and file-size limits, and verifiers that execute agent code do so in a subprocess with a timeout. An untouched workspace scores at most 0.25 on every task.
Calibration
Tasks were measured with Qwen3.8-27B (thinking off) inside a tool-calling GRPO training harness, because pass rates there differ a lot from a strict JSON-reply driver (the two correlate only 0.25 per task). Each task was sampled 8 to 56 times across profiling and training passes.
This version is built for the guide's default budgets (completion 4096, context 8192 tokens), not the maximum. In the same harness the model's median episode on our earlier, longer tasks was about 5,100 tokens, and only about a third fit under 4096. The 49 training tasks here are short: in profiling runs about 92% of episodes on this style of task fit under 4096 completion tokens (median about 650 to 900 tokens), and each states its full requirement in a prompt of roughly 350 to 700 characters with one small input file.
- Pooled mean pass@1 of the 49 tasks: about 0.47. Short tasks proved easy for this model even with five to eight stacked exact rules, so this blend starts higher than earlier versions (0.3 to 0.4). Selecting on low observed rates regresses upward, so the true start is probably a little higher.
- Every task runs bare: no reply-format reminder and no per-step hint.
- Prompt text: the task statements and the three global strings were rewritten by Qwen3.8-27B under strict gates (every number, file name, identifier, quoted string and JSON example must survive verbatim). Input files written by the task generators were not rewritten.
- No transcript in the profiling runs touched verifier or task source.
Domains
| Domain | Skills exercised |
|---|---|
| software | multi-file bug fixing against hidden tests, implementing modules from a spec, CLI conversion, migration scripts, simulator debugging, performance and dependency problems |
| industrial | sensor log analysis and fault reports, control tuning by simulation, pipe and pump sizing, tariff billing, inventory policy, flow-line simulation |
| science | curve and rate fitting, sequence analysis and protein mass, ODE integration, hypothesis tests, titration equilibria, spectral analysis |
| office | payables reconciliation, meeting-minutes extraction, room scheduling, expense audits, payroll, contact deduplication |
| finance | amortization, market-data analytics, ledger audit, bond pricing, tax lots, FX arbitrage |
| math | exact number theory, proofs by computation, assignment optimization, shortest routes, constructive codes, absorbing Markov chains, impartial games |
| security | log incident analysis, hardening vulnerable code, layered-encoding and crypto forensics, detection rules, advisory matching (all synthetic, defensive) |
| media | subtitle retiming, image pipelines, audio synthesis, SVG charts, timecode conformance, markdown to RSS |
Files
tasks.jsonl:task_id,domain,promptfor each of the 49 training tasks, in training order (domains interleaved).
- Downloads last month
- 53