Mangosteen, a 47 billion-token Thai corpus built with a Thai-adapted pipeline, improves language model performance on Thai benchmarks.
Wannaphong Phatthiyaphaibun PRO
wannaphong
AI & ML interests
None yet
Recent Activity
updated
a dataset
about 3 hours ago
wannaphong/ArgInstruct
published
a dataset
about 3 hours ago
wannaphong/ArgInstruct
updated
a dataset
1 day ago
wannaphong/thaisum-thaipbs
Organizations
models
54
wannaphong/SmolLM3-3B-Base-Thai-v0.1
Text Generation
•
3B
•
Updated
•
6
wannaphong/cat-tokenizer
Updated
wannaphong/mangosteen-gpt2-lab
Updated
wannaphong/mangosteen-cpt-checkpoint
Updated
wannaphong/thai-dolma-fasttext-model
Updated
wannaphong/typhoon2-qwen2.5-7b-instruct-1M
Text Generation
•
8B
•
Updated
•
11
wannaphong/Roman2Thai-transliterator
Translation
•
0.1B
•
Updated
•
8
wannaphong/ThaiAILab
Updated
wannaphong/word2word-model
Updated
wannaphong/Sailor2-20B-Chat-Q8_0-GGUF
Text Generation
•
19B
•
Updated
•
2
datasets
50
wannaphong/ArgInstruct
Viewer
•
Updated
•
52.4k
wannaphong/thaisum-thaipbs
Viewer
•
Updated
•
87.4k
•
142
wannaphong/thwikisource-20250901-cleaned
Viewer
•
Updated
•
4.16k
•
17
wannaphong/thwikiquote-20250901-cleaned
Viewer
•
Updated
•
649
•
13
wannaphong/thwikibooks-20250901-cleaned
Viewer
•
Updated
•
1.08k
•
13
wannaphong/thwiki-20250901-cleaned
Viewer
•
Updated
•
153k
•
23
wannaphong/Thai-TinyTextbook2
Viewer
•
Updated
•
1.03k
•
18
wannaphong/fineweb2-thai-cleaned
Preview
•
Updated
•
7
wannaphong/backup-thai_dolma-gamble_web_full
Viewer
•
Updated
•
201M
•
3
wannaphong/dhamma-text
Viewer
•
Updated
•
190
•
113