PotatoHD/github-pr-agent-trajectories

1---2dataset_info:3  features:4  - name: task_id5    dtype: string6  - name: repo7    dtype: string8  - name: pr_number9    dtype: int3210  - name: base_sha11    dtype: string12  - name: head_sha13    dtype: string14  - name: task15    dtype: string16  - name: plan17    list:18    - name: subtask_id19      dtype: string20    - name: goal21      dtype: string22    - name: scope_files23      list: string24  - name: subagents25    list:26    - name: subtask_id27      dtype: string28    - name: scope_files29      list: string30    - name: solution_diff31      dtype: string32    - name: summary33      dtype: string34  - name: revisions35    list:36    - name: subtask_id37      dtype: string38    - name: comment39      dtype: string40    - name: revised_diff41      dtype: string42  - name: final_diff43    dtype: string44  - name: num_subtasks45    dtype: int3246  - name: num_changed_files47    dtype: int3248  splits:49  - name: train50    num_bytes: 786762353251    num_examples: 3469652  download_size: 232602709353  dataset_size: 786762353254configs:55- config_name: default56  data_files:57  - split: train58    path: data/train-*59tags:60- code61- agents62- github63- swe64- trajectories65- planning66license: other67language:68- code69- en70size_categories:71- 10K<n<100K72task_categories:73- text-generation74pretty_name: GitHub PR Agent Trajectories75---76 77 78## Description79 80Reconstructed multi-step agent trajectories for resolving public GitHub PRs: a `plan` of subtasks and per-subtask `subagents` with scoped files and solution diffs.81 82> Derived dataset. Source material retains its original per-item licence (see `source`/`repo` columns); treat as **other / mixed**. Provided as-is.83 84## Usage85 86```python87from datasets import load_dataset88ds = load_dataset("PotatoHD/github-pr-agent-trajectories")89```90