Mangosteen

wannaphong 's Collections

updated Sep 2, 2025

Mangosteen, a 47 billion-token Thai corpus built with a Thai-adapted pipeline, improves language model performance on Thai benchmarks.

Upvote

Mangosteen: An Open Thai Corpus for Language Model Pretraining

Paper • 2507.14664 • Published Jul 19, 2025 • 7
wannaphong/dolma_web1_raw

Viewer • Updated Jul 28, 2025 • 325M • 4

Note Raw data for Thai Dolma - Commoncrawl - Fineweb2
wannaphong/fineweb2_base

Viewer • Updated Jul 28, 2025 • 37.5M • 2

Note Fineweb2 LD
wannaphong/fineweb2_quality

Viewer • Updated Jul 28, 2025 • 19.3M • 2

Note + Quality filtering
wannaphong/fineweb2_quality_dedup

Viewer • Updated Jul 28, 2025 • 17.7M • 2

Note +Deduplication
wannaphong/fineweb2_dolma

Viewer • Updated Jun 8, 2025 • 16.9M • 2

Note Fineweb2 by our pipeline (LD+ Quality filtering + Deduplication + Content filtering)
wannaphong/cc_base

Viewer • Updated Jul 28, 2025 • 165M • 2

Note Commoncrawl LD
wannaphong/cc_quality

Viewer • Updated Jul 28, 2025 • 34.5M • 2

Note + Quality filtering
wannaphong/cc_quality_dedup

Viewer • Updated Jul 28, 2025 • 25.4M • 2

Note +Deduplication
wannaphong/cc_dolma

Preview • Updated Oct 30, 2025 • 2

Note Commoncrawl by our pipeline (LD+ Quality filtering + Deduplication + Content filtering)
wannaphong/web-dataset

Viewer • Updated Apr 14, 2025 • 19.8M • 2
wannaphong/thai-dolma-fasttext-model

Updated Jul 2, 2025

Note FastText Model
aisingapore/WangchanLION-Curated

Viewer • Updated Sep 3, 2025 • 402k • 93 • 3

Note Non-common crawl subset
aisingapore/WangchanLION-Web

Viewer • Updated Sep 3, 2025 • 19.8M • 97 • 3

Note common crawl subset
aisingapore/WangchanLION-v3

Text Generation • 8B • Updated Sep 3, 2025 • 74 • 3

Note CPT base model
aisingapore/WangchanLION-v3-IT

Text Generation • 8B • Updated Sep 3, 2025 • 69 • 3

Note SFT model
wannaphong/mangosteen-gpt2-lab

Updated Jul 18, 2025
wannaphong/mangosteen-cpt-checkpoint

Updated Jul 14, 2025

Upvote