| --- |
| license: mit |
| task_categories: |
| - text-generation |
| language: |
| - en |
| tags: |
| - code |
| - python |
| - ai-safety |
| - control |
| - apps |
| pretty_name: APPS-cleaned for AI Control replication |
| size_categories: |
| - 1K<n<10K |
| --- |
| |
| # control-apps-cleaned |
|
|
| A curated subset of the [APPS dataset](https://github.com/hendrycks/apps) (Hendrycks et al., NeurIPS 2021), pre-filtered for use in the ARENA *AI Control* chapter — a teaching replication of Greenblatt et al. 2023 ([arXiv:2312.06942](https://arxiv.org/abs/2312.06942)). |
|
|
| ## What's in here |
|
|
| - **`cleaned_apps.jsonl`** — 1,202 problems from the APPS "interview" split, filtered to a uniform I/O schema (inputs and outputs are each one of `list[str]`, `list[int]`, `list[list[str]]`, `list[list[int]]`). Each line is a JSON record with the original APPS fields plus two added fields: |
| - `input_type` — one of `"list of strings"`, `"list of integers"`, `"list of lists of strings"`, `"list of lists of integers"` |
| - `output_type` — same vocabulary |
| |
| The filtering removes problems whose original I/O was strings-with-embedded-newlines (which broke the chapter's `main(input)` function-call interface). |
| |
| ## Why this exists |
| |
| The original APPS dataset has inputs as raw stdin strings, which doesn't play nicely with the inspect_ai sandbox calling `main(input)` directly. This filtered subset lets every problem be solved by a function `def main(input: list[str]) -> list[str]: ...` without per-problem dispatch on the input shape. |
| |
| ## Usage |
| |
| ```python |
| from huggingface_hub import hf_hub_download |
| import json |
| |
| path = hf_hub_download( |
| repo_id="styme3279/control-apps-cleaned", |
| filename="cleaned_apps.jsonl", |
| repo_type="dataset", |
| ) |
| |
| with open(path) as f: |
| apps = [json.loads(line) for line in f] |
| |
| print(len(apps), "problems") |
| print(apps[0].keys()) |
| ``` |
| |
| ## Provenance |
| |
| - Upstream source: [hendrycks/apps](https://github.com/hendrycks/apps), interview-difficulty split |
| - Filter script: `utils.clean_dataset` in the ARENA chapter `chapter3_llm_evals/exercises/part5_ai_control/utils.py` |
| - Used by: ARENA chapter 3.5 (AI Control) |
| |
| ## License |
| |
| MIT, inherited from the upstream APPS dataset. |
| |