PyGeoX

Datasets for training and evaluating language models on geometric constraint solving — reading a natural-language description of a diagram and producing the point coordinates / circle radii that satisfy the stated constraints — plus a supervised set for generating PyGeoX scene code.

Subsets

Config	Rows	Task
gcs-sft	15,738	SFT — solve geometry (NL → coordinates), merged master with per-variant reward labels
codegen-sft	46,977	SFT — NL diagram description → PyGeoX code
gcs-rl	46,977	RL — prompts for geometry constraint solving
bench	300	Evaluation benchmark (self-contained)
wild	200	Evaluation — real-world school-geometry questions

from datasets import load_dataset
ds = load_dataset("rafaelcabral96/PyGeoX", "gcs-rl")

How this data is meant to be used. Several subsets contain free-form nested objects (e.g. Objs, possible_solution) whose keys vary per example. The reward / evaluation harness downloads the raw .jsonl files and parses them line-by-line with json — that always works. The Hugging Face viewer may not fully render the nested fields.

Reward ground truth (PyGeoX-GCS-RL-Code.zip)

The problem definitions used to score RL and benchmark outputs are shipped as data/PyGeoX-GCS-RL-Code.zip. The source_file field of each gcs-rl row points into this folder. After downloading the repo, extract it inside data/ so the relative paths resolve:

cd data && unzip PyGeoX-GCS-RL-Code.zip      # -> data/PyGeoX-GCS-RL-Code/<id>.json

All source_file paths are written relative to the repo root (data/PyGeoX-GCS-RL-Code/<name>.json), so they are invariant to where the repo lives.

Subset details

gcs-sft — geometry-solving SFT (merged master)

One deduplicated file keyed by (problem, completion). Each row's splits map records, per training variant (sar, sar_sd, mse, mse_sd, sparse, plus the full base), whether that variant includes it and with what reward/weight. Reconstruct any variant:

rows    = [r for r in ds if r["splits"]["mse_sd"]["in"]]
weights = [r["splits"]["mse_sd"].get("weight", 1.0) for r in rows]

codegen-sft — NL → PyGeoX code

messages = (user: diagram description + "Please generate PyGeoX code…", assistant: python … block), plus problem_id, source_file, difficulty.

gcs-rl — RL prompts

messages (system + user), source_file (→ reward ground truth), difficulty (easy/medium/hard for 1/2/3 primary objects).

bench — evaluation benchmark (self-contained)

Each record carries everything needed to score a model inline: unique_id, nl_description, pygeox_code, Objs, Rels, Points, extra_rel, possible_solution. Rebuild the scene directly from a record (no file needed) and score predicted coordinates:

from pygeox.synthetic.llm_client import create_scene_from_json
scene = create_scene_from_json(domain=10, json_data=record, generate_objective_function=True)
reward, details = scene.reward.reward_function(pred_points, pred_circles)

wild — real-world school geometry

200 questions from MathVerse, ZhongKaoGeo, and MathVista (id, question, source, source_id). Ground truth is executable PyGeoX full_code, shipped in data/PyGeoX-Wild-Code.zip (one problem_<id>.json per question, keyed 1:1 by id, each with full_code + possible_solution). Extract inside data/:

cd data && unzip PyGeoX-Wild-Code.zip    # -> data/PyGeoX-Wild-Code/problem_<id>.json

Citation

@software{pygeox_datasets,
  title  = {PyGeoX: Datasets for Geometric Constraint Solving},
  author = {Rafael Cabral},
  year   = {2026},
  url    = {https://huggingface.co/datasets/rafaelcabral96/PyGeoX}
}

License

MIT