fineweb-edu-ibex-merged-260315

Arabic translations from the IBEX parallel translation pipeline. Documents have been reconstructed from translation chunks — each row is a complete document.

Dataset Details

• Total documents: 82,840,410
• Shards: 867 (zstd compressed)
• Merged from: 650 chunk-level shards
• Created: 2026-03-17 18:45

Schema

Column	Type	Description
doc_id	int64	Document identifier
text	string	Full Arabic translation (chunks merged, ordered by position)

Usage

from datasets import load_dataset

# Streaming
ds = load_dataset("SultanR/fineweb-edu-ibex-merged-260315", split="train", streaming=True)
for doc in ds:
    print(doc["text"][:200])
    break

# Full load
ds = load_dataset("SultanR/fineweb-edu-ibex-merged-260315", split="train")
print(f"{len(ds):,} documents")

License

Please refer to the original source dataset for license information.