examples/tahoe/source.json

{
  "dataset": "tahoebio/Tahoe-100M",
  "shards": [
    "data/train-00600-of-03388.parquet",
    "data/train-00601-of-03388.parquet",
    "data/train-02600-of-03388.parquet"
  ],
  "cells": 568,
  "conditions": 144,
  "reservoir_cap": 4,
  "seed": 0
}