task_id stringlengths 20 33 | component stringclasses 5
values | domain stringclasses 8
values | behavior stringlengths 48 100 | split stringclasses 1
value | source_dataset stringclasses 5
values | source_revision stringclasses 5
values | source_url stringclasses 5
values | skill_family stringclasses 8
values | reward stringclasses 2
values | instance_policy stringclasses 2
values | original_instruction stringlengths 56 73 | original_instruction_sha256 stringlengths 64 64 | source_record stringclasses 1
value | original_instruction_json_pointer stringclasses 1
value | computational_tools bool 1
class |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
candidate-1005-software-systems | mimo-code | software-engineering | Protocol library regression repair and functional checks | train | FineEnvs/MiMo-V2.6-RL-harbor-terminal | b74a53f26a1c7a4070a32a2c96b0dc301dce9472 | https://huggingface.co/datasets/FineEnvs/MiMo-V2.6-RL-harbor-terminal/tree/b74a53f26a1c7a4070a32a2c96b0dc301dce9472 | code_repair | Unchanged original binary success reward; 0 or 1. | One fixed original train instance per task_id; no random question substitution. | components/mimo-code/source_bundle/source/instruction.md | 44d9729328960cc04484e8ff23d492ac54edfc83f5910c24301742c895b96920 | null | null | null |
task_008667_2151f271 | tmax | finance-economics | Cross-account ownership mapping, transaction cycle detection, and deduplication | train | TMaxxx/TMax-15K-Harbor | 48a77eb0b017606c643ed905f96db41672914798 | https://huggingface.co/datasets/TMaxxx/TMax-15K-Harbor/tree/48a77eb0b017606c643ed905f96db41672914798 | database_workflows | Unchanged original binary success reward; 0 or 1. | One fixed original train instance per task_id; no random question substitution. | components/tmax/tasks/task_008667_2151f271/source/instruction.md | 50d1d10121dfa3b565842fd10f2589ceb91b4459126787af7793cec9cc0d06b1 | null | null | null |
construct-mc-ramsey-l4-s0 | math | mathematics-or-formal-reasoning | Combinatorial search and counterexample checking | train | amphora/MathConstructOptimize-Envs-harbor | 66cd5c87d51113c76faaef1b05397f1a6282a0f7 | https://huggingface.co/datasets/amphora/MathConstructOptimize-Envs-harbor/tree/66cd5c87d51113c76faaef1b05397f1a6282a0f7 | constraint_construction | Unchanged original binary success reward; 0 or 1. | One fixed original train instance per task_id; no random question substitution. | components/math/upstream/construct-mc-ramsey-l4-s0/instruction.md | dacf1d28faa912ac766cb9a8859b9a2d08940d2c52a9f8460c85ce0595a16c81 | null | null | null |
awm-accounting-1-invoice | awm | office-white-collar | Business entity lookup, invoice creation, discount and tax calculations, and parent-child line items | train | Snowflake/AgentWorldModel-1K | dde80a0283fe781bdc51656bce57063dc5650213 | https://huggingface.co/datasets/Snowflake/AgentWorldModel-1K/tree/dde80a0283fe781bdc51656bce57063dc5650213 | business_api | Unchanged original pure-code state verifier, binary 0/1. | Fixed original request and pristine database copied at reset. | components/awm/private/tasks/awm-accounting-1-invoice/SOURCE.json | a08b48d978d8a70cfea5c6e871d079622fc99ac8438fb38d6d6816bdfa399c99 | gen_tasks.jsonl#scenario=accounting_1;tasks[1] | /task | null |
task_000260_7b7cc286 | tmax | media-content-production | Frame-by-frame video analysis and structured output | train | TMaxxx/TMax-15K-Harbor | 48a77eb0b017606c643ed905f96db41672914798 | https://huggingface.co/datasets/TMaxxx/TMax-15K-Harbor/tree/48a77eb0b017606c643ed905f96db41672914798 | media_analysis | Unchanged original binary success reward; 0 or 1. | One fixed original train instance per task_id; no random question substitution. | components/tmax/tasks/task_000260_7b7cc286/source/instruction.md | 79c9fd4e4d41524db109d90b0415a4f0a2b9d24bf690fa8d7340ce3cb5ce0975 | null | null | null |
pdbthink-s09-70c0778d985e571dde37 | pdbthink | natural-science | Coordinate extraction and oriented geometry classification without tools | train | open-athena/pdbthink-coordinate-tasks | 3734406cb97b1702844319f9a5d860cbbf8fe660 | https://huggingface.co/datasets/open-athena/pdbthink-coordinate-tasks/tree/3734406cb97b1702844319f9a5d860cbbf8fe660 | tool_free_geometry | Unchanged original binary success reward; 0 or 1. | One fixed original train instance per task_id; no random question substitution. | upstream/pdbthink/pdbthink-s09-70c0778d985e571dde37/instruction.md | 33d86986d7b9d0a12678be7326d02048cf812cb536fc45fdb2837c9772c64ddf | null | null | false |
task_000663_dff2c340 | tmax | office-white-collar | Backup restoration dependencies, time boundaries, and resource aggregation | train | TMaxxx/TMax-15K-Harbor | 48a77eb0b017606c643ed905f96db41672914798 | https://huggingface.co/datasets/TMaxxx/TMax-15K-Harbor/tree/48a77eb0b017606c643ed905f96db41672914798 | database_workflows | Unchanged original binary success reward; 0 or 1. | One fixed original train instance per task_id; no random question substitution. | components/tmax/tasks/task_000663_dff2c340/source/instruction.md | feae25374139b4b1d8ec9759d553c591841d9330a071217f46a2e464bd32bbd7 | null | null | null |
candidate-1001-security-forensics | mimo-code | cybersecurity | Exception handling and event recovery for corrupted forensic data | train | FineEnvs/MiMo-V2.6-RL-harbor-terminal | b74a53f26a1c7a4070a32a2c96b0dc301dce9472 | https://huggingface.co/datasets/FineEnvs/MiMo-V2.6-RL-harbor-terminal/tree/b74a53f26a1c7a4070a32a2c96b0dc301dce9472 | code_repair | Unchanged original binary success reward; 0 or 1. | One fixed original train instance per task_id; no random question substitution. | components/mimo-code/source_bundle/source-other/instruction.md | 6d55a1b1cb882a951518a74f06dbd70bdb845a3669b4ad0210e4aff0c56d82a3 | null | null | null |
task_000056_f726766b | tmax | industrial-physical-systems | Waveform system identification, numerical fitting, and result verification | train | TMaxxx/TMax-15K-Harbor | 48a77eb0b017606c643ed905f96db41672914798 | https://huggingface.co/datasets/TMaxxx/TMax-15K-Harbor/tree/48a77eb0b017606c643ed905f96db41672914798 | system_identification | Unchanged original binary success reward; 0 or 1. | One fixed original train instance per task_id; no random question substitution. | components/tmax/tasks/task_000056_f726766b/source/instruction.md | 714f780883970365cfffe408e2c76e75f499005f87551fc84b55d79368e77197 | null | null | null |
construct-mc-social-golfers-l4-s0 | math | mathematics-or-formal-reasoning | Constraint scheduling and combinatorial coverage | train | amphora/MathConstructOptimize-Envs-harbor | 66cd5c87d51113c76faaef1b05397f1a6282a0f7 | https://huggingface.co/datasets/amphora/MathConstructOptimize-Envs-harbor/tree/66cd5c87d51113c76faaef1b05397f1a6282a0f7 | constraint_construction | Unchanged original binary success reward; 0 or 1. | One fixed original train instance per task_id; no random question substitution. | components/math/upstream/construct-mc-social-golfers-l4-s0/instruction.md | a638ef887b79167571086549b8d9438447d07b800772aac868b3e9a2e6899cd2 | null | null | null |
pdbthink-s09-7ba4aa076b6bb5633953 | pdbthink | natural-science | Coordinate extraction and oriented geometry classification without tools | train | open-athena/pdbthink-coordinate-tasks | 3734406cb97b1702844319f9a5d860cbbf8fe660 | https://huggingface.co/datasets/open-athena/pdbthink-coordinate-tasks/tree/3734406cb97b1702844319f9a5d860cbbf8fe660 | tool_free_geometry | Unchanged original binary success reward; 0 or 1. | One fixed original train instance per task_id; no random question substitution. | upstream/pdbthink/pdbthink-s09-7ba4aa076b6bb5633953/instruction.md | 21dc586748f23a8630118831399e1c40d62443444d454bf6f5e7f60ba334b06f | null | null | false |
task_005684_3c8594a3 | tmax | office-white-collar | Communication graph computation, database joins, and sensitive access checks | train | TMaxxx/TMax-15K-Harbor | 48a77eb0b017606c643ed905f96db41672914798 | https://huggingface.co/datasets/TMaxxx/TMax-15K-Harbor/tree/48a77eb0b017606c643ed905f96db41672914798 | database_workflows | Unchanged original binary success reward; 0 or 1. | One fixed original train instance per task_id; no random question substitution. | components/tmax/tasks/task_005684_3c8594a3/source/instruction.md | f4eee4e0643fcfb7399df10e1947ec9dbda28c74bacf307167f443506268f583 | null | null | null |
construct-mc-costas-array-l3-s0 | math | mathematics-or-formal-reasoning | Implementing constructive constraints and complete validation | train | amphora/MathConstructOptimize-Envs-harbor | 66cd5c87d51113c76faaef1b05397f1a6282a0f7 | https://huggingface.co/datasets/amphora/MathConstructOptimize-Envs-harbor/tree/66cd5c87d51113c76faaef1b05397f1a6282a0f7 | constraint_construction | Unchanged original binary success reward; 0 or 1. | One fixed original train instance per task_id; no random question substitution. | components/math/upstream/construct-mc-costas-array-l3-s0/instruction.md | 88e4a26e71d8483fbc89627210e4ed07f6dfef0fb16ce3e6d31e84a04bd050f9 | null | null | null |
task_008998_ae4cbe72 | tmax | industrial-physical-systems | State parsing from compressed manufacturing logs and consistent multi-file output | train | TMaxxx/TMax-15K-Harbor | 48a77eb0b017606c643ed905f96db41672914798 | https://huggingface.co/datasets/TMaxxx/TMax-15K-Harbor/tree/48a77eb0b017606c643ed905f96db41672914798 | stream_parsing | Unchanged original binary success reward; 0 or 1. | One fixed original train instance per task_id; no random question substitution. | components/tmax/tasks/task_008998_ae4cbe72/source/instruction.md | f81f86b07681931baabdba313cc4e988c01bff7a1f1d83349dfa939cdaa20669 | null | null | null |
pdbthink-s09-772d9c283d4255fc0a11 | pdbthink | natural-science | Coordinate extraction and oriented geometry classification without tools | train | open-athena/pdbthink-coordinate-tasks | 3734406cb97b1702844319f9a5d860cbbf8fe660 | https://huggingface.co/datasets/open-athena/pdbthink-coordinate-tasks/tree/3734406cb97b1702844319f9a5d860cbbf8fe660 | tool_free_geometry | Unchanged original binary success reward; 0 or 1. | One fixed original train instance per task_id; no random question substitution. | upstream/pdbthink/pdbthink-s09-772d9c283d4255fc0a11/instruction.md | 1eba6ac44775863993c8215f0d146cdccc7e8874a05b694e79db776e375dd7b1 | null | null | false |
This dataset contains 15 fixed training tasks served by five OpenEnv images.
| Task type | Count | Source |
|---|---|---|
| Software repair | 2 | MiMo |
| Database and financial data processing | 3 | TMax |
| Business API workflows | 1 | AgentWorldModel |
| Manufacturing log processing | 1 | TMax |
| Video analysis | 1 | TMax |
| Physical parameter identification | 1 | TMax |
| Mathematical constraint construction | 3 | MathConstruct |
| Protein geometry | 3 | PDBThink |
tasks.jsonl lists task IDs, original instruction paths, and pinned source revisions. Each ID represents one original task. Reset restores its original inputs or database. The preserved original checkers compute binary rewards of 0 or 1.
components/ contains runtime adapters, original tasks, inputs, checkers, and dependency records. upstream/ preserves original Harbor materials. All images serve schema.json. Episodes use one persistent WebSocket connection at /ws. The shell environments use the pinned SDK's stateless HTTP inspection routes. Software, data, and mathematical tasks use exec and finish; business tasks use call and finish; protein tasks accept one submit_answer call.
REGISTRATION.json lists the 15 registered training IDs and two unregistered alternatives available in the images. SOURCE_LOCK.json, MANIFEST.json, and SHA256SUMS record source provenance and file hashes. Licenses and attribution appear in LICENSES/README.txt and the component notices.
Validate the local dataset:
python3 -B validate_dataset.py
Restore required source directories after downloading from Hugging Face:
python3 -B prepare_source_tree.py
- Downloads last month
- 129