WebInstructSub

← All datasets

Author: hf-mirror · License: apache-2.0 · Format: parquet

0 downloads · 0 stars · 0 mirrors · 3,509,803,840 · 13 files

2.3M mined STEM QA pairs (forums + StackExchange) from the MAmmoTH2 pipeline (Apache-2.0).

Tags: stem · sft · web-mined · tiger-lab · mammoth2

Description

TIGER-Lab's WebInstructSub: 2.3M question-answer pairs mined from mathstackexchange/socratic/StackOverflow-style QA via the MAmmoTH2 pipeline — STEM reasoning at web scale, the base for Mantis/MAmmoTH2. Apache-2.0. (13 shards, ~3.5 GB.)

Download

Direct download: https://huggingface.co/datasets/TIGER-Lab/WebInstructSub/resolve/main/data/train-00000-of-00013.parquet

Or via the API: curl -L -o file "https://huggingface.co/datasets/TIGER-Lab/WebInstructSub/resolve/main/data/train-00000-of-00013.parquet"

Files

• data/train-00000-of-00013.parquet (138,656,648)
• data/train-00001-of-00013.parquet (138,679,495)
• data/train-00002-of-00013.parquet (143,536,404)
• data/train-00003-of-00013.parquet (293,121,757)
• data/train-00004-of-00013.parquet (293,056,798)
• data/train-00005-of-00013.parquet (294,010,845)
• data/train-00006-of-00013.parquet (293,463,004)
• data/train-00007-of-00013.parquet (293,365,089)
• data/train-00008-of-00013.parquet (293,901,300)
• data/train-00009-of-00013.parquet (293,238,370)
• data/train-00010-of-00013.parquet (293,405,417)
• data/train-00011-of-00013.parquet (360,077,470)
• data/train-00012-of-00013.parquet (381,291,243)

Added 2026-09-27. Platform: Distillio — the open source training data collective.