data: add JupiMind J1R A100x4 12B-token foundation pretrain mix

Completed snapshot. sequences=2931118, missing=0, target_tokens=12000000000.

.gitattributes

@@ -59,3 +59,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.mp4 filter=lfs diff=lfs merge=lfs -text
 *.webm filter=lfs diff=lfs merge=lfs -text
 pretrain/j1r_24k_7b_mix/j1r_t4x4_24k_mix.jsonl filter=lfs diff=lfs merge=lfs -text
+pretrain/j1r_a100x4_24k_foundation/j1r_a100x4_24k_foundation_mix.jsonl filter=lfs diff=lfs merge=lfs -text
+pretrain/j1r_a100x4_24k_foundation/j1r_a100x4_24k_foundation_mix.hashes.txt filter=lfs diff=lfs merge=lfs -text

pretrain/j1r_a100x4_24k_foundation/LICENSE_NOTES.md

@@ -0,0 +1,3 @@
+# License Notes
+
+Composite dataset snapshot assembled from multiple upstream datasets. Review upstream dataset licenses and terms before redistribution or commercial use.

pretrain/j1r_a100x4_24k_foundation/MANIFEST.json

@@ -0,0 +1,18 @@
+{
+  "name": "JupiMind Dense J1R A100x4 24k foundation pretrain mix",
+  "format": "jsonl",
+  "schema": {
+    "text": "string"
+  },
+  "tokenizer": "aevynt/JupiMind-Dense-J1R-Tokenizer-24k",
+  "max_seq_len": 4096,
+  "target_tokens": 12000000000,
+  "target_sequences": 2931118,
+  "generated_sequences": 2931118,
+  "missing_sequences": 0,
+  "duplicate_rows_removed": 935713,
+  "input_rows": 3866831,
+  "data_file_bytes": 14939943680,
+  "hash_file_bytes": 120175838,
+  "created_at_utc": "2026-04-27T04:48:21.012170+00:00"
+}

pretrain/j1r_a100x4_24k_foundation/README.md

@@ -0,0 +1,45 @@
+---
+language:
+- vi
+- en
+tags:
+- jupimind
+- pretrain
+- vietnamese
+- language-modeling
+pretty_name: JupiMind Dense J1R A100x4 24k Foundation Pretrain Mix
+size_categories:
+- 1M<n<10M
+task_categories:
+- text-generation
+---
+
+# JupiMind Dense J1R A100x4 24k Foundation Pretrain Mix
+
+Snapshot dataset used for JupiMind Dense J1R foundation pretraining.
+
+## Files
+
+- `j1r_a100x4_24k_foundation_mix.jsonl`: pretrain corpus, one JSON object per line, schema `{"text": "..."}`.
+- `j1r_a100x4_24k_foundation_mix.summary.json`: build summary and source stats.
+- `j1r_a100x4_24k_foundation_mix.hashes.txt`: SHA1 hashes used for dedupe/reproducibility.
+- `j1r_foundation_a100x4_24k.yaml`: source recipe.
+- `MANIFEST.json`: compact machine-readable metadata.
+
+## Snapshot Stats
+
+- Generated sequences: `2931118`
+- Missing sequences: `0`
+- Target tokens: `12000000000`
+- Max sequence length: `4096`
+- Tokenizer: `aevynt/JupiMind-Dense-J1R-Tokenizer-24k`
+- Data file size: `13.91 GB`
+- Duplicate rows removed: `935713`
+
+## Intended Use
+
+Foundation pretraining dataset snapshot for the JupiMind Dense J1R model family. This is not an instruction-tuning dataset.
+
+## License Notes
+
+Composite training snapshot. Review upstream source licenses in the recipe/summary before redistribution or commercial use.

pretrain/j1r_a100x4_24k_foundation/j1r_a100x4_24k_foundation_mix.hashes.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:303e3607bcc641a1336a94e80ec3191fa03b3005b754d8e3a8a2972d9a5cf549
+size 120175838

pretrain/j1r_a100x4_24k_foundation/j1r_a100x4_24k_foundation_mix.jsonl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:934793b4154fbadc5c15ff409e6e23fc0b2552a9c823fcdad47c19461505b59c
+size 14939943680

pretrain/j1r_a100x4_24k_foundation/j1r_a100x4_24k_foundation_mix.summary.json

@@ -0,0 +1,388 @@
+{
+  "recipe_path": "configs/data/j1r_foundation_a100x4_24k.yaml",
+  "output_path": "/home/trainer/jupimind/outputs/pretrain/j1r_a100x4_24k_foundation_mix.jsonl",
+  "shard_dir": "/home/trainer/jupimind/outputs/pretrain/j1r_a100x4_24k_foundation_shards",
+  "target_tokens": 12000000000,
+  "target_sequences": 2931118,
+  "generated_sequences": 2931118,
+  "missing_sequences": 0,
+  "duplicate_rows_removed": 935713,
+  "input_rows": 3866831,
+  "sources": {
+    "fineweb2_vi": 937958,
+    "vietnamese_curated": 674128,
+    "wikipedia_vi": 214805,
+    "fineweb_edu": 337079,
+    "stackexchange_filtered": 234490,
+    "stackv2_edu": 175867,
+    "ubuntu_irc_filtered": 102589,
+    "python_peps": 642,
+    "mubeen_devops": 27931,
+    "devops_v1": 12,
+    "tech_docs": 1043,
+    "kubernetes_qa": 2831,
+    "linux_commands": 72,
+    "backfill_214985": 214985,
+    "fineweb2_recovery": 6686
+  },
+  "shard_summaries": {
+    "fineweb2_vi": {
+      "recipe_path": "/home/trainer/jupimind/configs/data/j1r_foundation_a100x4_24k.yaml",
+      "tokenizer_path": "aevynt/JupiMind-Dense-J1R-Tokenizer-24k",
+      "output_path": "/home/trainer/jupimind/outputs/pretrain/j1r_a100x4_24k_foundation_shards/fineweb2_vi.jsonl",
+      "source": "fineweb2_vi",
+      "target_tokens": 12000000000,
+      "target_sequences": 2931118,
+      "source_target_sequences": 937958,
+      "max_seq_len": 4096,
+      "min_chunk_tokens": 256,
+      "parallel_workers": 8,
+      "parallel_batch_docs": 768,
+      "sources": {
+        "fineweb2_vi": {
+          "dataset_name": "epfml/FineWeb2-HQ",
+          "config_name": "vie_Latn",
+          "target_sequences": 937958,
+          "written_sequences": 937958,
+          "seen_docs": 1227264
+        }
+      },
+      "generated_sequences": 937958,
+      "missing_sequences": 0
+    },
+    "vietnamese_curated": {
+      "recipe_path": "/home/trainer/jupimind/configs/data/j1r_foundation_a100x4_24k.yaml",
+      "tokenizer_path": "aevynt/JupiMind-Dense-J1R-Tokenizer-24k",
+      "output_path": "/home/trainer/jupimind/outputs/pretrain/j1r_a100x4_24k_foundation_shards/vietnamese_curated.jsonl",
+      "source": "vietnamese_curated",
+      "target_tokens": 12000000000,
+      "target_sequences": 2931118,
+      "source_target_sequences": 674157,
+      "max_seq_len": 4096,
+      "min_chunk_tokens": 256,
+      "parallel_workers": 8,
+      "parallel_batch_docs": 768,
+      "sources": {
+        "vietnamese_curated": {
+          "dataset_name": "VTSNLP/vietnamese_curated_dataset",
+          "config_name": null,
+          "target_sequences": 674157,
+          "written_sequences": 674157,
+          "seen_docs": 758016
+        }
+      },
+      "generated_sequences": 674157,
+      "missing_sequences": 0
+    },
+    "wikipedia_vi": {
+      "recipe_path": "/home/trainer/jupimind/configs/data/j1r_foundation_a100x4_24k.yaml",
+      "tokenizer_path": "aevynt/JupiMind-Dense-J1R-Tokenizer-24k",
+      "output_path": "/home/trainer/jupimind/outputs/pretrain/j1r_a100x4_24k_foundation_shards/wikipedia_vi.jsonl",
+      "source": "wikipedia_vi",
+      "target_tokens": 12000000000,
+      "target_sequences": 2931118,
+      "source_target_sequences": 219834,
+      "max_seq_len": 4096,
+      "min_chunk_tokens": 256,
+      "parallel_workers": 8,
+      "parallel_batch_docs": 768,
+      "sources": {
+        "wikipedia_vi": {
+          "dataset_name": "wikimedia/wikipedia",
+          "config_name": "20231101.vi",
+          "target_sequences": 219834,
+          "written_sequences": 219217,
+          "seen_docs": 1288680
+        }
+      },
+      "generated_sequences": 219217,
+      "missing_sequences": 617
+    },
+    "fineweb_edu": {
+      "recipe_path": "/home/trainer/jupimind/configs/data/j1r_foundation_a100x4_24k.yaml",
+      "tokenizer_path": "aevynt/JupiMind-Dense-J1R-Tokenizer-24k",
+      "output_path": "/home/trainer/jupimind/outputs/pretrain/j1r_a100x4_24k_foundation_shards/fineweb_edu.jsonl",
+      "source": "fineweb_edu",
+      "target_tokens": 12000000000,
+      "target_sequences": 2931118,
+      "source_target_sequences": 337079,
+      "max_seq_len": 4096,
+      "min_chunk_tokens": 256,
+      "parallel_workers": 8,
+      "parallel_batch_docs": 768,
+      "sources": {
+        "fineweb_edu": {
+          "dataset_name": "HuggingFaceFW/fineweb-edu",
+          "config_name": "sample-10BT",
+          "target_sequences": 337079,
+          "written_sequences": 337079,
+          "seen_docs": 376320
+        }
+      },
+      "generated_sequences": 337079,
+      "missing_sequences": 0
+    },
+    "stackexchange_filtered": {
+      "recipe_path": "/home/trainer/jupimind/configs/data/j1r_foundation_a100x4_24k.yaml",
+      "tokenizer_path": "aevynt/JupiMind-Dense-J1R-Tokenizer-24k",
+      "output_path": "/home/trainer/jupimind/outputs/pretrain/j1r_a100x4_24k_foundation_shards/stackexchange_filtered.jsonl",
+      "source": "stackexchange_filtered",
+      "target_tokens": 12000000000,
+      "target_sequences": 2931118,
+      "source_target_sequences": 234490,
+      "max_seq_len": 4096,
+      "min_chunk_tokens": 256,
+      "parallel_workers": 8,
+      "parallel_batch_docs": 768,
+      "sources": {
+        "stackexchange_filtered": {
+          "dataset_name": "common-pile/stackexchange_filtered",
+          "config_name": null,
+          "target_sequences": 234490,
+          "written_sequences": 234490,
+          "seen_docs": 323328
+        }
+      },
+      "generated_sequences": 234490,
+      "missing_sequences": 0
+    },
+    "stackv2_edu": {
+      "recipe_path": "/home/trainer/jupimind/configs/data/j1r_foundation_a100x4_24k.yaml",
+      "tokenizer_path": "aevynt/JupiMind-Dense-J1R-Tokenizer-24k",
+      "output_path": "/home/trainer/jupimind/outputs/pretrain/j1r_a100x4_24k_foundation_shards/stackv2_edu.jsonl",
+      "source": "stackv2_edu",
+      "target_tokens": 12000000000,
+      "target_sequences": 2931118,
+      "source_target_sequences": 175867,
+      "max_seq_len": 4096,
+      "min_chunk_tokens": 256,
+      "parallel_workers": 8,
+      "parallel_batch_docs": 768,
+      "sources": {
+        "stackv2_edu": {
+          "dataset_name": "common-pile/stackv2_edu_filtered",
+          "config_name": null,
+          "target_sequences": 175867,
+          "written_sequences": 175867,
+          "seen_docs": 433481
+        }
+      },
+      "generated_sequences": 175867,
+      "missing_sequences": 0
+    },
+    "ubuntu_irc_filtered": {
+      "recipe_path": "/home/trainer/jupimind/configs/data/j1r_foundation_a100x4_24k.yaml",
+      "tokenizer_path": "aevynt/JupiMind-Dense-J1R-Tokenizer-24k",
+      "output_path": "/home/trainer/jupimind/outputs/pretrain/j1r_a100x4_24k_foundation_shards/ubuntu_irc_filtered.jsonl",
+      "source": "ubuntu_irc_filtered",
+      "target_tokens": 12000000000,
+      "target_sequences": 2931118,
+      "source_target_sequences": 102589,
+      "max_seq_len": 4096,
+      "min_chunk_tokens": 256,
+      "parallel_workers": 8,
+      "parallel_batch_docs": 768,
+      "sources": {
+        "ubuntu_irc_filtered": {
+          "dataset_name": "common-pile/ubuntu_irc_filtered",
+          "config_name": null,
+          "target_sequences": 102589,
+          "written_sequences": 102589,
+          "seen_docs": 112793
+        }
+      },
+      "generated_sequences": 102589,
+      "missing_sequences": 0
+    },
+    "python_peps": {
+      "recipe_path": "/home/trainer/jupimind/configs/data/j1r_foundation_a100x4_24k.yaml",
+      "tokenizer_path": "aevynt/JupiMind-Dense-J1R-Tokenizer-24k",
+      "output_path": "/home/trainer/jupimind/outputs/pretrain/j1r_a100x4_24k_foundation_shards/python_peps.jsonl",
+      "source": "python_peps",
+      "target_tokens": 12000000000,
+      "target_sequences": 2931118,
+      "source_target_sequences": 5862,
+      "max_seq_len": 4096,
+      "min_chunk_tokens": 256,
+      "parallel_workers": 8,
+      "parallel_batch_docs": 768,
+      "sources": {
+        "python_peps": {
+          "dataset_name": "common-pile/python_enhancement_proposals_filtered",
+          "config_name": null,
+          "target_sequences": 5862,
+          "written_sequences": 642,
+          "seen_docs": 655
+        }
+      },
+      "generated_sequences": 642,
+      "missing_sequences": 5220
+    },
+    "mubeen_devops": {
+      "recipe_path": "/home/trainer/jupimind/configs/data/j1r_foundation_a100x4_24k.yaml",
+      "tokenizer_path": "aevynt/JupiMind-Dense-J1R-Tokenizer-24k",
+      "output_path": "/home/trainer/jupimind/outputs/pretrain/j1r_a100x4_24k_foundation_shards/mubeen_devops.jsonl",
+      "source": "mubeen_devops",
+      "target_tokens": 12000000000,
+      "target_sequences": 2931118,
+      "source_target_sequences": 146556,
+      "max_seq_len": 4096,
+      "min_chunk_tokens": 256,
+      "parallel_workers": 8,
+      "parallel_batch_docs": 768,
+      "sources": {
+        "mubeen_devops": {
+          "dataset_name": "Mubeen161/DEVOPS",
+          "config_name": null,
+          "target_sequences": 146556,
+          "written_sequences": 27931,
+          "seen_docs": 42819
+        }
+      },
+      "generated_sequences": 27931,
+      "missing_sequences": 118625
+    },
+    "devops_v1": {
+      "recipe_path": "/home/trainer/jupimind/configs/data/j1r_foundation_a100x4_24k.yaml",
+      "tokenizer_path": "aevynt/JupiMind-Dense-J1R-Tokenizer-24k",
+      "output_path": "/home/trainer/jupimind/outputs/pretrain/j1r_a100x4_24k_foundation_shards/devops_v1.jsonl",
+      "source": "devops_v1",
+      "target_tokens": 12000000000,
+      "target_sequences": 2931118,
+      "source_target_sequences": 35173,
+      "max_seq_len": 4096,
+      "min_chunk_tokens": 256,
+      "parallel_workers": 8,
+      "parallel_batch_docs": 768,
+      "sources": {
+        "devops_v1": {
+          "dataset_name": "pavanmantha/devops-v1",
+          "config_name": null,
+          "target_sequences": 35173,
+          "written_sequences": 12,
+          "seen_docs": 312
+        }
+      },
+      "generated_sequences": 12,
+      "missing_sequences": 35161
+    },
+    "tech_docs": {
+      "recipe_path": "/home/trainer/jupimind/configs/data/j1r_foundation_a100x4_24k.yaml",
+      "tokenizer_path": "aevynt/JupiMind-Dense-J1R-Tokenizer-24k",
+      "output_path": "/home/trainer/jupimind/outputs/pretrain/j1r_a100x4_24k_foundation_shards/tech_docs.jsonl",
+      "source": "tech_docs",
+      "target_tokens": 12000000000,
+      "target_sequences": 2931118,
+      "source_target_sequences": 29311,
+      "max_seq_len": 4096,
+      "min_chunk_tokens": 256,
+      "parallel_workers": 8,
+      "parallel_batch_docs": 768,
+      "sources": {
+        "tech_docs": {
+          "dataset_name": "saidsef/tech-docs",
+          "config_name": null,
+          "target_sequences": 29311,
+          "written_sequences": 1043,
+          "seen_docs": 1397
+        }
+      },
+      "generated_sequences": 1043,
+      "missing_sequences": 28268
+    },
+    "kubernetes_qa": {
+      "recipe_path": "/home/trainer/jupimind/configs/data/j1r_foundation_a100x4_24k.yaml",
+      "tokenizer_path": "aevynt/JupiMind-Dense-J1R-Tokenizer-24k",
+      "output_path": "/home/trainer/jupimind/outputs/pretrain/j1r_a100x4_24k_foundation_shards/kubernetes_qa.jsonl",
+      "source": "kubernetes_qa",
+      "target_tokens": 12000000000,
+      "target_sequences": 2931118,
+      "source_target_sequences": 23449,
+      "max_seq_len": 4096,
+      "min_chunk_tokens": 256,
+      "parallel_workers": 8,
+      "parallel_batch_docs": 768,
+      "sources": {
+        "kubernetes_qa": {
+          "dataset_name": "sidddddddddddd/kubernetes",
+          "config_name": null,
+          "target_sequences": 23449,
+          "written_sequences": 2831,
+          "seen_docs": 5400
+        }
+      },
+      "generated_sequences": 2831,
+      "missing_sequences": 20618
+    },
+    "linux_commands": {
+      "recipe_path": "/home/trainer/jupimind/configs/data/j1r_foundation_a100x4_24k.yaml",
+      "tokenizer_path": "aevynt/JupiMind-Dense-J1R-Tokenizer-24k",
+      "output_path": "/home/trainer/jupimind/outputs/pretrain/j1r_a100x4_24k_foundation_shards/linux_commands.jsonl",
+      "source": "linux_commands",
+      "target_tokens": 12000000000,
+      "target_sequences": 2931118,
+      "source_target_sequences": 8793,
+      "max_seq_len": 4096,
+      "min_chunk_tokens": 256,
+      "parallel_workers": 8,
+      "parallel_batch_docs": 768,
+      "sources": {
+        "linux_commands": {
+          "dataset_name": "hotal/linux_commands",
+          "config_name": null,
+          "target_sequences": 8793,
+          "written_sequences": 72,
+          "seen_docs": 481
+        }
+      },
+      "generated_sequences": 72,
+      "missing_sequences": 8721
+    },
+    "backfill_214985": {
+      "recipe_path": "configs/data/j1r_foundation_a100x4_24k.yaml",
+      "tokenizer_path": "aevynt/JupiMind-Dense-J1R-Tokenizer-24k",
+      "output_path": "/home/trainer/jupimind/outputs/pretrain/j1r_a100x4_24k_foundation_shards/backfill_214985.jsonl",
+      "target_sequences": 214985,
+      "generated_sequences": 214985,
+      "missing_sequences": 0,
+      "parallel_workers": 8,
+      "parallel_batch_docs": 768,
+      "exclude_hash_path": "outputs/pretrain/j1r_a100x4_24k_foundation_mix.hashes.txt",
+      "sources": {
+        "vietnamese_curated": {
+          "dataset_name": "VTSNLP/vietnamese_curated_dataset",
+          "config_name": null,
+          "target_sequences": 214985,
+          "written_sequences": 214985,
+          "seen_docs": 996864,
+          "skip_rows": 758016
+        }
+      }
+    },
+    "fineweb2_recovery": {
+      "recipe_path": "configs/data/j1r_foundation_a100x4_24k.yaml",
+      "tokenizer_path": "aevynt/JupiMind-Dense-J1R-Tokenizer-24k",
+      "output_path": "/home/trainer/jupimind/outputs/pretrain/j1r_a100x4_24k_foundation_shards/fineweb2_recovery.jsonl",
+      "source": "fineweb2_vi",
+      "target_tokens": 12000000000,
+      "target_sequences": 2931118,
+      "source_target_sequences": 937958,
+      "max_seq_len": 4096,
+      "min_chunk_tokens": 256,
+      "parallel_workers": 8,
+      "parallel_batch_docs": 768,
+      "sources": {
+        "fineweb2_vi": {
+          "dataset_name": "epfml/FineWeb2-HQ",
+          "config_name": "vie_Latn",
+          "target_sequences": 937958,
+          "written_sequences": 937958,
+          "seen_docs": 1227264,
+          "skip_rows": 0
+        }
+      },
+      "generated_sequences": 937958,
+      "missing_sequences": 0
+    }
+  }
+}

pretrain/j1r_a100x4_24k_foundation/j1r_foundation_a100x4_24k.yaml

@@ -0,0 +1,111 @@
+tokenizer_path: aevynt/JupiMind-Dense-J1R-Tokenizer-24k
+output_path: outputs/pretrain/j1r_a100x4_24k_foundation_mix.jsonl
+summary_path: outputs/pretrain/j1r_a100x4_24k_foundation_mix.summary.json
+target_tokens: 12000000000
+max_seq_len: 4096
+min_chunk_tokens: 256
+min_text_chars: 256
+max_text_chars: 120000
+seed: 42
+seed_text_paths:
+  - data/tokenizer/j1r_server_seed.txt
+parallel:
+  workers: 8
+  batch_docs: 768
+backfill_order:
+  - vietnamese_curated
+  - fineweb_edu
+  - stackexchange_filtered
+  - stackv2_edu
+  - ubuntu_irc_filtered
+  - mubeen_devops
+  - wikipedia_vi
+  - fineweb2_vi
+quality:
+  max_url_ratio: 0.02
+  max_non_word_ratio: 0.35
+  max_line_repeat_ratio: 0.35
+  max_char_repeat_run: 12
+  max_upper_ratio: 0.35
+sources:
+  - name: fineweb2_vi
+    dataset_name: epfml/FineWeb2-HQ
+    config_name: vie_Latn
+    split: train
+    weight: 0.32
+    extractor: text
+    backfill_priority: 100
+  - name: vietnamese_curated
+    dataset_name: VTSNLP/vietnamese_curated_dataset
+    split: train
+    weight: 0.23
+    extractor: text
+    backfill_priority: 95
+  - name: wikipedia_vi
+    dataset_name: wikimedia/wikipedia
+    config_name: 20231101.vi
+    split: train
+    weight: 0.075
+    extractor: wikipedia
+    backfill_priority: 90
+  - name: fineweb_edu
+    dataset_name: HuggingFaceFW/fineweb-edu
+    config_name: sample-10BT
+    split: train
+    weight: 0.115
+    extractor: text
+    backfill_priority: 85
+  - name: stackexchange_filtered
+    dataset_name: common-pile/stackexchange_filtered
+    split: train
+    weight: 0.08
+    extractor: text
+    backfill_priority: 80
+  - name: stackv2_edu
+    dataset_name: common-pile/stackv2_edu_filtered
+    split: train
+    weight: 0.06
+    extractor: text
+    backfill_priority: 75
+  - name: ubuntu_irc_filtered
+    dataset_name: common-pile/ubuntu_irc_filtered
+    split: train
+    weight: 0.035
+    extractor: text
+    backfill_priority: 70
+  - name: python_peps
+    dataset_name: common-pile/python_enhancement_proposals_filtered
+    split: train
+    weight: 0.002
+    extractor: text
+    backfill_priority: 65
+  - name: mubeen_devops
+    dataset_name: Mubeen161/DEVOPS
+    split: train
+    weight: 0.05
+    extractor: instruction_response
+    backfill_priority: 88
+  - name: devops_v1
+    dataset_name: pavanmantha/devops-v1
+    split: train
+    weight: 0.012
+    extractor: question_answer
+    backfill_priority: 78
+  - name: tech_docs
+    dataset_name: saidsef/tech-docs
+    split: train
+    weight: 0.01
+    extractor: question_answer
+    backfill_priority: 76
+  - name: kubernetes_qa
+    dataset_name: sidddddddddddd/kubernetes
+    split: train
+    weight: 0.008
+    extractor: question_answer
+    backfill_priority: 74
+  - name: linux_commands
+    dataset_name: hotal/linux_commands
+    split: train
+    weight: 0.003
+    extractor: question_answer
+    backfill_priority: 72