TagaLLM
Collection
Tagalog and Taglish research models and dataset • 5 items • Updated
The dataset is currently empty. Upload or create new data files. Then, you will be able to explore them in the Dataset Viewer.
YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
TagaLLM dataset documentation
This repository documents a Tagalog and Taglish conversational dataset candidate. Dialogue records are withheld pending generation-provider clearance for redistribution and downstream model training. No dataset license has been applied.
A native Tagalog speaker reviewed selected dataset examples for grammatical accuracy, natural phrasing, and fit with everyday Metro Manila Tagalog.
| Split | Candidate records |
|---|---|
| Train | 35,033 |
| Validation | 1,230 |
| Test | 1,320 |
| Total | 37,583 |
The candidate excludes the four VOA Learning English source families and restores 4,265 missing skill labels from source metadata. Historical research used separately pinned snapshots.
Each record contains id, family, topic, skill, messages (an ordered array of role and content objects), source_pool, editorial_reviewed, and payload_tokens. This card contains aggregate statistics and schema information only.
Project links
- Downloads last month
- 19