Datasets:

gowitheflowlab
/

physics-bench-solid-w-hardnegs

dataset_info:
  features:
    - name: text
      dtype: string
    - name: video
      dtype: string
    - name: hard_negative_texts
      list: string
    - name: hard_negative_videos
      list: string
  splits:
    - name: train
      num_bytes: 956287081
      num_examples: 2700
  download_size: 956287081
  dataset_size: 956287081
configs:
  - config_name: default
    data_files:
      - split: train
        path: metadata.parquet

Physics Bench Solid With Hard Negatives

Repository: gowitheflowlab/physics-bench-solid-w-hardnegs

Source dataset: gowitheflowlab/physics-bench-solid-train-2700

This repository follows the path-based layout of gowitheflowlab/physics-bench-optics-w-hardnegs.

rows: 2700
metadata columns: text, video, hard_negative_texts, hard_negative_videos
positive text: query 1 (__query_1) from the source case
hard negatives per row: 5
video paths: repository-relative videos/<family>/<case_id>.mp4
list alignment: hard_negative_texts[i] and hard_negative_videos[i] come from the same case
candidate pool: only the other 99 cases in the positive case's family

Reproducibility

global seed: 42
traversal: family alphabetical order, then case_id order
sampling: one random.Random(42) stream for all 2700 rows
selection: random.sample(sorted(other_99_case_ids), 5)

Loading

from pathlib import Path
import pyarrow.parquet as pq
from huggingface_hub import snapshot_download

root = Path(snapshot_download("gowitheflowlab/physics-bench-solid-w-hardnegs", repo_type="dataset"))
rows = pq.read_table(root / "metadata.parquet").to_pylist()
row = rows[0]
positive_video = root / row["video"]
negative_videos = [root / path for path in row["hard_negative_videos"]]

The training metadata intentionally contains only the four retrieval columns. Case IDs and query IDs are retained in source_metadata/sampling_manifest.json for auditing.