metadata
configs:
- config_name: pt
data_files:
- split: train
path: pt/train-*.parquet
- config_name: es
data_files:
- split: train
path: es/train-*.parquet
- config_name: hi
data_files:
- split: train
path: hi/train-*.parquet
- config_name: all
data_files:
- split: train
path: all/train-*.parquet
llm-und/pretrain-pt-es-hi (PILOT-SCALE TEST BUILD -- not the final corpus)
Built from data/processed/{pt,es,hi}.parquet, themselves produced by a
300-docs-per-source pilot run across the 10 currently ingestable sources
(see TASK1-PLAN.md sec 2.1 and README.md for the 2 blocked ones). This is
here to validate the Etapa 6 aggregation shape end-to-end, not to publish.
| config | docs |
|---|---|
| pt | 882 |
| es | 821 |
| hi | 791 |
| all | 2,494 |