Datasets:
File size: 2,460 Bytes
b2a0961 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 | {
"repo_folder": "diversity_oriented",
"paper_setting": "DIVERSITY-ORIENTED",
"internal_run_name": "diversity-first",
"selection_rule": "Top-ranked documents taken within each of the 24 WebOrganizer topics, each topic receiving a share of the source budget proportional to its share of the remaining pool, then rewritten.",
"content": "rewritten",
"composition": {
"note": "This folder is the complete 10B-token mixture the models actually trained on: the shared 5B raw anchor plus this setting's 5B component, already merged and shuffled. The anchor is included in every setting folder.",
"shuffle_seed": 42,
"anchor_docs": 4120164,
"anchor_tokens": 5000002332,
"strategy_docs": 8336411,
"strategy_tokens": 4889635504,
"strategy_token_target": 5000000000,
"strategy_token_shortfall": 110362164,
"shortfall_reason": "Per-topic quotas could not be filled for 4 of 24 topics (Adult, History, Literature, Religion) and the assembly policy was 'A_no_cross_topic_backfill' -- backfilling from other topics would have destroyed the topical balance that defines this setting. This arm therefore ships 4.89B strategy tokens rather than 5.00B.",
"sort_key": "fasttext-ranking-v2 (DESC, within topic)",
"topup_policy": "A_no_cross_topic_backfill"
},
"rewriter_model": "Qwen2.5-7B-Instruct (greedy decoding)",
"prompt_style": "Nemotron-CC wikipedia-style rephrasing and distillation prompts, as released in the FinePhrase repository",
"doc_count": 12456575,
"token_count_llama2": 9889637833,
"shard_count": 25,
"columns": [
"doc_id",
"orig_doc_id",
"text",
"source_prompt",
"train_tokens",
"topic"
],
"provenance": {
"source_folder": "/scratch/bvandur1/zhuicon1/data_rewrite/pretrain/diversity-first/shuffled",
"tokenized_folder": "/scratch/bvandur1/zhuicon1/nanotron_tokenized/diversity-first/tokenized",
"source_evidence": "datatrove executor.json (reader data_folder), confirmed by decoding 50 documents out of the .ds and matching them into this folder (50/50)",
"trained_from_config": "config_diversity-first_10B_1.5B.yaml",
"config_dir": "/scratch/bvandur1/zhuicon1/projects/nanotron/examples",
"seeds": [
42,
43,
44
],
"doc_content_digest_sha256": "6153917f59d5564da5720f94ce4636c0a9b0dddd532a6490ab83de18fd0529ad",
"doc_id_digest_sha256": "fc0ca392c84f4a22b02272e5f60738bad18965f4112105237d790bbc91e86d6b"
}
} |