| { |
| "format": "packed-uint32-blocks", |
| "dtype": "uint32", |
| "seq_len": 4096, |
| "train_blocks": 28074124, |
| "val_blocks": 512, |
| "train_tokens": 114991611904, |
| "val_tokens": 2097152, |
| "model": "nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16", |
| "vocab_size": 131072, |
| "doc_sep_id": 11, |
| "bos_per_doc": null, |
| "one_doc_per_block": false, |
| "docs_read": 92527181, |
| "docs_kept": 92527181, |
| "docs_duplicate": 0, |
| "docs_too_short": 0, |
| "records_no_text": 0, |
| "tokens_from_docs": 114993709234, |
| "tokens_separator_fill": 0, |
| "tokens_dropped_tail": 178, |
| "dedup": "off", |
| "val_ratio": 0.002, |
| "seed": 1234, |
| "sources": { |
| "0:006a557a548b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/006a557a548b.jsonl", |
| "docs": 191944, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 221266424 |
| }, |
| "1:00e87840c5c6.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/00e87840c5c6.jsonl", |
| "docs": 189927, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 218647315 |
| }, |
| "2:013e6fcaa8d1.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/013e6fcaa8d1.jsonl", |
| "docs": 44759, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 154327514 |
| }, |
| "3:0239d94f7a75.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/0239d94f7a75.jsonl", |
| "docs": 135129, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 141972513 |
| }, |
| "4:02c3c631b724.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/02c3c631b724.jsonl", |
| "docs": 131353, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 168541197 |
| }, |
| "5:038e2895c16b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/038e2895c16b.jsonl", |
| "docs": 177570, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 203737063 |
| }, |
| "6:0398729c97d0.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/0398729c97d0.jsonl", |
| "docs": 172395, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 195434417 |
| }, |
| "7:03b70c888a7c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/03b70c888a7c.jsonl", |
| "docs": 175707, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 230599543 |
| }, |
| "8:03dd357d901e.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/03dd357d901e.jsonl", |
| "docs": 181160, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 238077052 |
| }, |
| "9:049387659f2f.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/049387659f2f.jsonl", |
| "docs": 132424, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 171301700 |
| }, |
| "10:0543ba65ff3b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/0543ba65ff3b.jsonl", |
| "docs": 128760, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 204263612 |
| }, |
| "11:058cac8c4174.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/058cac8c4174.jsonl", |
| "docs": 183890, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 210696154 |
| }, |
| "12:061800b1d43c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/061800b1d43c.jsonl", |
| "docs": 131136, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 201256239 |
| }, |
| "13:067e756905e0.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/067e756905e0.jsonl", |
| "docs": 158630, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 201556853 |
| }, |
| "14:0697940c34d5.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/0697940c34d5.jsonl", |
| "docs": 77626, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 200225544 |
| }, |
| "15:06e0307edfc8.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/06e0307edfc8.jsonl", |
| "docs": 210912, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 246732417 |
| }, |
| "16:06f6fe21eac5.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/06f6fe21eac5.jsonl", |
| "docs": 197456, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 228784792 |
| }, |
| "17:07241244fffb.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/07241244fffb.jsonl", |
| "docs": 162610, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 211726311 |
| }, |
| "18:073e8999e274.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/073e8999e274.jsonl", |
| "docs": 188141, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 215463039 |
| }, |
| "19:07662f65a89d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/07662f65a89d.jsonl", |
| "docs": 133894, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 172735023 |
| }, |
| "20:07a0f645905f.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/07a0f645905f.jsonl", |
| "docs": 207335, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 242894460 |
| }, |
| "21:07ad96d38787.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/07ad96d38787.jsonl", |
| "docs": 240523, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 213153397 |
| }, |
| "22:07dc650a6661.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/07dc650a6661.jsonl", |
| "docs": 168027, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 190648470 |
| }, |
| "23:098947c93551.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/098947c93551.jsonl", |
| "docs": 142688, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 185303217 |
| }, |
| "24:0b3f206ea275.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/0b3f206ea275.jsonl", |
| "docs": 158096, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 206372081 |
| }, |
| "25:0bb6e41f313b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/0bb6e41f313b.jsonl", |
| "docs": 125798, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 184532250 |
| }, |
| "26:0c9dd05ff6b6.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/0c9dd05ff6b6.jsonl", |
| "docs": 160623, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 209804739 |
| }, |
| "27:0cadf0935f51.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/0cadf0935f51.jsonl", |
| "docs": 81457, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 208040956 |
| }, |
| "28:0cd21295b7e4.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/0cd21295b7e4.jsonl", |
| "docs": 157251, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 204620403 |
| }, |
| "29:0ce005b98304.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/0ce005b98304.jsonl", |
| "docs": 145700, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 187835451 |
| }, |
| "30:0d869da9f65d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/0d869da9f65d.jsonl", |
| "docs": 135120, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 173548496 |
| }, |
| "31:0e13cddbcf7f.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/0e13cddbcf7f.jsonl", |
| "docs": 156565, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 204842465 |
| }, |
| "32:0e2a86a838cb.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/0e2a86a838cb.jsonl", |
| "docs": 148655, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 193197406 |
| }, |
| "33:0e930e21d2f7.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/0e930e21d2f7.jsonl", |
| "docs": 176968, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 200837470 |
| }, |
| "34:0f1ff16f2562.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/0f1ff16f2562.jsonl", |
| "docs": 145398, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 188674232 |
| }, |
| "35:0f53b29ef334.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/0f53b29ef334.jsonl", |
| "docs": 159080, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 217778247 |
| }, |
| "36:0f8125619083.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/0f8125619083.jsonl", |
| "docs": 181675, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 207616660 |
| }, |
| "37:0fcd717d436e.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/0fcd717d436e.jsonl", |
| "docs": 164428, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 208633203 |
| }, |
| "38:100b54035ce7.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/100b54035ce7.jsonl", |
| "docs": 128881, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 88960394 |
| }, |
| "39:100de73a067b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/100de73a067b.jsonl", |
| "docs": 172580, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 196419348 |
| }, |
| "40:100ea269ad6b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/100ea269ad6b.jsonl", |
| "docs": 100541, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 236085928 |
| }, |
| "41:102223161d3f.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/102223161d3f.jsonl", |
| "docs": 96824, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 235146865 |
| }, |
| "42:113869c09b9d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/113869c09b9d.jsonl", |
| "docs": 55569, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 159127046 |
| }, |
| "43:1215e9704de3.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/1215e9704de3.jsonl", |
| "docs": 127308, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 164563798 |
| }, |
| "44:124895a79422.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/124895a79422.jsonl", |
| "docs": 203721, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 237088008 |
| }, |
| "45:1308c26610f0.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/1308c26610f0.jsonl", |
| "docs": 178347, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 203435041 |
| }, |
| "46:130905232836.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/130905232836.jsonl", |
| "docs": 130529, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 169416084 |
| }, |
| "47:131cbc0171f7.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/131cbc0171f7.jsonl", |
| "docs": 97008, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 198207156 |
| }, |
| "48:13419f40ce3b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/13419f40ce3b.jsonl", |
| "docs": 143604, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 185374328 |
| }, |
| "49:13665261a7e6.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/13665261a7e6.jsonl", |
| "docs": 65941, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 114335884 |
| }, |
| "50:13685cb025b4.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/13685cb025b4.jsonl", |
| "docs": 180154, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 205995337 |
| }, |
| "51:13f269d268be.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/13f269d268be.jsonl", |
| "docs": 70022, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 190134546 |
| }, |
| "52:147778bf10c6.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/147778bf10c6.jsonl", |
| "docs": 156880, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 201858023 |
| }, |
| "53:14b2458d8ed8.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/14b2458d8ed8.jsonl", |
| "docs": 174706, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 199855708 |
| }, |
| "54:14f611302c97.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/14f611302c97.jsonl", |
| "docs": 177485, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 213286222 |
| }, |
| "55:154320e790d9.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/154320e790d9.jsonl", |
| "docs": 146581, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 191102697 |
| }, |
| "56:15fd5ac0f1d8.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/15fd5ac0f1d8.jsonl", |
| "docs": 198865, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 230380108 |
| }, |
| "57:1602ffea88c7.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/1602ffea88c7.jsonl", |
| "docs": 141317, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 183147572 |
| }, |
| "58:16c06ae40e2a.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/16c06ae40e2a.jsonl", |
| "docs": 207881, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 242124202 |
| }, |
| "59:1722d20b6011.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/1722d20b6011.jsonl", |
| "docs": 86682, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 214805469 |
| }, |
| "60:178ec1e97892.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/178ec1e97892.jsonl", |
| "docs": 181986, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 208623073 |
| }, |
| "61:17a1d3669019.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/17a1d3669019.jsonl", |
| "docs": 155200, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 202234395 |
| }, |
| "62:17fb2d023cf3.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/17fb2d023cf3.jsonl", |
| "docs": 206103, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 241301676 |
| }, |
| "63:1845cd155452.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/1845cd155452.jsonl", |
| "docs": 193174, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 222251141 |
| }, |
| "64:18747c04f3a9.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/18747c04f3a9.jsonl", |
| "docs": 157430, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 206392462 |
| }, |
| "65:1876e074a79d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/1876e074a79d.jsonl", |
| "docs": 141421, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 181061976 |
| }, |
| "66:1905fdbe06ce.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/1905fdbe06ce.jsonl", |
| "docs": 185266, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 212953974 |
| }, |
| "67:194e9855e434.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/194e9855e434.jsonl", |
| "docs": 205107, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 241289006 |
| }, |
| "68:197880c8d348.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/197880c8d348.jsonl", |
| "docs": 209951, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 247280548 |
| }, |
| "69:19a5936841e6.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/19a5936841e6.jsonl", |
| "docs": 144843, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 187408947 |
| }, |
| "70:1a6ea558245a.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/1a6ea558245a.jsonl", |
| "docs": 151020, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 108136187 |
| }, |
| "71:1c162ef796d9.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/1c162ef796d9.jsonl", |
| "docs": 178188, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 234417990 |
| }, |
| "72:1c16fe8a0fc9.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/1c16fe8a0fc9.jsonl", |
| "docs": 58029, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 189989538 |
| }, |
| "73:1c996eaaef69.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/1c996eaaef69.jsonl", |
| "docs": 133325, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 173169002 |
| }, |
| "74:1e5a3905fca9.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/1e5a3905fca9.jsonl", |
| "docs": 208995, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 244703684 |
| }, |
| "75:1f0facb83393.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/1f0facb83393.jsonl", |
| "docs": 154964, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 193085310 |
| }, |
| "76:1f551a567958.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/1f551a567958.jsonl", |
| "docs": 150519, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 195345541 |
| }, |
| "77:20ee195b01eb.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/20ee195b01eb.jsonl", |
| "docs": 148116, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 105579613 |
| }, |
| "78:210303145457.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/210303145457.jsonl", |
| "docs": 176584, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 201161613 |
| }, |
| "79:2103627c302b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2103627c302b.jsonl", |
| "docs": 133036, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 171372330 |
| }, |
| "80:2168d896e172.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2168d896e172.jsonl", |
| "docs": 156271, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 191890007 |
| }, |
| "81:21b8d8a282d7.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/21b8d8a282d7.jsonl", |
| "docs": 131541, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 171496953 |
| }, |
| "82:21d797f1a9c9.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/21d797f1a9c9.jsonl", |
| "docs": 208757, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 243684037 |
| }, |
| "83:2263989a0681.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2263989a0681.jsonl", |
| "docs": 135942, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 176026838 |
| }, |
| "84:241800191c9d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/241800191c9d.jsonl", |
| "docs": 158163, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 203488461 |
| }, |
| "85:2431999e289f.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2431999e289f.jsonl", |
| "docs": 177322, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 202049990 |
| }, |
| "86:2436a15b8055.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2436a15b8055.jsonl", |
| "docs": 247090, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 137116313 |
| }, |
| "87:25188e4cab2f.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/25188e4cab2f.jsonl", |
| "docs": 168808, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 192397878 |
| }, |
| "88:252af67c2a96.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/252af67c2a96.jsonl", |
| "docs": 204172, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 238204299 |
| }, |
| "89:2592b728e544.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2592b728e544.jsonl", |
| "docs": 154334, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 200780264 |
| }, |
| "90:259f605e3753.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/259f605e3753.jsonl", |
| "docs": 169552, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 221240091 |
| }, |
| "91:25b1c0606f63.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/25b1c0606f63.jsonl", |
| "docs": 178790, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 204251974 |
| }, |
| "92:2603d9a4a969.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2603d9a4a969.jsonl", |
| "docs": 196962, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 226328589 |
| }, |
| "93:267e774d5321.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/267e774d5321.jsonl", |
| "docs": 193851, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 223422149 |
| }, |
| "94:27f86ea79fab.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/27f86ea79fab.jsonl", |
| "docs": 171544, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 195042216 |
| }, |
| "95:2808761ee4e6.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2808761ee4e6.jsonl", |
| "docs": 158863, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 206713619 |
| }, |
| "96:2876c4c25fea.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2876c4c25fea.jsonl", |
| "docs": 178871, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 237628109 |
| }, |
| "97:28881755e790.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/28881755e790.jsonl", |
| "docs": 177589, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 202159994 |
| }, |
| "98:298018034c99.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/298018034c99.jsonl", |
| "docs": 148923, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 178319579 |
| }, |
| "99:2a4b21e48137.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2a4b21e48137.jsonl", |
| "docs": 137656, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 178696595 |
| }, |
| "100:2a5868d38fbe.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2a5868d38fbe.jsonl", |
| "docs": 154081, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 200749734 |
| }, |
| "101:2a705e97ee1c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2a705e97ee1c.jsonl", |
| "docs": 159811, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 210845815 |
| }, |
| "102:2a9fea55e145.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2a9fea55e145.jsonl", |
| "docs": 222133, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 163435442 |
| }, |
| "103:2bbb90429216.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2bbb90429216.jsonl", |
| "docs": 187747, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 215364700 |
| }, |
| "104:2bd332ce8eac.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2bd332ce8eac.jsonl", |
| "docs": 179128, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 203646587 |
| }, |
| "105:2c25323f1ddf.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2c25323f1ddf.jsonl", |
| "docs": 149668, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 195581565 |
| }, |
| "106:2c5e52202a84.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2c5e52202a84.jsonl", |
| "docs": 86976, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 122593700 |
| }, |
| "107:2c7bfe8be2c2.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2c7bfe8be2c2.jsonl", |
| "docs": 200693, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 232127535 |
| }, |
| "108:2cd767d36b16.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2cd767d36b16.jsonl", |
| "docs": 179590, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 235213328 |
| }, |
| "109:2cfd77b6b896.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2cfd77b6b896.jsonl", |
| "docs": 169275, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 193117234 |
| }, |
| "110:2dd2d4f00420.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2dd2d4f00420.jsonl", |
| "docs": 144389, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 169525022 |
| }, |
| "111:2de57a11d20c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2de57a11d20c.jsonl", |
| "docs": 151322, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 108550073 |
| }, |
| "112:2e2a349d3c07.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2e2a349d3c07.jsonl", |
| "docs": 166722, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 189448428 |
| }, |
| "113:2eb0a5dd2132.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2eb0a5dd2132.jsonl", |
| "docs": 176887, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 200608214 |
| }, |
| "114:2ec9a90a629c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2ec9a90a629c.jsonl", |
| "docs": 170757, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 222053189 |
| }, |
| "115:2f4e8fa9d3f3.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2f4e8fa9d3f3.jsonl", |
| "docs": 176250, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 200653091 |
| }, |
| "116:2ffaa51f4355.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/2ffaa51f4355.jsonl", |
| "docs": 183475, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 236615916 |
| }, |
| "117:3041cf64b5b6.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/3041cf64b5b6.jsonl", |
| "docs": 121591, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 171985292 |
| }, |
| "118:30568139da4d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/30568139da4d.jsonl", |
| "docs": 175908, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 201275197 |
| }, |
| "119:30a8d2e35fdb.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/30a8d2e35fdb.jsonl", |
| "docs": 175768, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 203455360 |
| }, |
| "120:30e5af924eb3.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/30e5af924eb3.jsonl", |
| "docs": 180736, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 207867684 |
| }, |
| "121:328ac22b1d13.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/328ac22b1d13.jsonl", |
| "docs": 187498, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 215094862 |
| }, |
| "122:32f88137c3e7.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/32f88137c3e7.jsonl", |
| "docs": 130153, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 166601292 |
| }, |
| "123:3423e07b6e8a.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/3423e07b6e8a.jsonl", |
| "docs": 146630, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 190083013 |
| }, |
| "124:351d664e8f49.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/351d664e8f49.jsonl", |
| "docs": 120330, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 83146884 |
| }, |
| "125:3560e6cf4e7c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/3560e6cf4e7c.jsonl", |
| "docs": 210927, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 247426728 |
| }, |
| "126:38063869d82c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/38063869d82c.jsonl", |
| "docs": 174451, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 198855286 |
| }, |
| "127:39ed54d636bb.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/39ed54d636bb.jsonl", |
| "docs": 168513, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 190445219 |
| }, |
| "128:3a18d6a32ee4.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/3a18d6a32ee4.jsonl", |
| "docs": 171482, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 194871530 |
| }, |
| "129:3ac0073c2ac0.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/3ac0073c2ac0.jsonl", |
| "docs": 128964, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 166479603 |
| }, |
| "130:3ac0c028277c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/3ac0c028277c.jsonl", |
| "docs": 145085, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 187208083 |
| }, |
| "131:3b30eb0329ce.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/3b30eb0329ce.jsonl", |
| "docs": 163276, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 213465287 |
| }, |
| "132:3b51de0f1a83.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/3b51de0f1a83.jsonl", |
| "docs": 154483, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 201704467 |
| }, |
| "133:3b73186902a7.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/3b73186902a7.jsonl", |
| "docs": 179986, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 205421206 |
| }, |
| "134:3bd460139936.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/3bd460139936.jsonl", |
| "docs": 166746, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 188924157 |
| }, |
| "135:3c1d8261d3fc.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/3c1d8261d3fc.jsonl", |
| "docs": 140046, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 179383191 |
| }, |
| "136:3c5727111949.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/3c5727111949.jsonl", |
| "docs": 167727, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 189874756 |
| }, |
| "137:3ca8a2b1acc8.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/3ca8a2b1acc8.jsonl", |
| "docs": 198605, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 228530294 |
| }, |
| "138:3d69d09f6b5a.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/3d69d09f6b5a.jsonl", |
| "docs": 190994, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 222244458 |
| }, |
| "139:3d8c8ef8d50e.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/3d8c8ef8d50e.jsonl", |
| "docs": 178449, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 233970670 |
| }, |
| "140:3ea2751a1bd2.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/3ea2751a1bd2.jsonl", |
| "docs": 180105, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 206565737 |
| }, |
| "141:3ea62ecc3683.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/3ea62ecc3683.jsonl", |
| "docs": 134453, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 168564115 |
| }, |
| "142:3ec8a750d259.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/3ec8a750d259.jsonl", |
| "docs": 190604, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 218562132 |
| }, |
| "143:3ecddcc7dd4c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/3ecddcc7dd4c.jsonl", |
| "docs": 149228, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 194778014 |
| }, |
| "144:3f06bd6ff4ad.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/3f06bd6ff4ad.jsonl", |
| "docs": 69637, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 208787713 |
| }, |
| "145:4056c276f74e.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/4056c276f74e.jsonl", |
| "docs": 141790, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 184383259 |
| }, |
| "146:409813c8b53e.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/409813c8b53e.jsonl", |
| "docs": 240075, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 178608381 |
| }, |
| "147:40a1a3603ada.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/40a1a3603ada.jsonl", |
| "docs": 185614, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 213141228 |
| }, |
| "148:40d800ee1dd3.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/40d800ee1dd3.jsonl", |
| "docs": 155562, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 202628947 |
| }, |
| "149:40db4c254cd2.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/40db4c254cd2.jsonl", |
| "docs": 132901, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 172038427 |
| }, |
| "150:410fe90e41f8.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/410fe90e41f8.jsonl", |
| "docs": 71613, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 199026898 |
| }, |
| "151:412264e0310c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/412264e0310c.jsonl", |
| "docs": 179634, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 235722685 |
| }, |
| "152:421b1a79ec8d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/421b1a79ec8d.jsonl", |
| "docs": 146588, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 189596676 |
| }, |
| "153:421e270a4bed.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/421e270a4bed.jsonl", |
| "docs": 214349, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 156836554 |
| }, |
| "154:4249bd5c5d6e.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/4249bd5c5d6e.jsonl", |
| "docs": 155717, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 203034645 |
| }, |
| "155:42658ca048f0.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/42658ca048f0.jsonl", |
| "docs": 148048, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 193670156 |
| }, |
| "156:4321f0564e1c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/4321f0564e1c.jsonl", |
| "docs": 200335, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 232484184 |
| }, |
| "157:436627065233.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/436627065233.jsonl", |
| "docs": 197793, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 228442304 |
| }, |
| "158:4396beb47065.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/4396beb47065.jsonl", |
| "docs": 171972, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 194678442 |
| }, |
| "159:43bc2bea42f1.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/43bc2bea42f1.jsonl", |
| "docs": 142797, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 183710444 |
| }, |
| "160:43d2b20d3218.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/43d2b20d3218.jsonl", |
| "docs": 355847, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 167264665 |
| }, |
| "161:444f86359689.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/444f86359689.jsonl", |
| "docs": 200190, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 230806299 |
| }, |
| "162:44d972da570e.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/44d972da570e.jsonl", |
| "docs": 133091, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 170912616 |
| }, |
| "163:44febfa6dfd0.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/44febfa6dfd0.jsonl", |
| "docs": 177697, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 238856444 |
| }, |
| "164:45cdee5b20a6.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/45cdee5b20a6.jsonl", |
| "docs": 68745, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 239988442 |
| }, |
| "165:45d676704ae4.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/45d676704ae4.jsonl", |
| "docs": 130042, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 167547649 |
| }, |
| "166:462989c454a6.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/462989c454a6.jsonl", |
| "docs": 141316, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 98034314 |
| }, |
| "167:46fa28e13b05.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/46fa28e13b05.jsonl", |
| "docs": 152374, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 206645432 |
| }, |
| "168:47fdaccdd97d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/47fdaccdd97d.jsonl", |
| "docs": 145966, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 189367238 |
| }, |
| "169:487a119b1a93.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/487a119b1a93.jsonl", |
| "docs": 167952, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 218721365 |
| }, |
| "170:48eecaae33e2.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/48eecaae33e2.jsonl", |
| "docs": 53521, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 206253629 |
| }, |
| "171:49ef22d216d6.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/49ef22d216d6.jsonl", |
| "docs": 202096, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 236620104 |
| }, |
| "172:4a8d4ca498f5.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/4a8d4ca498f5.jsonl", |
| "docs": 144568, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 100869332 |
| }, |
| "173:4afcbfc83266.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/4afcbfc83266.jsonl", |
| "docs": 138111, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 179299052 |
| }, |
| "174:4b0e45949d20.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/4b0e45949d20.jsonl", |
| "docs": 178697, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 205294807 |
| }, |
| "175:4b21e32cb0e3.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/4b21e32cb0e3.jsonl", |
| "docs": 181149, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 237613902 |
| }, |
| "176:4b40bd01292d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/4b40bd01292d.jsonl", |
| "docs": 154038, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 201643177 |
| }, |
| "177:4df11ca8d1ff.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/4df11ca8d1ff.jsonl", |
| "docs": 152686, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 200948154 |
| }, |
| "178:4e16a660d24a.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/4e16a660d24a.jsonl", |
| "docs": 169776, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 220807228 |
| }, |
| "179:4e4c1e667168.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/4e4c1e667168.jsonl", |
| "docs": 176673, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 199881953 |
| }, |
| "180:4e79afdb4362.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/4e79afdb4362.jsonl", |
| "docs": 180292, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 232935120 |
| }, |
| "181:4e847e8b1b91.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/4e847e8b1b91.jsonl", |
| "docs": 176017, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 200232628 |
| }, |
| "182:4e9f47187836.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/4e9f47187836.jsonl", |
| "docs": 205343, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 238802361 |
| }, |
| "183:4ec814590987.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/4ec814590987.jsonl", |
| "docs": 192728, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 138620385 |
| }, |
| "184:4ed457dcd0a3.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/4ed457dcd0a3.jsonl", |
| "docs": 188400, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 216982049 |
| }, |
| "185:4f6265935021.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/4f6265935021.jsonl", |
| "docs": 163326, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 213553460 |
| }, |
| "186:4f7a1a669724.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/4f7a1a669724.jsonl", |
| "docs": 168229, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 218744028 |
| }, |
| "187:4fa2548aa658.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/4fa2548aa658.jsonl", |
| "docs": 184088, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 247405877 |
| }, |
| "188:5043368e5722.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/5043368e5722.jsonl", |
| "docs": 77165, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 202509836 |
| }, |
| "189:50d95ba54406.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/50d95ba54406.jsonl", |
| "docs": 139237, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 179697622 |
| }, |
| "190:51cae0f3f902.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/51cae0f3f902.jsonl", |
| "docs": 181554, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 207726020 |
| }, |
| "191:521317c8792d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/521317c8792d.jsonl", |
| "docs": 198969, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 229623630 |
| }, |
| "192:52a85db81e21.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/52a85db81e21.jsonl", |
| "docs": 155671, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 201111604 |
| }, |
| "193:5321b6bd8cd4.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/5321b6bd8cd4.jsonl", |
| "docs": 174358, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 234151672 |
| }, |
| "194:54b1c1c9fae3.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/54b1c1c9fae3.jsonl", |
| "docs": 168815, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 219599904 |
| }, |
| "195:54c91dc6b0b5.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/54c91dc6b0b5.jsonl", |
| "docs": 170253, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 193266093 |
| }, |
| "196:5534ed01e86d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/5534ed01e86d.jsonl", |
| "docs": 129342, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 165741603 |
| }, |
| "197:558a762fd0fa.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/558a762fd0fa.jsonl", |
| "docs": 204565, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 240165579 |
| }, |
| "198:56612aecc9b0.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/56612aecc9b0.jsonl", |
| "docs": 230980, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 171018693 |
| }, |
| "199:566eedabec8a.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/566eedabec8a.jsonl", |
| "docs": 195887, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 226152781 |
| }, |
| "200:571df6e89820.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/571df6e89820.jsonl", |
| "docs": 164518, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 213738548 |
| }, |
| "201:586eef3e7388.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/586eef3e7388.jsonl", |
| "docs": 184821, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 188564021 |
| }, |
| "202:590cadc98a13.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/590cadc98a13.jsonl", |
| "docs": 170396, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 194248841 |
| }, |
| "203:59729a057c76.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/59729a057c76.jsonl", |
| "docs": 142506, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 182964554 |
| }, |
| "204:5a1e7825061e.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/5a1e7825061e.jsonl", |
| "docs": 147480, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 192042305 |
| }, |
| "205:5a5276bced4b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/5a5276bced4b.jsonl", |
| "docs": 146361, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 188389073 |
| }, |
| "206:5b02e2627a8a.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/5b02e2627a8a.jsonl", |
| "docs": 165559, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 215542120 |
| }, |
| "207:5c26af0f4604.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/5c26af0f4604.jsonl", |
| "docs": 176532, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 202201632 |
| }, |
| "208:5ccf4d18ab55.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/5ccf4d18ab55.jsonl", |
| "docs": 148545, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 192137269 |
| }, |
| "209:5d97655ca79d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/5d97655ca79d.jsonl", |
| "docs": 125886, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 162647065 |
| }, |
| "210:5e6031468f03.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/5e6031468f03.jsonl", |
| "docs": 161267, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 210865007 |
| }, |
| "211:5f4c253bea92.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/5f4c253bea92.jsonl", |
| "docs": 172325, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 196596003 |
| }, |
| "212:5f7843934b23.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/5f7843934b23.jsonl", |
| "docs": 174300, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 197548114 |
| }, |
| "213:60aabca832cd.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/60aabca832cd.jsonl", |
| "docs": 166299, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 217294252 |
| }, |
| "214:60c7124b3d0a.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/60c7124b3d0a.jsonl", |
| "docs": 153857, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 107998715 |
| }, |
| "215:60f981ee6018.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/60f981ee6018.jsonl", |
| "docs": 149705, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 195362943 |
| }, |
| "216:613e01f4ec60.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/613e01f4ec60.jsonl", |
| "docs": 127016, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 162845185 |
| }, |
| "217:6183dc302c20.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/6183dc302c20.jsonl", |
| "docs": 162297, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 222070760 |
| }, |
| "218:620fa0b2cd8e.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/620fa0b2cd8e.jsonl", |
| "docs": 194652, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 224033055 |
| }, |
| "219:6250a5a887cf.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/6250a5a887cf.jsonl", |
| "docs": 206535, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 239887556 |
| }, |
| "220:63597b3c12a8.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/63597b3c12a8.jsonl", |
| "docs": 264459, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 201197230 |
| }, |
| "221:6374937a255b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/6374937a255b.jsonl", |
| "docs": 209885, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 245930157 |
| }, |
| "222:6374e0867921.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/6374e0867921.jsonl", |
| "docs": 158879, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 212229705 |
| }, |
| "223:637eae08184b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/637eae08184b.jsonl", |
| "docs": 186669, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 213835662 |
| }, |
| "224:64031bbcd771.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/64031bbcd771.jsonl", |
| "docs": 161646, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 115970568 |
| }, |
| "225:641721a16a10.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/641721a16a10.jsonl", |
| "docs": 73299, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 183596435 |
| }, |
| "226:64a0898c599a.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/64a0898c599a.jsonl", |
| "docs": 198278, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 229105760 |
| }, |
| "227:661ad5009abe.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/661ad5009abe.jsonl", |
| "docs": 136562, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 177087698 |
| }, |
| "228:6652cf20dc87.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/6652cf20dc87.jsonl", |
| "docs": 162527, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 211582234 |
| }, |
| "229:668947d4f028.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/668947d4f028.jsonl", |
| "docs": 175251, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 199296166 |
| }, |
| "230:66e7db3ffaa3.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/66e7db3ffaa3.jsonl", |
| "docs": 166091, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 216798650 |
| }, |
| "231:66f8de1d63c1.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/66f8de1d63c1.jsonl", |
| "docs": 165835, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 217285185 |
| }, |
| "232:67affcd75de7.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/67affcd75de7.jsonl", |
| "docs": 149307, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 197876511 |
| }, |
| "233:67eed0085d3d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/67eed0085d3d.jsonl", |
| "docs": 148810, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 197038816 |
| }, |
| "234:696f48587d33.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/696f48587d33.jsonl", |
| "docs": 188624, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 216926970 |
| }, |
| "235:69921981e3ef.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/69921981e3ef.jsonl", |
| "docs": 139321, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 180182938 |
| }, |
| "236:6a0b2edecd64.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/6a0b2edecd64.jsonl", |
| "docs": 169849, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 193260605 |
| }, |
| "237:6a22a0323b6e.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/6a22a0323b6e.jsonl", |
| "docs": 209084, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 245681614 |
| }, |
| "238:6c869c7fa700.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/6c869c7fa700.jsonl", |
| "docs": 144559, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 186101298 |
| }, |
| "239:6cf218ae7d3e.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/6cf218ae7d3e.jsonl", |
| "docs": 177732, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 231979489 |
| }, |
| "240:6d5c10a2a578.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/6d5c10a2a578.jsonl", |
| "docs": 198445, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 228471833 |
| }, |
| "241:6dd852cc7f01.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/6dd852cc7f01.jsonl", |
| "docs": 151633, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 197493083 |
| }, |
| "242:6f3e34afe4e7.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/6f3e34afe4e7.jsonl", |
| "docs": 140596, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 180752124 |
| }, |
| "243:6f5da0d82ecf.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/6f5da0d82ecf.jsonl", |
| "docs": 153903, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 199670430 |
| }, |
| "244:6f801da2d4f8.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/6f801da2d4f8.jsonl", |
| "docs": 130014, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 189595931 |
| }, |
| "245:6fa5486dee67.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/6fa5486dee67.jsonl", |
| "docs": 199913, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 231604475 |
| }, |
| "246:6fb3ca54077a.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/6fb3ca54077a.jsonl", |
| "docs": 187957, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 215897222 |
| }, |
| "247:70ccd35bde64.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/70ccd35bde64.jsonl", |
| "docs": 181874, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 220673493 |
| }, |
| "248:7100148db806.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/7100148db806.jsonl", |
| "docs": 132985, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 91987586 |
| }, |
| "249:710c932cca6f.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/710c932cca6f.jsonl", |
| "docs": 128549, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 199468775 |
| }, |
| "250:711355cc767b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/711355cc767b.jsonl", |
| "docs": 201714, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 233702543 |
| }, |
| "251:71e7882f50bc.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/71e7882f50bc.jsonl", |
| "docs": 183038, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 209341745 |
| }, |
| "252:722504a3700c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/722504a3700c.jsonl", |
| "docs": 175054, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 173627368 |
| }, |
| "253:72601503a0bc.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/72601503a0bc.jsonl", |
| "docs": 157646, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 203189092 |
| }, |
| "254:7322ddce9722.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/7322ddce9722.jsonl", |
| "docs": 169314, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 192336788 |
| }, |
| "255:732cda4a4aae.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/732cda4a4aae.jsonl", |
| "docs": 127601, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 163635374 |
| }, |
| "256:73810cab18c0.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/73810cab18c0.jsonl", |
| "docs": 153009, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 200629083 |
| }, |
| "257:759c17de0efc.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/759c17de0efc.jsonl", |
| "docs": 54933, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 174657171 |
| }, |
| "258:75bac606ae40.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/75bac606ae40.jsonl", |
| "docs": 206609, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 150512321 |
| }, |
| "259:75c25d61e873.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/75c25d61e873.jsonl", |
| "docs": 144589, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 177651172 |
| }, |
| "260:768e8acad690.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/768e8acad690.jsonl", |
| "docs": 81721, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 199343959 |
| }, |
| "261:769e4b256de9.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/769e4b256de9.jsonl", |
| "docs": 200116, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 233160670 |
| }, |
| "262:776b44190cd2.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/776b44190cd2.jsonl", |
| "docs": 171588, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 195341877 |
| }, |
| "263:7862d8516d3c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/7862d8516d3c.jsonl", |
| "docs": 181124, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 240498168 |
| }, |
| "264:78693d5b81bb.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/78693d5b81bb.jsonl", |
| "docs": 173241, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 198206179 |
| }, |
| "265:792a0459ccf1.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/792a0459ccf1.jsonl", |
| "docs": 159570, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 208019558 |
| }, |
| "266:797768b0fe7b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/797768b0fe7b.jsonl", |
| "docs": 173032, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 226504121 |
| }, |
| "267:797fea15337f.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/797fea15337f.jsonl", |
| "docs": 208528, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 243915131 |
| }, |
| "268:7a3732995971.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/7a3732995971.jsonl", |
| "docs": 150729, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 196085082 |
| }, |
| "269:7bb29e2c7f88.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/7bb29e2c7f88.jsonl", |
| "docs": 250322, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 191249737 |
| }, |
| "270:7cc3a1f61228.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/7cc3a1f61228.jsonl", |
| "docs": 199294, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 229506535 |
| }, |
| "271:7ccbc2d4bd7e.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/7ccbc2d4bd7e.jsonl", |
| "docs": 169767, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 192154744 |
| }, |
| "272:7cd09111457c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/7cd09111457c.jsonl", |
| "docs": 166120, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 216212568 |
| }, |
| "273:7ce08200b820.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/7ce08200b820.jsonl", |
| "docs": 171725, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 194617768 |
| }, |
| "274:7cfc31dd18fc.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/7cfc31dd18fc.jsonl", |
| "docs": 174536, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 228071869 |
| }, |
| "275:7cfd43cf2150.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/7cfd43cf2150.jsonl", |
| "docs": 124858, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 86190009 |
| }, |
| "276:7d172dcd8fb2.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/7d172dcd8fb2.jsonl", |
| "docs": 208217, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 243061041 |
| }, |
| "277:7dcb4e502d10.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/7dcb4e502d10.jsonl", |
| "docs": 160675, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 209753577 |
| }, |
| "278:7df9c20d2aef.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/7df9c20d2aef.jsonl", |
| "docs": 202697, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 234398510 |
| }, |
| "279:7e1289dc8d6a.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/7e1289dc8d6a.jsonl", |
| "docs": 183798, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 210540777 |
| }, |
| "280:7f095e006828.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/7f095e006828.jsonl", |
| "docs": 145785, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 239162697 |
| }, |
| "281:807961f3046c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/807961f3046c.jsonl", |
| "docs": 210080, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 246652334 |
| }, |
| "282:80d5733f1506.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/80d5733f1506.jsonl", |
| "docs": 90796, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 181471359 |
| }, |
| "283:8138cd1cfe2f.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/8138cd1cfe2f.jsonl", |
| "docs": 177879, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 202498478 |
| }, |
| "284:814a9194d74b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/814a9194d74b.jsonl", |
| "docs": 200211, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 231853827 |
| }, |
| "285:8207b7655103.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/8207b7655103.jsonl", |
| "docs": 130923, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 232175341 |
| }, |
| "286:8277f1e5614e.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/8277f1e5614e.jsonl", |
| "docs": 336029, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 148256909 |
| }, |
| "287:837bf8c0fb02.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/837bf8c0fb02.jsonl", |
| "docs": 135597, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 175286529 |
| }, |
| "288:83bb14989bc0.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/83bb14989bc0.jsonl", |
| "docs": 52312, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 159888695 |
| }, |
| "289:84c03428cb1d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/84c03428cb1d.jsonl", |
| "docs": 208290, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 244869146 |
| }, |
| "290:84c8626f450d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/84c8626f450d.jsonl", |
| "docs": 136823, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 176979275 |
| }, |
| "291:852e66c96b59.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/852e66c96b59.jsonl", |
| "docs": 146357, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 166766087 |
| }, |
| "292:85342b695d1c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/85342b695d1c.jsonl", |
| "docs": 207955, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 244193597 |
| }, |
| "293:853f3710ff9f.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/853f3710ff9f.jsonl", |
| "docs": 161647, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 210039522 |
| }, |
| "294:85c150ada652.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/85c150ada652.jsonl", |
| "docs": 152797, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 109273210 |
| }, |
| "295:85e01d7eeafe.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/85e01d7eeafe.jsonl", |
| "docs": 60521, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 190439248 |
| }, |
| "296:8661f8fa8fbb.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/8661f8fa8fbb.jsonl", |
| "docs": 166048, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 215893097 |
| }, |
| "297:875db5e143ef.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/875db5e143ef.jsonl", |
| "docs": 137626, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 178069461 |
| }, |
| "298:87e473b210d5.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/87e473b210d5.jsonl", |
| "docs": 136177, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 94882886 |
| }, |
| "299:880083b30bf6.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/880083b30bf6.jsonl", |
| "docs": 203768, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 235623640 |
| }, |
| "300:88043842534e.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/88043842534e.jsonl", |
| "docs": 138893, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 179036618 |
| }, |
| "301:880f0b26f94b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/880f0b26f94b.jsonl", |
| "docs": 117845, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 224541537 |
| }, |
| "302:88d8867ceb28.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/88d8867ceb28.jsonl", |
| "docs": 131371, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 169616093 |
| }, |
| "303:88dbb4a1be0d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/88dbb4a1be0d.jsonl", |
| "docs": 128245, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 166238502 |
| }, |
| "304:89e41221c602.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/89e41221c602.jsonl", |
| "docs": 173434, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 198627485 |
| }, |
| "305:89fc4211c762.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/89fc4211c762.jsonl", |
| "docs": 86715, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 178633493 |
| }, |
| "306:8ae3a2dd7039.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/8ae3a2dd7039.jsonl", |
| "docs": 157707, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 205564679 |
| }, |
| "307:8b08fe29fa9d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/8b08fe29fa9d.jsonl", |
| "docs": 140948, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 181474301 |
| }, |
| "308:8b56fdf7cebb.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/8b56fdf7cebb.jsonl", |
| "docs": 18936, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 60617293 |
| }, |
| "309:8b5b5e0bb919.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/8b5b5e0bb919.jsonl", |
| "docs": 200500, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 232224209 |
| }, |
| "310:8b82468442a5.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/8b82468442a5.jsonl", |
| "docs": 149628, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 192623664 |
| }, |
| "311:8bac5bbc9f4d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/8bac5bbc9f4d.jsonl", |
| "docs": 80165, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 216527476 |
| }, |
| "312:8bf6149914c2.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/8bf6149914c2.jsonl", |
| "docs": 149749, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 106895418 |
| }, |
| "313:8c838b1a563d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/8c838b1a563d.jsonl", |
| "docs": 138062, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 178605150 |
| }, |
| "314:8c9cd1992392.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/8c9cd1992392.jsonl", |
| "docs": 97680, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 207322293 |
| }, |
| "315:8dd9f5e4adc4.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/8dd9f5e4adc4.jsonl", |
| "docs": 200696, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 232457335 |
| }, |
| "316:8de0bd6adfb5.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/8de0bd6adfb5.jsonl", |
| "docs": 161611, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 211399314 |
| }, |
| "317:8e269b4feb2a.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/8e269b4feb2a.jsonl", |
| "docs": 158633, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 206737622 |
| }, |
| "318:8f2264d59296.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/8f2264d59296.jsonl", |
| "docs": 160426, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 211636965 |
| }, |
| "319:8f51c2d8f18a.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/8f51c2d8f18a.jsonl", |
| "docs": 151324, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 197241924 |
| }, |
| "320:8f94c8e348bf.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/8f94c8e348bf.jsonl", |
| "docs": 184555, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 210791018 |
| }, |
| "321:8f9e03f3ccb3.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/8f9e03f3ccb3.jsonl", |
| "docs": 46110, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 214736549 |
| }, |
| "322:9041d2957d08.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/9041d2957d08.jsonl", |
| "docs": 32079, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 129814470 |
| }, |
| "323:91df34a1298f.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/91df34a1298f.jsonl", |
| "docs": 74304, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 213418520 |
| }, |
| "324:92075d567959.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/92075d567959.jsonl", |
| "docs": 171948, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 196010779 |
| }, |
| "325:921929d61298.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/921929d61298.jsonl", |
| "docs": 205657, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 240796688 |
| }, |
| "326:9290a55d9d22.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/9290a55d9d22.jsonl", |
| "docs": 101471, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 244361235 |
| }, |
| "327:929e36c0e2d1.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/929e36c0e2d1.jsonl", |
| "docs": 175715, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 228793976 |
| }, |
| "328:9329b0775bb7.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/9329b0775bb7.jsonl", |
| "docs": 144041, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 185614185 |
| }, |
| "329:93fc9ca6c930.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/93fc9ca6c930.jsonl", |
| "docs": 188319, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 216716039 |
| }, |
| "330:943acf0c3ca7.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/943acf0c3ca7.jsonl", |
| "docs": 61697, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 193555708 |
| }, |
| "331:94b1467290a8.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/94b1467290a8.jsonl", |
| "docs": 183313, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 209775908 |
| }, |
| "332:95ea9861c60b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/95ea9861c60b.jsonl", |
| "docs": 208251, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 244899996 |
| }, |
| "333:972502b45314.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/972502b45314.jsonl", |
| "docs": 166746, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 216638092 |
| }, |
| "334:97a0ec0c32ce.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/97a0ec0c32ce.jsonl", |
| "docs": 154723, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 201051849 |
| }, |
| "335:97bf14ea23bf.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/97bf14ea23bf.jsonl", |
| "docs": 210380, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 246502972 |
| }, |
| "336:983e3bc97ec5.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/983e3bc97ec5.jsonl", |
| "docs": 83432, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 196452559 |
| }, |
| "337:98670c065699.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/98670c065699.jsonl", |
| "docs": 167100, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 231277673 |
| }, |
| "338:98e451d30f18.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/98e451d30f18.jsonl", |
| "docs": 142707, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 225491427 |
| }, |
| "339:990931b67f49.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/990931b67f49.jsonl", |
| "docs": 184748, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 212298443 |
| }, |
| "340:9aac45946228.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/9aac45946228.jsonl", |
| "docs": 159621, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 208537813 |
| }, |
| "341:9b2b650628cd.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/9b2b650628cd.jsonl", |
| "docs": 130332, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 169964525 |
| }, |
| "342:9b2dfaa3ae96.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/9b2dfaa3ae96.jsonl", |
| "docs": 177622, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 203410473 |
| }, |
| "343:9c4ac3593445.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/9c4ac3593445.jsonl", |
| "docs": 180354, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 237329342 |
| }, |
| "344:9cc398f97e62.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/9cc398f97e62.jsonl", |
| "docs": 183438, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 208883282 |
| }, |
| "345:9ccd588a4581.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/9ccd588a4581.jsonl", |
| "docs": 158938, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 213107959 |
| }, |
| "346:9d05487c77d1.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/9d05487c77d1.jsonl", |
| "docs": 192869, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 221003040 |
| }, |
| "347:9d31e42cb2e5.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/9d31e42cb2e5.jsonl", |
| "docs": 191911, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 220398671 |
| }, |
| "348:9d776f653dc3.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/9d776f653dc3.jsonl", |
| "docs": 250943, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 188189139 |
| }, |
| "349:9daba54cd98b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/9daba54cd98b.jsonl", |
| "docs": 130395, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 167445629 |
| }, |
| "350:9dad799ed481.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/9dad799ed481.jsonl", |
| "docs": 169258, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 192677916 |
| }, |
| "351:9ddce9e6f311.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/9ddce9e6f311.jsonl", |
| "docs": 173790, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 197322784 |
| }, |
| "352:9dfeff82e265.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/9dfeff82e265.jsonl", |
| "docs": 171121, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 194453444 |
| }, |
| "353:9e0deb5bdde2.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/9e0deb5bdde2.jsonl", |
| "docs": 149778, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 221790817 |
| }, |
| "354:9e1ce79b9eb9.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/9e1ce79b9eb9.jsonl", |
| "docs": 179445, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 204801451 |
| }, |
| "355:9eabb0a06ad3.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/9eabb0a06ad3.jsonl", |
| "docs": 189399, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 217504327 |
| }, |
| "356:9f39dc89df58.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/9f39dc89df58.jsonl", |
| "docs": 197403, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 226755778 |
| }, |
| "357:9f54fec26d64.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/9f54fec26d64.jsonl", |
| "docs": 153291, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 199999072 |
| }, |
| "358:a06fa490b04f.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a06fa490b04f.jsonl", |
| "docs": 135763, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 174283766 |
| }, |
| "359:a0a814c7ff63.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a0a814c7ff63.jsonl", |
| "docs": 179611, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 174513672 |
| }, |
| "360:a0efd7a5d3f4.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a0efd7a5d3f4.jsonl", |
| "docs": 126603, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 163657608 |
| }, |
| "361:a13ca156b4ee.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a13ca156b4ee.jsonl", |
| "docs": 79568, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 212072542 |
| }, |
| "362:a13fa5e1c9c3.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a13fa5e1c9c3.jsonl", |
| "docs": 176268, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 229929142 |
| }, |
| "363:a15982bb884a.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a15982bb884a.jsonl", |
| "docs": 203371, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 235797341 |
| }, |
| "364:a18fbffee1b0.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a18fbffee1b0.jsonl", |
| "docs": 173478, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 226475585 |
| }, |
| "365:a1ec326ea994.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a1ec326ea994.jsonl", |
| "docs": 205069, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 239292067 |
| }, |
| "366:a1feb4bc89b8.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a1feb4bc89b8.jsonl", |
| "docs": 170866, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 222575277 |
| }, |
| "367:a243bb3f402d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a243bb3f402d.jsonl", |
| "docs": 193344, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 224038479 |
| }, |
| "368:a2515f7de5e6.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a2515f7de5e6.jsonl", |
| "docs": 186410, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 214948517 |
| }, |
| "369:a28f81263972.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a28f81263972.jsonl", |
| "docs": 149822, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 107250757 |
| }, |
| "370:a2b90dfac67c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a2b90dfac67c.jsonl", |
| "docs": 180319, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 206971632 |
| }, |
| "371:a36bcff432f4.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a36bcff432f4.jsonl", |
| "docs": 173891, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 123336609 |
| }, |
| "372:a3aa73e0a5fe.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a3aa73e0a5fe.jsonl", |
| "docs": 141685, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 181718815 |
| }, |
| "373:a3b24a046f5c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a3b24a046f5c.jsonl", |
| "docs": 169789, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 191723879 |
| }, |
| "374:a4df06ffae81.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a4df06ffae81.jsonl", |
| "docs": 191646, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 219985990 |
| }, |
| "375:a5524bc8170d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a5524bc8170d.jsonl", |
| "docs": 174574, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 227601262 |
| }, |
| "376:a5cc0a033349.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a5cc0a033349.jsonl", |
| "docs": 160405, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 208000570 |
| }, |
| "377:a6baf45a01a7.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a6baf45a01a7.jsonl", |
| "docs": 167874, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 119260288 |
| }, |
| "378:a76b31a9c877.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a76b31a9c877.jsonl", |
| "docs": 198878, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 229967148 |
| }, |
| "379:a7849855c5d6.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a7849855c5d6.jsonl", |
| "docs": 143959, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 185161132 |
| }, |
| "380:a7d499b872f4.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a7d499b872f4.jsonl", |
| "docs": 164282, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 213950691 |
| }, |
| "381:a8006392c71b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a8006392c71b.jsonl", |
| "docs": 281163, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 205331261 |
| }, |
| "382:a9242a43257e.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a9242a43257e.jsonl", |
| "docs": 116444, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 80295877 |
| }, |
| "383:a94b1151300b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a94b1151300b.jsonl", |
| "docs": 183127, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 211203768 |
| }, |
| "384:a97d0c404655.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a97d0c404655.jsonl", |
| "docs": 210385, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 245917636 |
| }, |
| "385:a9ff2f60acfa.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/a9ff2f60acfa.jsonl", |
| "docs": 148656, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 194083483 |
| }, |
| "386:aa182271d5c4.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/aa182271d5c4.jsonl", |
| "docs": 170477, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 222940989 |
| }, |
| "387:ab193417df73.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ab193417df73.jsonl", |
| "docs": 180366, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 128500874 |
| }, |
| "388:ab69590cbcbf.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ab69590cbcbf.jsonl", |
| "docs": 200980, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 231847280 |
| }, |
| "389:abd4ec744a1b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/abd4ec744a1b.jsonl", |
| "docs": 126182, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 210873079 |
| }, |
| "390:abf9da4e0b35.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/abf9da4e0b35.jsonl", |
| "docs": 126999, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 164092834 |
| }, |
| "391:ac2dc1bb9953.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ac2dc1bb9953.jsonl", |
| "docs": 200878, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 233009350 |
| }, |
| "392:acb56dae19d3.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/acb56dae19d3.jsonl", |
| "docs": 179282, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 205426276 |
| }, |
| "393:ad1ef24a508b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ad1ef24a508b.jsonl", |
| "docs": 178877, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 239995750 |
| }, |
| "394:ad4ceb680104.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ad4ceb680104.jsonl", |
| "docs": 181431, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 208314814 |
| }, |
| "395:ae868dd3bbb9.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ae868dd3bbb9.jsonl", |
| "docs": 155025, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 202001743 |
| }, |
| "396:af5d994c475b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/af5d994c475b.jsonl", |
| "docs": 97872, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 180074931 |
| }, |
| "397:afc13a714f52.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/afc13a714f52.jsonl", |
| "docs": 154365, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 200317222 |
| }, |
| "398:b09436bdb45f.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b09436bdb45f.jsonl", |
| "docs": 113546, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 194142290 |
| }, |
| "399:b09ae7afdd07.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b09ae7afdd07.jsonl", |
| "docs": 171449, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 225419448 |
| }, |
| "400:b0c4a12c3944.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b0c4a12c3944.jsonl", |
| "docs": 167233, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 244886629 |
| }, |
| "401:b16c1b3db86d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b16c1b3db86d.jsonl", |
| "docs": 173989, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 199020571 |
| }, |
| "402:b2bfc4591a86.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b2bfc4591a86.jsonl", |
| "docs": 169059, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 192346428 |
| }, |
| "403:b320f5e1b383.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b320f5e1b383.jsonl", |
| "docs": 152302, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 197413016 |
| }, |
| "404:b44c33b2614e.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b44c33b2614e.jsonl", |
| "docs": 167491, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 190896479 |
| }, |
| "405:b581ff8463c0.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b581ff8463c0.jsonl", |
| "docs": 206990, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 240649845 |
| }, |
| "406:b58bc74bb0aa.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b58bc74bb0aa.jsonl", |
| "docs": 156575, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 204482823 |
| }, |
| "407:b5fd58091529.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b5fd58091529.jsonl", |
| "docs": 147798, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 191152069 |
| }, |
| "408:b60a8431d5b5.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b60a8431d5b5.jsonl", |
| "docs": 143176, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 183478191 |
| }, |
| "409:b63674d50c90.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b63674d50c90.jsonl", |
| "docs": 189115, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 218018723 |
| }, |
| "410:b6d97e171341.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b6d97e171341.jsonl", |
| "docs": 208770, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 243721825 |
| }, |
| "411:b7471c3ca7eb.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b7471c3ca7eb.jsonl", |
| "docs": 158226, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 206134538 |
| }, |
| "412:b796180484e4.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b796180484e4.jsonl", |
| "docs": 150380, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 193840853 |
| }, |
| "413:b7bdef3bec9a.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b7bdef3bec9a.jsonl", |
| "docs": 133520, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 171862538 |
| }, |
| "414:b7fec9a98e87.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b7fec9a98e87.jsonl", |
| "docs": 76257, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 217209035 |
| }, |
| "415:b80455f2d9b0.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b80455f2d9b0.jsonl", |
| "docs": 49679, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 192388459 |
| }, |
| "416:b8247449604b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b8247449604b.jsonl", |
| "docs": 209642, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 244249397 |
| }, |
| "417:b831f07173a8.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b831f07173a8.jsonl", |
| "docs": 74182, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 201831870 |
| }, |
| "418:b88d44d5eb27.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b88d44d5eb27.jsonl", |
| "docs": 153345, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 199646098 |
| }, |
| "419:b90f332737ab.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b90f332737ab.jsonl", |
| "docs": 157488, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 205647681 |
| }, |
| "420:b91cbaf70e2f.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b91cbaf70e2f.jsonl", |
| "docs": 91651, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 223430714 |
| }, |
| "421:b93f621ab8eb.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b93f621ab8eb.jsonl", |
| "docs": 195860, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 225403393 |
| }, |
| "422:b9e04cdd5411.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/b9e04cdd5411.jsonl", |
| "docs": 173800, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 198907312 |
| }, |
| "423:ba05d788d28a.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ba05d788d28a.jsonl", |
| "docs": 144978, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 188091185 |
| }, |
| "424:ba50dd2e2955.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ba50dd2e2955.jsonl", |
| "docs": 154093, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 200860239 |
| }, |
| "425:ba6a14e6eaad.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ba6a14e6eaad.jsonl", |
| "docs": 122198, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 227954925 |
| }, |
| "426:bb74b2920805.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/bb74b2920805.jsonl", |
| "docs": 261271, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 163689065 |
| }, |
| "427:bc57f20030e3.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/bc57f20030e3.jsonl", |
| "docs": 150458, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 202110020 |
| }, |
| "428:bc5952c17525.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/bc5952c17525.jsonl", |
| "docs": 168548, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 190857061 |
| }, |
| "429:bcd7d40ea7a7.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/bcd7d40ea7a7.jsonl", |
| "docs": 182144, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 208894644 |
| }, |
| "430:bd25e8e41ced.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/bd25e8e41ced.jsonl", |
| "docs": 164751, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 214343377 |
| }, |
| "431:bd38e95282e7.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/bd38e95282e7.jsonl", |
| "docs": 209792, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 245638504 |
| }, |
| "432:bd686d1d127a.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/bd686d1d127a.jsonl", |
| "docs": 198661, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 229427777 |
| }, |
| "433:bdb41fecb82c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/bdb41fecb82c.jsonl", |
| "docs": 134595, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 174643833 |
| }, |
| "434:be760ea0e43f.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/be760ea0e43f.jsonl", |
| "docs": 150628, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 195718529 |
| }, |
| "435:beb1e9226777.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/beb1e9226777.jsonl", |
| "docs": 196590, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 226481162 |
| }, |
| "436:bfb10681059d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/bfb10681059d.jsonl", |
| "docs": 101801, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 187388391 |
| }, |
| "437:bfe5dbbbe5ba.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/bfe5dbbbe5ba.jsonl", |
| "docs": 141982, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 183013636 |
| }, |
| "438:bfe61613786c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/bfe61613786c.jsonl", |
| "docs": 168029, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 190392937 |
| }, |
| "439:c000eb754a80.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/c000eb754a80.jsonl", |
| "docs": 184031, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 211295486 |
| }, |
| "440:c14fb3a33163.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/c14fb3a33163.jsonl", |
| "docs": 153126, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 198278455 |
| }, |
| "441:c191e969c1b8.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/c191e969c1b8.jsonl", |
| "docs": 186442, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 213337808 |
| }, |
| "442:c2f327a7532f.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/c2f327a7532f.jsonl", |
| "docs": 176111, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 200017863 |
| }, |
| "443:c374507d9ad6.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/c374507d9ad6.jsonl", |
| "docs": 131010, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 173070064 |
| }, |
| "444:c458fdc2b0a7.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/c458fdc2b0a7.jsonl", |
| "docs": 171932, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 195671486 |
| }, |
| "445:c63b74b07437.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/c63b74b07437.jsonl", |
| "docs": 187153, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 214623591 |
| }, |
| "446:c6b909d6a5f0.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/c6b909d6a5f0.jsonl", |
| "docs": 203177, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 236827881 |
| }, |
| "447:c717c571c396.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/c717c571c396.jsonl", |
| "docs": 209616, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 248246507 |
| }, |
| "448:c785f4106354.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/c785f4106354.jsonl", |
| "docs": 167991, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 220487894 |
| }, |
| "449:c7a89e3fe5d4.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/c7a89e3fe5d4.jsonl", |
| "docs": 172582, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 225385291 |
| }, |
| "450:c84a16a7db82.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/c84a16a7db82.jsonl", |
| "docs": 179625, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 204805933 |
| }, |
| "451:c85cc1511945.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/c85cc1511945.jsonl", |
| "docs": 181804, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 207067137 |
| }, |
| "452:c8799b1cc101.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/c8799b1cc101.jsonl", |
| "docs": 161351, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 210255268 |
| }, |
| "453:c8d9367f1b4f.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/c8d9367f1b4f.jsonl", |
| "docs": 166967, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 218216389 |
| }, |
| "454:c8fbfd02f328.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/c8fbfd02f328.jsonl", |
| "docs": 177590, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 232764267 |
| }, |
| "455:c94d690d3079.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/c94d690d3079.jsonl", |
| "docs": 195383, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 225882798 |
| }, |
| "456:c9894a464c20.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/c9894a464c20.jsonl", |
| "docs": 204986, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 237548204 |
| }, |
| "457:c9a1626c5e2b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/c9a1626c5e2b.jsonl", |
| "docs": 202882, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 233283162 |
| }, |
| "458:ca4b62476a7d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ca4b62476a7d.jsonl", |
| "docs": 202424, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 234484126 |
| }, |
| "459:ca7a4f19a8ff.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ca7a4f19a8ff.jsonl", |
| "docs": 130003, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 169029644 |
| }, |
| "460:cb156a7eca77.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/cb156a7eca77.jsonl", |
| "docs": 151052, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 195805257 |
| }, |
| "461:cb2803d0ffdc.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/cb2803d0ffdc.jsonl", |
| "docs": 125841, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 162177351 |
| }, |
| "462:cbec0c649a91.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/cbec0c649a91.jsonl", |
| "docs": 206610, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 241454092 |
| }, |
| "463:cbf5a86d375c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/cbf5a86d375c.jsonl", |
| "docs": 142508, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 200109846 |
| }, |
| "464:ccb50eecd359.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ccb50eecd359.jsonl", |
| "docs": 183953, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 210775535 |
| }, |
| "465:cd347cc4595f.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/cd347cc4595f.jsonl", |
| "docs": 191602, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 220967783 |
| }, |
| "466:cd7e526820f1.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/cd7e526820f1.jsonl", |
| "docs": 140460, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 182624218 |
| }, |
| "467:cdbce18ecf8c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/cdbce18ecf8c.jsonl", |
| "docs": 134531, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 174393021 |
| }, |
| "468:cf3b337eff85.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/cf3b337eff85.jsonl", |
| "docs": 148562, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 104293424 |
| }, |
| "469:cf6ce3eeb4ca.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/cf6ce3eeb4ca.jsonl", |
| "docs": 170309, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 193441911 |
| }, |
| "470:cf83b8a6c018.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/cf83b8a6c018.jsonl", |
| "docs": 153148, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 199449536 |
| }, |
| "471:d163f82d1c2d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/d163f82d1c2d.jsonl", |
| "docs": 59659, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 170563902 |
| }, |
| "472:d27b518ebf4e.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/d27b518ebf4e.jsonl", |
| "docs": 134365, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 173356879 |
| }, |
| "473:d2818e297c3e.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/d2818e297c3e.jsonl", |
| "docs": 159637, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 208057998 |
| }, |
| "474:d350d2a3eb50.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/d350d2a3eb50.jsonl", |
| "docs": 75597, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 204524063 |
| }, |
| "475:d445c14152a1.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/d445c14152a1.jsonl", |
| "docs": 147782, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 192187602 |
| }, |
| "476:d58b2953c1fc.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/d58b2953c1fc.jsonl", |
| "docs": 172263, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 196244129 |
| }, |
| "477:d5e07c15af39.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/d5e07c15af39.jsonl", |
| "docs": 179475, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 205001860 |
| }, |
| "478:d68f09142ecb.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/d68f09142ecb.jsonl", |
| "docs": 149969, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 194792868 |
| }, |
| "479:d8247ca6ac2e.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/d8247ca6ac2e.jsonl", |
| "docs": 152768, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 198109771 |
| }, |
| "480:da1590656101.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/da1590656101.jsonl", |
| "docs": 131716, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 170259669 |
| }, |
| "481:da1ced34e517.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/da1ced34e517.jsonl", |
| "docs": 186012, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 132881847 |
| }, |
| "482:da387c2ace5c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/da387c2ace5c.jsonl", |
| "docs": 188846, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 218033087 |
| }, |
| "483:db69e8cbf10c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/db69e8cbf10c.jsonl", |
| "docs": 184775, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 212181929 |
| }, |
| "484:db6a1e461cef.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/db6a1e461cef.jsonl", |
| "docs": 167485, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 188483544 |
| }, |
| "485:dc0dbc945d57.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/dc0dbc945d57.jsonl", |
| "docs": 73186, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 196170636 |
| }, |
| "486:dc944e64cd23.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/dc944e64cd23.jsonl", |
| "docs": 157513, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 113218723 |
| }, |
| "487:dca4277ea333.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/dca4277ea333.jsonl", |
| "docs": 191562, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 219744013 |
| }, |
| "488:dcfb17685095.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/dcfb17685095.jsonl", |
| "docs": 174800, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 229016795 |
| }, |
| "489:dd087bca6df7.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/dd087bca6df7.jsonl", |
| "docs": 177910, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 204232918 |
| }, |
| "490:dd5fbf3a8349.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/dd5fbf3a8349.jsonl", |
| "docs": 135701, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 169619231 |
| }, |
| "491:dd77604eaae6.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/dd77604eaae6.jsonl", |
| "docs": 172473, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 226135512 |
| }, |
| "492:dde001bdf8c4.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/dde001bdf8c4.jsonl", |
| "docs": 148519, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 106027436 |
| }, |
| "493:ddfd91684fa0.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ddfd91684fa0.jsonl", |
| "docs": 181731, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 209924512 |
| }, |
| "494:de4924089a56.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/de4924089a56.jsonl", |
| "docs": 173220, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 196196586 |
| }, |
| "495:de5c4b7572f4.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/de5c4b7572f4.jsonl", |
| "docs": 98846, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 241930977 |
| }, |
| "496:dea0c7d2038c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/dea0c7d2038c.jsonl", |
| "docs": 176352, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 231922748 |
| }, |
| "497:df2c6192c3a6.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/df2c6192c3a6.jsonl", |
| "docs": 152474, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 198871536 |
| }, |
| "498:e046ed901283.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e046ed901283.jsonl", |
| "docs": 196075, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 227140135 |
| }, |
| "499:e0914a744571.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e0914a744571.jsonl", |
| "docs": 60357, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 156231919 |
| }, |
| "500:e14d25f28d50.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e14d25f28d50.jsonl", |
| "docs": 122465, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 173203564 |
| }, |
| "501:e17c1a935c91.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e17c1a935c91.jsonl", |
| "docs": 150336, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 195916759 |
| }, |
| "502:e1b58a4ba3bc.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e1b58a4ba3bc.jsonl", |
| "docs": 64239, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 177864586 |
| }, |
| "503:e349b99206c0.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e349b99206c0.jsonl", |
| "docs": 131798, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 171178881 |
| }, |
| "504:e3a004791c98.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e3a004791c98.jsonl", |
| "docs": 13476, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 229836415 |
| }, |
| "505:e3feac35b4a8.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e3feac35b4a8.jsonl", |
| "docs": 156828, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 203894435 |
| }, |
| "506:e40801ad3e6f.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e40801ad3e6f.jsonl", |
| "docs": 185514, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 234086954 |
| }, |
| "507:e4b801f384d1.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e4b801f384d1.jsonl", |
| "docs": 63274, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 178624498 |
| }, |
| "508:e546bc6cdd9e.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e546bc6cdd9e.jsonl", |
| "docs": 205504, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 239075164 |
| }, |
| "509:e610670781f6.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e610670781f6.jsonl", |
| "docs": 148170, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 192173817 |
| }, |
| "510:e61fcdcc2ea1.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e61fcdcc2ea1.jsonl", |
| "docs": 148564, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 106166185 |
| }, |
| "511:e696245c2542.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e696245c2542.jsonl", |
| "docs": 140916, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 182740187 |
| }, |
| "512:e696d53329e8.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e696d53329e8.jsonl", |
| "docs": 204225, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 239169316 |
| }, |
| "513:e6cf94797185.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e6cf94797185.jsonl", |
| "docs": 177602, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 203211542 |
| }, |
| "514:e6f765306809.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e6f765306809.jsonl", |
| "docs": 197626, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 227192402 |
| }, |
| "515:e7075eccf7f9.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e7075eccf7f9.jsonl", |
| "docs": 207858, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 244859823 |
| }, |
| "516:e7cdf70d87bc.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e7cdf70d87bc.jsonl", |
| "docs": 191159, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 219931572 |
| }, |
| "517:e85ee614a01d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e85ee614a01d.jsonl", |
| "docs": 170487, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 193437832 |
| }, |
| "518:e8dd95d96887.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e8dd95d96887.jsonl", |
| "docs": 193950, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 222686259 |
| }, |
| "519:e8e3822f18bc.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e8e3822f18bc.jsonl", |
| "docs": 125527, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 163370375 |
| }, |
| "520:e8e565bfa5e1.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e8e565bfa5e1.jsonl", |
| "docs": 197309, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 227973505 |
| }, |
| "521:e8ea01a5f44f.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e8ea01a5f44f.jsonl", |
| "docs": 57775, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 183109442 |
| }, |
| "522:e8f74033510f.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e8f74033510f.jsonl", |
| "docs": 200739, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 230709398 |
| }, |
| "523:e93350bc7e0b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/e93350bc7e0b.jsonl", |
| "docs": 172176, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 224273265 |
| }, |
| "524:ea08072a50be.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ea08072a50be.jsonl", |
| "docs": 150649, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 196468441 |
| }, |
| "525:ea2996f52648.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ea2996f52648.jsonl", |
| "docs": 136506, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 176385124 |
| }, |
| "526:ea52dc030440.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ea52dc030440.jsonl", |
| "docs": 67123, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 186880087 |
| }, |
| "527:ea934ab5a8e0.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ea934ab5a8e0.jsonl", |
| "docs": 60972, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 149960281 |
| }, |
| "528:eb1ca12e28c3.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/eb1ca12e28c3.jsonl", |
| "docs": 146103, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 189427456 |
| }, |
| "529:eba8b7a95b54.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/eba8b7a95b54.jsonl", |
| "docs": 153595, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 110129590 |
| }, |
| "530:ebb497bf9f08.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ebb497bf9f08.jsonl", |
| "docs": 137702, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 176733794 |
| }, |
| "531:ec2fb2ab06b5.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ec2fb2ab06b5.jsonl", |
| "docs": 200977, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 145196469 |
| }, |
| "532:ec7c199fe5dd.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ec7c199fe5dd.jsonl", |
| "docs": 162403, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 210830314 |
| }, |
| "533:ec82b7bd848e.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ec82b7bd848e.jsonl", |
| "docs": 190994, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 219272387 |
| }, |
| "534:ec9a1a75e8fc.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ec9a1a75e8fc.jsonl", |
| "docs": 180703, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 236345327 |
| }, |
| "535:ecc7e2dee1de.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ecc7e2dee1de.jsonl", |
| "docs": 87855, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 178811039 |
| }, |
| "536:ed1e8872a379.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ed1e8872a379.jsonl", |
| "docs": 174621, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 199116892 |
| }, |
| "537:edd449bf54c2.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/edd449bf54c2.jsonl", |
| "docs": 191678, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 221428102 |
| }, |
| "538:ee3bf6475966.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ee3bf6475966.jsonl", |
| "docs": 155704, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 202900933 |
| }, |
| "539:ee7ba49b7b47.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ee7ba49b7b47.jsonl", |
| "docs": 195006, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 224371190 |
| }, |
| "540:ee9203ad0da2.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ee9203ad0da2.jsonl", |
| "docs": 148999, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 106618354 |
| }, |
| "541:eedeca57f50d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/eedeca57f50d.jsonl", |
| "docs": 203278, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 235340488 |
| }, |
| "542:eff45e19c253.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/eff45e19c253.jsonl", |
| "docs": 136666, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 176430653 |
| }, |
| "543:f1060ef7c7f6.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/f1060ef7c7f6.jsonl", |
| "docs": 138881, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 178776885 |
| }, |
| "544:f129124edbf9.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/f129124edbf9.jsonl", |
| "docs": 207878, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 243178075 |
| }, |
| "545:f141d2f857ec.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/f141d2f857ec.jsonl", |
| "docs": 176676, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 201881280 |
| }, |
| "546:f198f609c1fc.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/f198f609c1fc.jsonl", |
| "docs": 162097, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 114784714 |
| }, |
| "547:f244b14a6647.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/f244b14a6647.jsonl", |
| "docs": 76594, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 147533083 |
| }, |
| "548:f462cd086e7a.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/f462cd086e7a.jsonl", |
| "docs": 187262, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 216500460 |
| }, |
| "549:f510578ad8cf.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/f510578ad8cf.jsonl", |
| "docs": 92351, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 179773026 |
| }, |
| "550:f58b52a38ac8.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/f58b52a38ac8.jsonl", |
| "docs": 166534, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 188347227 |
| }, |
| "551:f5b0841ed99c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/f5b0841ed99c.jsonl", |
| "docs": 162935, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 213106179 |
| }, |
| "552:f5d5efe68a8d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/f5d5efe68a8d.jsonl", |
| "docs": 154269, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 110721588 |
| }, |
| "553:f5f4b8441d17.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/f5f4b8441d17.jsonl", |
| "docs": 141175, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 182379439 |
| }, |
| "554:f61b20357975.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/f61b20357975.jsonl", |
| "docs": 193300, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 223672979 |
| }, |
| "555:f62e3e5ab154.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/f62e3e5ab154.jsonl", |
| "docs": 135798, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 174949057 |
| }, |
| "556:f72820db3532.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/f72820db3532.jsonl", |
| "docs": 128742, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 167509275 |
| }, |
| "557:f7ba5e6b8b28.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/f7ba5e6b8b28.jsonl", |
| "docs": 193107, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 222928477 |
| }, |
| "558:f866f8bd413c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/f866f8bd413c.jsonl", |
| "docs": 204639, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 238248025 |
| }, |
| "559:f9be195a5c19.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/f9be195a5c19.jsonl", |
| "docs": 96019, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 186091602 |
| }, |
| "560:fb167729d868.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/fb167729d868.jsonl", |
| "docs": 97697, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 229804008 |
| }, |
| "561:fb2a7c863531.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/fb2a7c863531.jsonl", |
| "docs": 185114, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 212462355 |
| }, |
| "562:fb66b1b4ab6b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/fb66b1b4ab6b.jsonl", |
| "docs": 196620, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 227981033 |
| }, |
| "563:fbce5024a5b7.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/fbce5024a5b7.jsonl", |
| "docs": 158407, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 111719287 |
| }, |
| "564:fbfc7fbdb2c8.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/fbfc7fbdb2c8.jsonl", |
| "docs": 157079, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 214354207 |
| }, |
| "565:fcc777fb1207.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/fcc777fb1207.jsonl", |
| "docs": 170587, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 194735426 |
| }, |
| "566:fcd0bfb1438d.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/fcd0bfb1438d.jsonl", |
| "docs": 75478, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 177480818 |
| }, |
| "567:fcee8014c85f.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/fcee8014c85f.jsonl", |
| "docs": 80559, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 116112638 |
| }, |
| "568:fcff23c6789a.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/fcff23c6789a.jsonl", |
| "docs": 167443, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 224888756 |
| }, |
| "569:fd42531f898c.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/fd42531f898c.jsonl", |
| "docs": 160290, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 209102870 |
| }, |
| "570:fd48964754de.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/fd48964754de.jsonl", |
| "docs": 155207, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 200318848 |
| }, |
| "571:fd706c29708b.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/fd706c29708b.jsonl", |
| "docs": 172878, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 225541478 |
| }, |
| "572:fde66eafa837.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/fde66eafa837.jsonl", |
| "docs": 128018, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 166400304 |
| }, |
| "573:fedd84509b5e.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/fedd84509b5e.jsonl", |
| "docs": 256911, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 173774501 |
| }, |
| "574:ff19a1aca0bc.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ff19a1aca0bc.jsonl", |
| "docs": 127411, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 165399961 |
| }, |
| "575:ff4afd5be782.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ff4afd5be782.jsonl", |
| "docs": 144598, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 224118450 |
| }, |
| "576:ffa90c21d6e5.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ffa90c21d6e5.jsonl", |
| "docs": 150245, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 201710215 |
| }, |
| "577:ffe648a90133.jsonl": { |
| "path": "/var/account/yhb2/Nemotron/data_curator/03_deduped/ffe648a90133.jsonl", |
| "docs": 205209, |
| "duplicates": 0, |
| "short": 0, |
| "records_no_text": 0, |
| "tokens": 239643827 |
| } |
| } |
| } |
