AbstractPhil/cc-prompts-sharded

Conceptual Captions — sharded for the Qwen task-conversion pipeline

Source: Conceptual Captions (conceptual_captions), deduplicated, minimum 4 words, split into three roughly-equal shards plus a "long" shard for captions over 50 words (reserved for later high-capacity model processing).

Each row:

{"id": "cc_00000123", "caption": "<text>", "n_words": <int>}

id is the position of the row in the CC stream, zero-padded to 8 digits. Stable across re-runs. Used by the downstream inference cell for resume tracking.

Usage

from datasets import load_dataset
shard_0 = load_dataset("AbstractPhil/cc-prompts-sharded", "shard_0", split="train")
# Configs: shard_0, shard_1, shard_2, long

Built by

cc_dataset_prep.py — single Colab cell, streams CC, dedupes, shards round-robin, uploads chunked part files (5000 rows each) for safe resumability. Re-running the cell picks up where the last upload left off without re-processing rows.