Upload OpenAster-1 20B pretraining data

README.md

@@ -1,3 +1,42 @@
 ---
 license: mit
+language:
+- en
+- zh
+task_categories:
+- text-generation
+tags:
+- openaster
+- pretraining
+- llm
 ---
+
+# OpenAster-1 20B Pretraining Data
+
+This dataset contains the final merged 20B-token scratch-pretraining mixture used for OpenAster-1. Files are released as line-preserving `jsonl.zst` shards. Each record contains `text`, `source`, and `language`.
+
+## Sources
+
+- **DCLM Edu**: 4B target tokens, English educational web.
+- **OpenCSG FineWeb Edu 4/5**: 5B target tokens, Chinese educational web.
+- **FineMath 4+**: 2B target tokens, English math corpus.
+- **FineWeb Edu EN**: 4B target tokens, English educational web.
+- **Chinese FineWeb Edu**: 4B target tokens, Chinese educational web.
+- **Pile Deduplicated**: 1B target tokens, general/reference replay.
+
+See `metadata/mix_manifest.json` for document/token counts and source metadata.
+
+# OpenAster-1 20B 预训练数据
+
+该数据集为 OpenAster-1 从零预训练阶段使用的最终 20B token 混合数据。发布格式为按行切分的 `jsonl.zst` 分片。每条记录包含 `text`、`source`、`language` 三个字段。
+
+## 数据来源
+
+- **DCLM Edu**：目标 4B tokens，英文教育类网页。
+- **OpenCSG FineWeb Edu 4/5**：目标 5B tokens，中文教育类网页。
+- **FineMath 4+**：目标 2B tokens，英文数学语料。
+- **FineWeb Edu EN**：目标 4B tokens，英文教育类网页。
+- **Chinese FineWeb Edu**：目标 4B tokens，中文教育类网页。
+- **Pile Deduplicated**：目标 1B tokens，通用/知识 replay 数据。
+
+详细 token 和文档统计见 `metadata/mix_manifest.json`。

SHA256SUMS

@@ -0,0 +1,26 @@
+6647f59e8e2a16a46342fc017b22e78dedf5b09145d6e3c3e4613da870f3b807  data/train-00000.jsonl.zst
+8279084fba21ffce211d7044e93ec7e5063c457217a3c0ac03b66cb0d9b13a3f  data/train-00001.jsonl.zst
+dbec94b34c61d3e99f7d2ea6296361ae4f953c7b918ff8ccab4879e39bf161ff  data/train-00002.jsonl.zst
+ec05d182764c54c219ce2c2bf024f28e1d1d27357810003ed972101160b6b9b4  data/train-00003.jsonl.zst
+e42d1cfc9c966a8db6a36b0d458dc785cb1a1ced8b453aa649290fd819ef8b63  data/train-00004.jsonl.zst
+8d68f48173855085d40cf461c97d6f891f21e83b243dd05c8299d6063d78974a  data/train-00005.jsonl.zst
+e5afccaee05fb8d90d7cfeb378f9a501fa4903e321b31e41e9ca4f927208453c  data/train-00006.jsonl.zst
+4900b3c8b63d24d577414379c4e8b3e90c5c22e78dd19917920bbc8088426c30  data/train-00007.jsonl.zst
+107fafc5dd62df961bc64c5e1ee2b6bc6b8082e98d7817e7325041ebb7e16fa5  data/train-00008.jsonl.zst
+7695a2b275fab3d7cf8903a083455cbb0ea5140d29cce8724b3d288bcc3113a2  data/train-00009.jsonl.zst
+d0841fd81fbc3e46efd8b77c4a33642ca16b65bb10c22d37a0f2ba25afc15d51  data/train-00010.jsonl.zst
+3ed97d8f9e74a1e56ecd881bb2449fb3def5bdb30b490220f6a7d489f926a4c8  data/train-00011.jsonl.zst
+ccace20008c2268f2a4e24fe444dfe963dabad6115939e195aaa40b900b4cf72  data/train-00012.jsonl.zst
+ff47203a2916d447bd3746dcba1f72add6ce7c0b5595c3b9098e61acff0655bf  data/train-00013.jsonl.zst
+53316fa993e3158bb8ec6b242472fb801b767d8354bfe356b9d30442c7a895ac  data/train-00014.jsonl.zst
+0baa9cedc70e5427511827d1fc38f5c85fe9b50a9db1134d8a45765e595d55b1  data/train-00015.jsonl.zst
+d30e8f898ee2ca87779e2f4c62f2a9c0c4c1a53d919df79173aeba4e198669f0  data/train-00016.jsonl.zst
+cb2a22d9967576860cf6517208df2426d0cb972995ff5cfd96ff639fa6b57bef  data/train-00017.jsonl.zst
+05b32d9d7b47d09d72070162caabbaf39ca40ced16304d2864607307f26be410  data/train-00018.jsonl.zst
+c7e8aa7f46e20a780b7f72c459ed81d60c2d704c2fd663049b2cefec9bcdda41  data/train-00019.jsonl.zst
+484610f11da8713727e81ac9ae6f239761c4dbaf35d56e77ebc9948cc0a97556  data/train-00020.jsonl.zst
+2f101761ec4a9396db5b79fa558638f03362143b2d148eb2cc356e2a4b04f10e  data/train-00021.jsonl.zst
+586235f2d88d8e4023b2d3b1ae86546e72ac6eb7edd043f19b78d84393531c78  metadata/mix_manifest.json
+c9c7f8149c7770bca941ba7ea088962238e8488e7e2098e88aea3b7c36267fb8  metadata/shards.json
+793273d1528c398e5a0b39be8858b5a62da99ddbe63259a5d30a08200d4cae23  metadata/source_summary.json
+bf697a7b0b6cfc9398a8d6251914b05029350270919372432b29ceadc129027f  README.md

data/train-00000.jsonl.zst

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6647f59e8e2a16a46342fc017b22e78dedf5b09145d6e3c3e4613da870f3b807
+size 1854030764

data/train-00001.jsonl.zst

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8279084fba21ffce211d7044e93ec7e5063c457217a3c0ac03b66cb0d9b13a3f
+size 1749350202

data/train-00002.jsonl.zst

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dbec94b34c61d3e99f7d2ea6296361ae4f953c7b918ff8ccab4879e39bf161ff
+size 1778255290

data/train-00003.jsonl.zst

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ec05d182764c54c219ce2c2bf024f28e1d1d27357810003ed972101160b6b9b4
+size 1810440550

data/train-00004.jsonl.zst

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e42d1cfc9c966a8db6a36b0d458dc785cb1a1ced8b453aa649290fd819ef8b63
+size 1660975662

data/train-00005.jsonl.zst

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8d68f48173855085d40cf461c97d6f891f21e83b243dd05c8299d6063d78974a
+size 1586082175

data/train-00006.jsonl.zst

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e5afccaee05fb8d90d7cfeb378f9a501fa4903e321b31e41e9ca4f927208453c
+size 1584310369

data/train-00007.jsonl.zst

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4900b3c8b63d24d577414379c4e8b3e90c5c22e78dd19917920bbc8088426c30
+size 1589029441

data/train-00008.jsonl.zst

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:107fafc5dd62df961bc64c5e1ee2b6bc6b8082e98d7817e7325041ebb7e16fa5
+size 1497688531

data/train-00009.jsonl.zst

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7695a2b275fab3d7cf8903a083455cbb0ea5140d29cce8724b3d288bcc3113a2
+size 1374347280

data/train-00010.jsonl.zst

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d0841fd81fbc3e46efd8b77c4a33642ca16b65bb10c22d37a0f2ba25afc15d51
+size 1536086889

data/train-00011.jsonl.zst

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3ed97d8f9e74a1e56ecd881bb2449fb3def5bdb30b490220f6a7d489f926a4c8
+size 1574422663

data/train-00012.jsonl.zst

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ccace20008c2268f2a4e24fe444dfe963dabad6115939e195aaa40b900b4cf72
+size 1575238814

data/train-00013.jsonl.zst

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ff47203a2916d447bd3746dcba1f72add6ce7c0b5595c3b9098e61acff0655bf
+size 1575765784

data/train-00014.jsonl.zst

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:53316fa993e3158bb8ec6b242472fb801b767d8354bfe356b9d30442c7a895ac
+size 1620796375

data/train-00015.jsonl.zst

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0baa9cedc70e5427511827d1fc38f5c85fe9b50a9db1134d8a45765e595d55b1
+size 1659984941

data/train-00016.jsonl.zst

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d30e8f898ee2ca87779e2f4c62f2a9c0c4c1a53d919df79173aeba4e198669f0
+size 1661054210

data/train-00017.jsonl.zst

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:cb2a22d9967576860cf6517208df2426d0cb972995ff5cfd96ff639fa6b57bef
+size 1661671657

data/train-00018.jsonl.zst

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:05b32d9d7b47d09d72070162caabbaf39ca40ced16304d2864607307f26be410
+size 1569915109

data/train-00019.jsonl.zst

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c7e8aa7f46e20a780b7f72c459ed81d60c2d704c2fd663049b2cefec9bcdda41
+size 1551014999

data/train-00020.jsonl.zst

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:484610f11da8713727e81ac9ae6f239761c4dbaf35d56e77ebc9948cc0a97556
+size 1476190707

data/train-00021.jsonl.zst

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2f101761ec4a9396db5b79fa558638f03362143b2d148eb2cc356e2a4b04f10e
+size 440532604

metadata/mix_manifest.json

@@ -0,0 +1,78 @@
+{
+  "name": "OpenAster-1 20B pretraining mix",
+  "format": "JSONL compressed with zstd; each record has text, source, and language fields.",
+  "target_total_tokens": 20000000000,
+  "completed_tokens": 20000429318,
+  "completed_docs": 15495122,
+  "tokenizer_reference": "Qwen3.5-4B tokenizer used for token counting during preprocessing.",
+  "sources": [
+    {
+      "source": "dclm_edu",
+      "upstream_dataset": "HuggingFaceTB/dclm-edu",
+      "target_tokens": 4000000000,
+      "written_tokens": 4000000514,
+      "written_docs": 2954893,
+      "written_chars": 17819801584,
+      "language": "en",
+      "data_type": "English educational web",
+      "status": "completed"
+    },
+    {
+      "source": "opencsg_fineweb_edu_45",
+      "upstream_dataset": "opencsg/Fineweb-Edu-Chinese-V2.1",
+      "target_tokens": 5000000000,
+      "written_tokens": 5000007374,
+      "written_docs": 2873552,
+      "written_chars": 8889749082,
+      "language": "zh",
+      "data_type": "Chinese educational web",
+      "status": "completed"
+    },
+    {
+      "source": "finemath_4plus",
+      "upstream_dataset": "HuggingFaceTB/finemath",
+      "target_tokens": 2000000000,
+      "written_tokens": 2000411098,
+      "written_docs": 1305494,
+      "written_chars": 6554829219,
+      "language": "en",
+      "data_type": "English math corpus",
+      "status": "completed"
+    },
+    {
+      "source": "fineweb_edu_en",
+      "upstream_dataset": "AI-ModelScope/fineweb-edu",
+      "target_tokens": 4000000000,
+      "written_tokens": 4000003289,
+      "written_docs": 3742852,
+      "written_chars": 18050007746,
+      "language": "en",
+      "data_type": "English educational web",
+      "status": "completed"
+    },
+    {
+      "source": "chinese_fineweb_edu",
+      "upstream_dataset": "AI-ModelScope/chinese-fineweb-edu",
+      "target_tokens": 4000000000,
+      "written_tokens": 4000000142,
+      "written_docs": 3900669,
+      "written_chars": 6964548330,
+      "language": "zh",
+      "data_type": "Chinese educational web",
+      "status": "completed"
+    },
+    {
+      "source": "pile_deduplicated",
+      "upstream_dataset": "EleutherAI/the_pile_deduplicated via ModelScope mirror",
+      "target_tokens": 1000000000,
+      "written_tokens": 1000006901,
+      "written_docs": 717662,
+      "written_chars": 4095830408,
+      "language": "en",
+      "data_type": "general/reference replay",
+      "status": "completed"
+    }
+  ],
+  "created_from": "final merged raw/train.jsonl from the OpenAster-1 20B scratch pretraining pipeline",
+  "prepared_at_unix": 1782316670
+}

metadata/shards.json

@@ -0,0 +1,93 @@
+{
+  "num_shards": 22,
+  "shards": [
+    {
+      "path": "data/train-00000.jsonl.zst",
+      "bytes": 1854030764
+    },
+    {
+      "path": "data/train-00001.jsonl.zst",
+      "bytes": 1749350202
+    },
+    {
+      "path": "data/train-00002.jsonl.zst",
+      "bytes": 1778255290
+    },
+    {
+      "path": "data/train-00003.jsonl.zst",
+      "bytes": 1810440550
+    },
+    {
+      "path": "data/train-00004.jsonl.zst",
+      "bytes": 1660975662
+    },
+    {
+      "path": "data/train-00005.jsonl.zst",
+      "bytes": 1586082175
+    },
+    {
+      "path": "data/train-00006.jsonl.zst",
+      "bytes": 1584310369
+    },
+    {
+      "path": "data/train-00007.jsonl.zst",
+      "bytes": 1589029441
+    },
+    {
+      "path": "data/train-00008.jsonl.zst",
+      "bytes": 1497688531
+    },
+    {
+      "path": "data/train-00009.jsonl.zst",
+      "bytes": 1374347280
+    },
+    {
+      "path": "data/train-00010.jsonl.zst",
+      "bytes": 1536086889
+    },
+    {
+      "path": "data/train-00011.jsonl.zst",
+      "bytes": 1574422663
+    },
+    {
+      "path": "data/train-00012.jsonl.zst",
+      "bytes": 1575238814
+    },
+    {
+      "path": "data/train-00013.jsonl.zst",
+      "bytes": 1575765784
+    },
+    {
+      "path": "data/train-00014.jsonl.zst",
+      "bytes": 1620796375
+    },
+    {
+      "path": "data/train-00015.jsonl.zst",
+      "bytes": 1659984941
+    },
+    {
+      "path": "data/train-00016.jsonl.zst",
+      "bytes": 1661054210
+    },
+    {
+      "path": "data/train-00017.jsonl.zst",
+      "bytes": 1661671657
+    },
+    {
+      "path": "data/train-00018.jsonl.zst",
+      "bytes": 1569915109
+    },
+    {
+      "path": "data/train-00019.jsonl.zst",
+      "bytes": 1551014999
+    },
+    {
+      "path": "data/train-00020.jsonl.zst",
+      "bytes": 1476190707
+    },
+    {
+      "path": "data/train-00021.jsonl.zst",
+      "bytes": 440532604
+    }
+  ]
+}

metadata/source_summary.json

@@ -0,0 +1,70 @@
+{
+  "sources": [
+    {
+      "source": "dclm_edu",
+      "upstream_dataset": "HuggingFaceTB/dclm-edu",
+      "target_tokens": 4000000000,
+      "written_tokens": 4000000514,
+      "written_docs": 2954893,
+      "written_chars": 17819801584,
+      "language": "en",
+      "data_type": "English educational web",
+      "status": "completed"
+    },
+    {
+      "source": "opencsg_fineweb_edu_45",
+      "upstream_dataset": "opencsg/Fineweb-Edu-Chinese-V2.1",
+      "target_tokens": 5000000000,
+      "written_tokens": 5000007374,
+      "written_docs": 2873552,
+      "written_chars": 8889749082,
+      "language": "zh",
+      "data_type": "Chinese educational web",
+      "status": "completed"
+    },
+    {
+      "source": "finemath_4plus",
+      "upstream_dataset": "HuggingFaceTB/finemath",
+      "target_tokens": 2000000000,
+      "written_tokens": 2000411098,
+      "written_docs": 1305494,
+      "written_chars": 6554829219,
+      "language": "en",
+      "data_type": "English math corpus",
+      "status": "completed"
+    },
+    {
+      "source": "fineweb_edu_en",
+      "upstream_dataset": "AI-ModelScope/fineweb-edu",
+      "target_tokens": 4000000000,
+      "written_tokens": 4000003289,
+      "written_docs": 3742852,
+      "written_chars": 18050007746,
+      "language": "en",
+      "data_type": "English educational web",
+      "status": "completed"
+    },
+    {
+      "source": "chinese_fineweb_edu",
+      "upstream_dataset": "AI-ModelScope/chinese-fineweb-edu",
+      "target_tokens": 4000000000,
+      "written_tokens": 4000000142,
+      "written_docs": 3900669,
+      "written_chars": 6964548330,
+      "language": "zh",
+      "data_type": "Chinese educational web",
+      "status": "completed"
+    },
+    {
+      "source": "pile_deduplicated",
+      "upstream_dataset": "EleutherAI/the_pile_deduplicated via ModelScope mirror",
+      "target_tokens": 1000000000,
+      "written_tokens": 1000006901,
+      "written_docs": 717662,
+      "written_chars": 4095830408,
+      "language": "en",
+      "data_type": "general/reference replay",
+      "status": "completed"
+    }
+  ]
+}