{ "audit_drain_documents": 0, "contract": "D:\\llm\\hummingbird\\configs\\flightmix_10b_balanced_corpus_contract.yaml", "contract_sha256": "73cd71ff4f7849d939606bd77434df4697d590bdfe2202ba1681b267c63f6786", "contract_validation": { "contract": "D:\\llm\\hummingbird\\configs\\flightmix_10b_balanced_corpus_contract.yaml", "curriculum": "constant_mixture", "pilot_only": false, "source_count": 9, "source_share_sum": 1.0, "source_tokens": { "code": 100000000, "cosmopedia_v2": 1500000000, "dense_pdf_exposition": 300000000, "general_web": 2050000000, "general_web_v2": 3450000000, "natural_text": 1000000000, "science_exposition": 400000000, "science_pdf_qa": 200000000, "worked_mathematics": 1000000000 }, "status": "passed", "train_tokens": 10000000000 }, "curriculum": [ { "document_boundary": true, "end_token": 10000000000, "per_source_tokens": { "code": 100000000, "cosmopedia_v2": 1500000000, "dense_pdf_exposition": 300000000, "general_web": 2050000000, "general_web_v2": 3450000000, "natural_text": 1000000000, "science_exposition": 400000000, "science_pdf_qa": 200000000, "worked_mathematics": 1000000000 }, "phase": "constant_mixture", "start_token": 0, "tokens": 10000000000 } ], "document_end_id": 8, "dtype": "uint16", "elapsed_seconds": 1166.1, "format_version": 5, "mix_policy": "three_phase_exact_token_curriculum", "per_source_documents_all_splits": { "code": { "held_out": 269, "train": 110098, "validation": 257 }, "cosmopedia_v2": { "held_out": 3687, "train": 1482589, "validation": 3700 }, "dense_pdf_exposition": { "held_out": 516, "train": 211198, "validation": 508 }, "general_web": { "held_out": 5209, "train": 2085750, "validation": 5245 }, "general_web_v2": { "held_out": 8644, "train": 3366937, "validation": 8545 }, "natural_text": { "held_out": 2344, "train": 927051, "validation": 2356 }, "science_exposition": { "held_out": 877, "train": 338512, "validation": 862 }, "science_pdf_qa": { "held_out": 1732, "train": 680086, "validation": 1688 }, "worked_mathematics": { "held_out": 2041, "train": 831431, "validation": 2117 } }, "per_source_tokens_all_splits": { "code": { "held_out": 243511, "train": 100000000, "validation": 250292 }, "cosmopedia_v2": { "held_out": 3750173, "train": 1500000000, "validation": 3750712 }, "dense_pdf_exposition": { "held_out": 751319, "train": 300000000, "validation": 750424 }, "general_web": { "held_out": 5123293, "train": 2050000000, "validation": 5125099 }, "general_web_v2": { "held_out": 8626828, "train": 3450000000, "validation": 8625843 }, "natural_text": { "held_out": 2500508, "train": 1000000000, "validation": 2501686 }, "science_exposition": { "held_out": 1001743, "train": 400000000, "validation": 995177 }, "science_pdf_qa": { "held_out": 500686, "train": 200000000, "validation": 500188 }, "worked_mathematics": { "held_out": 2501939, "train": 1000000000, "validation": 2500579 } }, "per_source_train_tokens": { "code": 100000000, "cosmopedia_v2": 1500000000, "dense_pdf_exposition": 300000000, "general_web": 2050000000, "general_web_v2": 3450000000, "natural_text": 1000000000, "science_exposition": 400000000, "science_pdf_qa": 200000000, "worked_mathematics": 1000000000 }, "phase_boundaries": [ 10000000000 ], "source_files": { "code": "E:\\hummingbird\\flightmix_10b_balanced\\materialized\\cleaned_code.jsonl", "cosmopedia_v2": "E:\\hummingbird\\flightmix_10b_balanced\\materialized\\cleaned_cosmopedia_v2.jsonl", "dense_pdf_exposition": "E:\\hummingbird\\flightmix_10b_balanced\\materialized\\cleaned_dense_pdf_exposition.jsonl", "general_web": "E:\\hummingbird\\flightmix_10b_balanced\\materialized\\cleaned_general_web.jsonl", "general_web_v2": "E:\\hummingbird\\flightmix_10b_balanced\\materialized\\cleaned_general_web_v2.jsonl", "natural_text": "E:\\hummingbird\\flightmix_10b_balanced\\materialized\\cleaned_natural_text.jsonl", "science_exposition": "E:\\hummingbird\\flightmix_10b_balanced\\materialized\\cleaned_science_exposition.jsonl", "science_pdf_qa": "E:\\hummingbird\\flightmix_10b_balanced\\materialized\\cleaned_science_pdf_qa.jsonl", "worked_mathematics": "E:\\hummingbird\\flightmix_10b_balanced\\materialized\\cleaned_worked_mathematics.jsonl" }, "source_order": [ "general_web", "general_web_v2", "cosmopedia_v2", "worked_mathematics", "natural_text", "science_exposition", "science_pdf_qa", "dense_pdf_exposition", "code" ], "source_rows_read": { "code": 110629, "cosmopedia_v2": 1489976, "dense_pdf_exposition": 212222, "general_web": 2096207, "general_web_v2": 3384126, "natural_text": 931751, "science_exposition": 340254, "science_pdf_qa": 683506, "worked_mathematics": 835589 }, "split_policy": { "hash": "sha256_first_8_bytes_big_endian", "held_out_buckets": [ 5, 6, 7, 8, 9 ], "modulus": 1000, "validation_buckets": [ 0, 1, 2, 3, 4 ] }, "splits": { "held_out": { "document_ends_file": "held_out_document_ends.npy", "documents": 25319, "quota_tail_fragments": 1, "token_file": "held_out_tokens.uint16", "tokens": 25000000 }, "train": { "document_ends_file": "train_document_ends.npy", "documents": 10033652, "quota_tail_fragments": 9, "token_file": "train_tokens.uint16", "tokens": 10000000000 }, "validation": { "document_ends_file": "validation_document_ends.npy", "documents": 25278, "quota_tail_fragments": 1, "token_file": "validation_tokens.uint16", "tokens": 25000000 } }, "tokenized_source_files": { "code": "E:\\hummingbird\\flightmix_10b_balanced\\materialized\\tokenized_code.uint16", "cosmopedia_v2": "E:\\hummingbird\\flightmix_10b_balanced\\materialized\\tokenized_cosmopedia_v2.uint16", "dense_pdf_exposition": "E:\\hummingbird\\flightmix_10b_balanced\\materialized\\tokenized_dense_pdf_exposition.uint16", "general_web": "E:\\hummingbird\\flightmix_10b_balanced\\materialized\\tokenized_general_web.uint16", "general_web_v2": "E:\\hummingbird\\flightmix_10b_balanced\\materialized\\tokenized_general_web_v2.uint16", "natural_text": "E:\\hummingbird\\flightmix_10b_balanced\\materialized\\tokenized_natural_text.uint16", "science_exposition": "E:\\hummingbird\\flightmix_10b_balanced\\materialized\\tokenized_science_exposition.uint16", "science_pdf_qa": "E:\\hummingbird\\flightmix_10b_balanced\\materialized\\tokenized_science_pdf_qa.uint16", "worked_mathematics": "E:\\hummingbird\\flightmix_10b_balanced\\materialized\\tokenized_worked_mathematics.uint16" }, "tokenizer_path": "D:\\llm\\hummingbird\\artifacts\\runs\\e3_tokenizers\\tok_4k_digit", "tokenizer_sha256": "a788340e936445a7c6a463b11e87f8025be1e63c2fd47a072901328d08ef8818" }