seta-env

# SETA-Env  [](https://github.com/camel-ai/seta) [](https://huggingface.co/camel-ai/Qwen3-8B-SETA-Env-RL) 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 its environment, execute a reference solution, and run tests. ## Dataset Structure ```text 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__...__b1 ``` A typical released task folder looks like: ```text {task_id}/ ├── task.toml ├── instruction.md ├── environment/ │ ├── Dockerfile │ └── ... ├── solution/ │ └── solve.sh └── tests/ ├── test.sh └── test_outputs.py ``` ## Source Breakdown ### SETA_Synth | Source | Count | |---|---:| | `ask_ubuntu` | 1013 | | `kaggle_notebook` | 637 | | `nl2bash` | 276 | | `stack_overflow` | 884 | | `unix_linux_se` | 445 | | **Total** | **3255** | ### SETA_Evolve | Source | Count | |---|---:| | `ask_ubuntu` | 783 | | `kaggle_notebook` | 147 | | `nl2bash` | 9 | | `stack_overflow` | 256 | | `unix_linux_se` | 117 | | **Total** | **1312** | ## Notes - Task folder names are the primary identifiers in this release. - `SETA_Evolve` task names include suffixes such as `__b1` and `__d1` to distinguish evolved variants. - `__b1` means context change round 1. - `__d1` means difficulty change round 1. ## Suggested Usage ```python from huggingface_hub import snapshot_download snapshot_download( repo_id="camel-ai/seta-env-release", repo_type="dataset", local_dir="./seta_release", ) ``` To download only one subset: ```python from huggingface_hub import snapshot_download snapshot_download( repo_id="camel-ai/seta-env-release", repo_type="dataset", allow_patterns="SETA_Synth/**", local_dir="./seta_release", ) ```