datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
c4-en-html-with-metadatac4-en-html-with-metadata-ppl-cleanFile list:
"c4-en-html_cc-main-2019-18_pq00-000.jsonl.gz",
"c4-en-html_cc-main-2019-18_pq00-001.jsonl.gz",
"c4-en-html_cc-main-2019-18_pq00-002.jsonl.gz",
"c4-en-html_cc-main-2019-18_pq00-003.jsonl.gz",
"c4-en-html_cc-main-2019-18_pq00-004.jsonl.gz",
"c4-en-html_cc-main-2019-18_pq00-005.jsonl.gz",
"c4-en-html_cc-main-2019-18_pq00-006.jsonl.gz",
"c4-en-html_cc-main-2019-18_pq00-007.jsonl.gz",
"c4-en-html_cc-main-2019-18_pq00-008.jsonl.gz",
"c4-en-html_cc-main-2019-18_pq00-009.jsonl.gz"… See the full description on the dataset page: https://huggingface.co/datasets/masoudjs/c4-en-html-with-metadata-ppl-clean.c4-en-html-with-training_metadata_allsynthetic-kv-qwen3-8b-with-metadata
Synthetic KV Qwen3 8B — metadata-enhanced 64K
This dataset is an exact key-value retrieval benchmark. The context begins
with a short schema and task description, followed by records in the form
[KEY: VALUE]. Each question asks for the value belonging to one exact key.
The context is intentionally stored once in compact JSONL format. The
questions[i] entry corresponds to answers[i].