AnonymousSubmissionUnderDouble-BlindRevi/Anonymous_Submission_Release
SETA-Env SETA-Env is an open-source verifiable RL terminal environment dataset for community training and evaluation. This release contains two top-level subsets: SETA_Synth: synthesized tasks SETA_Evolve: evolved variants of terminal-agent tasks The current release contains 4567 environments: SETA_Synth: 3255 SETA_Evolve: 1312 What Is Included Each task is packaged as a self-contained Harbor-style task directory with the files needed to run the task, build… See the full description on the dataset page: https://huggingface.co/datasets/AnonymousSubmissionUnderDouble-BlindRevi/Anonymous_Submission_Release.
SETA-Env
SETA-Env is an open-source verifiable RL terminal environment dataset for community training and evaluation.
This release contains two top-level subsets:
SETA_Synth: synthesized tasksSETA_Evolve: evolved variants of terminal-agent tasks
The current release contains 4567 environments:
SETA_Synth:3255SETA_Evolve:1312
What Is Included
Each task is packaged as a self-contained Harbor-style task directory with the files needed to run the task, build its environment, execute a reference solution, and run tests.
Dataset Structure
SETA_Synth/
├── ask_ubuntu__...
├── kaggle_notebook__...
├── nl2bash__...
├── stack_overflow__...
└── unix_linux_se__...
SETA_Evolve/
├── ask_ubuntu__...__b1
├── ask_ubuntu__...__d1
├── kaggle_notebook__...__b1
├── nl2bash__...__b1
├── stack_overflow__...__d1
└── unix_linux_se__...__b1A typical released task folder looks like:
{task_id}/
├── task.toml
├── instruction.md
├── environment/
│ ├── Dockerfile
│ └── ...
├── solution/
│ └── solve.sh
└── tests/
├── test.sh
└── test_outputs.pySource Breakdown
SETA_Synth
SETA_Evolve
Notes
- Task folder names are the primary identifiers in this release.
SETA_Evolvetask names include suffixes such as__b1and__d1to distinguish evolved variants.__b1means context change round 1.__d1means difficulty change round 1.
Suggested Usage
from huggingface_hub import snapshot_download
snapshot_download(
repo_id="AnonymousSubmissionUnderDouble-BlindRevi/Anonymous_Submission_Release",
repo_type="dataset",
local_dir="./seta_release",
)To download only one subset:
from huggingface_hub import snapshot_download
snapshot_download(
repo_id="AnonymousSubmissionUnderDouble-BlindRevi/Anonymous_Submission_Release",
repo_type="dataset",
allow_patterns="SETA_Synth/**",
local_dir="./seta_release",
)License
This release contains mixed-source data.
stack_overflow,unix_linux_se, andask_ubuntuare derived from Stack Exchange content licensed under CC BY-SA, with the exact version depending on post date. See also Ask Ubuntu licensing: https://askubuntu.com/help/licensingnl2bashis derived from the originalNL2Bashdataset, whose upstreamdata/bashsplit is released under MIT: https://github.com/TellinaTool/nl2bashkaggle_notebookis derived from public Kaggle notebooks and related public assets; licensing may vary by item. See: https://www.kaggle.com/terms
