Datasets:

IO-AIR
/

filterme-sft

+# IO-AIR/filterme-sft — v1.7 build appendix
+Machine-facing detail supplementing the public README. Generated from `build/v1.7/state/` (progress, pushed_sources, qa_report) — the same files that make every release reproducible and auditable.
+## Per-source pools and kept rows
+| source | kept rows | kept gpt2 | pool rows (shard) |
+|---|---:|---:|---:|
+| longctx_forge | 241 | 386,971,361 | 451 |
+| openmath_cot | 47,510 | 381,589,384 | 66,787 |
+| opensci | 49,751 | 294,581,048 | 65,779 |
+| toolmind | 13,139 | 171,025,036 | 13,351 |
+| ultradata_code | 2,764 | 163,636,053 | 4,200 |
+| ollvm_re | 704 | 115,095,930 | 704 |
+| nemotron_swe | 1,694 | 95,892,622 | 4,702 |
+| swesmith | 2,880 | 94,358,439 | 5,174 |
+| coderforge | 948 | 78,556,491 | 2,184 |
+| pyromind_long | 987 | 58,427,013 | 1,000 |
+| ultradata_search | 1,306 | 39,506,375 | 2,576 |
+| sophia_svg | 15,948 | 38,153,242 | 15,948 |
+| ultradata_general | 1,338 | 33,996,539 | 2,692 |
+| pyromind_short | 1,846 | 14,409,738 | 1,932 |
+| re_synth_forge | 5,100 | 12,075,013 | 6,059 |
+| ultradata_tooluse | 1,226 | 8,740,063 | 2,095 |
+| agenticsft | 500 | 8,115,596 | 500 |
+| cad_evo_forge | 3,593 | 3,389,137 | 5,820 |
+| icons_forge | 6,354 | 2,261,814 | 6,354 |
+| nemotron_cuda | 88 | 847,831 | 2,276 |
+| mcp_atlas | 31 | 729,868 | 462 |
+| cad_synth_forge | 3,204 | 709,604 | 3,204 |
+| demyagent | 228 | 628,420 | 500 |
+| vision_docmatix | 1,617 | 225,903 | 1,617 |
+| plot2code | 379 | 212,787 | 379 |
+| multiturn | 177 | 177,528 | 192 |
+| unified | 758 | 148,433 | 758 |
+| vision_forge | 1,232 | 68,406 | 1,232 |
+| cadforge_cq | 1 | 2,667 | 3 |
+| cadforge_3d | 1 | 2,028 | 14 |
+| **total** | **165,545** | **2,004,534,369** | 218,945 |
+## Quota drops (rebalance)
+| family | rows dropped | share before |
+|---|---:|---:|
+| reasoning_math | 19,277 | 0.1922 |
+| reasoning_sci | 16,028 | 0.1484 |
+| agentic_swe | 2,361 | 0.1811 |
+| general_sft | 1,584 | 0.1002 |
+| code_agent | 410 | 0.1111 |
+| reverse_engineering | 118 | 0.1429 |
+## Quota policy (`quota_keep`)
+```json
+{
+  "policy": "normalized-caps cascade, tolerance exit",
+  "rounds": 4,
+  "max_violation": 0.0114,
+  "tolerance": 0.0125,
+  "caps_normalized": {
+    "reasoning_math": 0.17900426266960132,
+    "reasoning_sci": 0.13818850472236924,
+    "agentic_swe": 0.16869584598668264,
+    "reverse_engineering": 0.13359150809350012,
+    "code_agent": 0.1035020756136294,
+    "general_sft": 0.09333296185885827,
+    "agentic_tool": 0.15267600924971442,
+    "vision_svg": 0.02813919148580503,
+    "cad_3d": 0.0020895439222132447,
+    "style_house": 0.0004179087844426489,
+    "vision_image": 0.00020895439222132445,
+    "vision_plot": 0.00015323322096230462
+  }
+}
+```
+## Generator diversity (per source, post-pareto)
+| source | distinct tasks | top task | top share |
+|---|---:|---|---:|
+| openmath_cot | 6 | math | 55.1% |
+| opensci | 2 | science | 57.4% |
+| mcp_atlas | 1 | mcp | 100.0% |
+| pyromind_long | 1 | tooluse | 100.0% |
+| pyromind_short | 2 | tooluse | 56.7% |
+| toolmind | 2 | tooluse | 54.3% |
+| ultradata_tooluse | 1 | tooluse | 100.0% |
+| nemotron_swe | 2310 | swe | 7.9% |
+| swesmith | 1685 | swe | 41.6% |
+| coderforge | 801 | swe | 29.2% |
+| nemotron_cuda | 1 | cuda | 100.0% |
+| ultradata_code | 3001 | coding | 3.0% |
+| agenticsft | 1 | agent | 100.0% |
+| demyagent | 1 | agent | 100.0% |
+| ultradata_general | 2001 | general | 7.6% |
+| ultradata_search | 2001 | search | 0.0% |
+| ollvm_re | 2 | re | 99.9% |
+| cadforge_3d | 1 | threedee | 100.0% |
+| cadforge_cq | 1 | cadquery | 100.0% |
+| multiturn | 1 | style | 100.0% |
+| unified | 1 | style | 100.0% |
+| vision_docmatix | 2 | vision | 88.3% |
+| vision_forge | 1 | vision | 100.0% |
+| plot2code | 2 | vision | 97.1% |
+| sophia_svg | 1 | text2svg | 100.0% |
+| icons_forge | 9 | tabler-outline | 41.5% |
+| re_synth_forge | 2937 | rle | 2.2% |
+| cad_synth_forge | 23 | plate_holes | 9.6% |
+| cad_evo_forge | 298 | box_stack_pocket_boss | 14.6% |
+| longctx_forge | 14 | utility library header rebuild | 37.0% |
+## Top tags
+| tag | rows |
+|---|---:|
+| `cot` | 122,820 |
+| `tok:1k-5k` | 77,983 |
+| `turns:short` | 73,056 |
+| `turns:mono` | 58,094 |
+| `tok:5k-20k` | 50,119 |
+| `domain:reasoning_math` | 47,510 |
+| `tools` | 41,452 |
+| `domain:reasoning_sci` | 33,147 |
+| `domain:vision_svg` | 22,302 |
+| `src:openmath_cot` | 21,385 |
+| `src:opensci` | 21,183 |
+| `tok:1k` | 19,257 |
+| `domain:reasoning_science` | 16,604 |
+| `domain:agentic_tool` | 16,069 |
+| `src:sophia_svg` | 15,948 |
+| `turns:mega` | 12,104 |
+| `turns:deep` | 11,233 |
+| `turns:mid` | 11,058 |
+| `toolcalls:10+` | 9,721 |
+| `toolcalls:1-3` | 9,165 |
+| `tok:20k-50k` | 8,043 |
+| `domain:cad_3d` | 6,799 |
+| `tok:<1k` | 6,411 |
+| `src:icons_forge` | 6,354 |
+## Pipeline (verbatim)
+`carry vetted v1.1 shards -> top-up fetch (rows+parquet bulk, offset-cursor resume) -> per-row gates (shape, non_latin, think-balance, dup, license, no_image) -> ONE global dedup index seeded with the carried corpus -> real gpt2 BPE per row -> per-slice pareto (strict) -> domain token-share caps -> deterministic interleave.`

ENSEMBLE_CARD.json ADDED Viewed

	@@ -0,0 +1,412 @@

+{
+ "version": "v1.7",
+ "rows": 165545,
+ "gpt2_tokens_total": 2004534369,
+ "sources": {
+  "openmath_cot": {
+   "rows": 66787,
+   "gpt2": 535300110
+  },
+  "opensci": {
+   "rows": 65779,
+   "gpt2": 389184121
+  },
+  "mcp_atlas": {
+   "rows": 462,
+   "gpt2": 7583866
+  },
+  "pyromind_long": {
+   "rows": 1000,
+   "gpt2": 58991782
+  },
+  "pyromind_short": {
+   "rows": 1932,
+   "gpt2": 14639677
+  },
+  "toolmind": {
+   "rows": 13351,
+   "gpt2": 171755540
+  },
+  "ultradata_tooluse": {
+   "rows": 2095,
+   "gpt2": 13520358
+  },
+  "nemotron_swe": {
+   "rows": 4702,
+   "gpt2": 271084502
+  },
+  "swesmith": {
+   "rows": 5174,
+   "gpt2": 164528053
+  },
+  "coderforge": {
+   "rows": 2184,
+   "gpt2": 199432411
+  },
+  "nemotron_cuda": {
+   "rows": 2276,
+   "gpt2": 23010820
+  },
+  "ultradata_code": {
+   "rows": 4200,
+   "gpt2": 276443679
+  },
+  "agenticsft": {
+   "rows": 500,
+   "gpt2": 8115596
+  },
+  "demyagent": {
+   "rows": 500,
+   "gpt2": 1179050
+  },
+  "ultradata_general": {
+   "rows": 2692,
+   "gpt2": 69243262
+  },
+  "ultradata_search": {
+   "rows": 2576,
+   "gpt2": 76121807
+  },
+  "ollvm_re": {
+   "rows": 704,
+   "gpt2": 115095930
+  },
+  "cadforge_3d": {
+   "rows": 14,
+   "gpt2": 35785
+  },
+  "cadforge_cq": {
+   "rows": 3,
+   "gpt2": 9347
+  },
+  "multiturn": {
+   "rows": 192,
+   "gpt2": 194013
+  },
+  "unified": {
+   "rows": 758,
+   "gpt2": 148433
+  },
+  "vision_docmatix": {
+   "rows": 1617,
+   "gpt2": 225903
+  },
+  "vision_forge": {
+   "rows": 1232,
+   "gpt2": 68406
+  },
+  "plot2code": {
+   "rows": 379,
+   "gpt2": 212787
+  },
+  "sophia_svg": {
+   "rows": 15948,
+   "gpt2": 38153242
+  },
+  "icons_forge": {
+   "rows": 6354,
+   "gpt2": 2261814
+  },
+  "re_synth_forge": {
+   "rows": 6059,
+   "gpt2": 17042338
+  },
+  "cad_synth_forge": {
+   "rows": 3204,
+   "gpt2": 709604
+  },
+  "cad_evo_forge": {
+   "rows": 5820,
+   "gpt2": 6612782
+  },
+  "longctx_forge": {
+   "rows": 451,
+   "gpt2": 752712105
+  }
+ },
+ "domains": {
+  "reasoning_math": {
+   "gpt2_tokens": 381589384,
+   "share": 0.1904
+  },
+  "agentic_swe": {
+   "gpt2_tokens": 359611343,
+   "share": 0.1794
+  },
+  "reasoning_sci": {
+   "gpt2_tokens": 294581048,
+   "share": 0.147
+  },
+  "reverse_engineering": {
+   "gpt2_tokens": 284560356,
+   "share": 0.142
+  },
+  "code_agent": {
+   "gpt2_tokens": 220638595,
+   "share": 0.1101
+  },
+  "agentic_tool": {
+   "gpt2_tokens": 219243982,
+   "share": 0.1094
+  },
+  "general_sft": {
+   "gpt2_tokens": 198958112,
+   "share": 0.0993
+  },
+  "vision_svg": {
+   "gpt2_tokens": 40415056,
+   "share": 0.0202
+  },
+  "cad_3d": {
+   "gpt2_tokens": 4103436,
+   "share": 0.002
+  },
+  "style_house": {
+   "gpt2_tokens": 325961,
+   "share": 0.0002
+  },
+  "vision_image": {
+   "gpt2_tokens": 294309,
+   "share": 0.0001
+  },
+  "vision_plot": {
+   "gpt2_tokens": 212787,
+   "share": 0.0001
+  }
+ },
+ "top_tags": {
+  "cot": 122820,
+  "tok:1k-5k": 77983,
+  "turns:short": 73056,
+  "turns:mono": 58094,
+  "tok:5k-20k": 50119,
+  "domain:reasoning_math": 47510,
+  "tools": 41452,
+  "domain:reasoning_sci": 33147,
+  "domain:vision_svg": 22302,
+  "src:openmath_cot": 21385,
+  "src:opensci": 21183,
+  "tok:1k": 19257,
+  "domain:reasoning_science": 16604,
+  "domain:agentic_tool": 16069,
+  "src:sophia_svg": 15948,
+  "turns:mega": 12104,
+  "turns:deep": 11233,
+  "turns:mid": 11058,
+  "toolcalls:10+": 9721,
+  "toolcalls:1-3": 9165,
+  "tok:20k-50k": 8043,
+  "domain:cad_3d": 6799,
+  "tok:<1k": 6411,
+  "src:icons_forge": 6354
+ },
+ "pareto_dropped_total": 13622,
+ "quota_dropped": {
+  "reasoning_math": {
+   "rows": 19277,
+   "tokens": 153710726,
+   "share_before": 0.1922
+  },
+  "reasoning_sci": {
+   "rows": 16028,
+   "tokens": 94603073,
+   "share_before": 0.1484
+  },
+  "general_sft": {
+   "rows": 1584,
+   "tokens": 100403486,
+   "share_before": 0.1002
+  },
+  "agentic_swe": {
+   "rows": 2361,
+   "tokens": 194499359,
+   "share_before": 0.1811
+  },
+  "reverse_engineering": {
+   "rows": 118,
+   "tokens": 213109436,
+   "share_before": 0.1429
+  },
+  "code_agent": {
+   "rows": 410,
+   "tokens": 24940129,
+   "share_before": 0.1111
+  }
+ },
+ "quota_meta": {
+  "policy": "normalized-caps cascade, tolerance exit",
+  "rounds": 4,
+  "max_violation": 0.0114,
+  "tolerance": 0.0125,
+  "caps_normalized": {
+   "reasoning_math": 0.17900426266960132,
+   "reasoning_sci": 0.13818850472236924,
+   "agentic_swe": 0.16869584598668264,
+   "reverse_engineering": 0.13359150809350012,
+   "code_agent": 0.1035020756136294,
+   "general_sft": 0.09333296185885827,
+   "agentic_tool": 0.15267600924971442,
+   "vision_svg": 0.02813919148580503,
+   "cad_3d": 0.0020895439222132447,
+   "style_house": 0.0004179087844426489,
+   "vision_image": 0.00020895439222132445,
+   "vision_plot": 0.00015323322096230462
+  }
+ },
+ "gen_diversity": {
+  "openmath_cot": {
+   "tasks": 6,
+   "top_task": "math",
+   "top_share": 0.551
+  },
+  "opensci": {
+   "tasks": 2,
+   "top_task": "science",
+   "top_share": 0.574
+  },
+  "mcp_atlas": {
+   "tasks": 1,
+   "top_task": "mcp",
+   "top_share": 1.0
+  },
+  "pyromind_long": {
+   "tasks": 1,
+   "top_task": "tooluse",
+   "top_share": 1.0
+  },
+  "pyromind_short": {
+   "tasks": 2,
+   "top_task": "tooluse",
+   "top_share": 0.567
+  },
+  "toolmind": {
+   "tasks": 2,
+   "top_task": "tooluse",
+   "top_share": 0.543
+  },
+  "ultradata_tooluse": {
+   "tasks": 1,
+   "top_task": "tooluse",
+   "top_share": 1.0
+  },
+  "nemotron_swe": {
+   "tasks": 2310,
+   "top_task": "swe",
+   "top_share": 0.079
+  },
+  "swesmith": {
+   "tasks": 1685,
+   "top_task": "swe",
+   "top_share": 0.416
+  },
+  "coderforge": {
+   "tasks": 801,
+   "top_task": "swe",
+   "top_share": 0.292
+  },
+  "nemotron_cuda": {
+   "tasks": 1,
+   "top_task": "cuda",
+   "top_share": 1.0
+  },
+  "ultradata_code": {
+   "tasks": 3001,
+   "top_task": "coding",
+   "top_share": 0.03
+  },
+  "agenticsft": {
+   "tasks": 1,
+   "top_task": "agent",
+   "top_share": 1.0
+  },
+  "demyagent": {
+   "tasks": 1,
+   "top_task": "agent",
+   "top_share": 1.0
+  },
+  "ultradata_general": {
+   "tasks": 2001,
+   "top_task": "general",
+   "top_share": 0.076
+  },
+  "ultradata_search": {
+   "tasks": 2001,
+   "top_task": "search",
+   "top_share": 0.0
+  },
+  "ollvm_re": {
+   "tasks": 2,
+   "top_task": "re",
+   "top_share": 0.999
+  },
+  "cadforge_3d": {
+   "tasks": 1,
+   "top_task": "threedee",
+   "top_share": 1.0
+  },
+  "cadforge_cq": {
+   "tasks": 1,
+   "top_task": "cadquery",
+   "top_share": 1.0
+  },
+  "multiturn": {
+   "tasks": 1,
+   "top_task": "style",
+   "top_share": 1.0
+  },
+  "unified": {
+   "tasks": 1,
+   "top_task": "style",
+   "top_share": 1.0
+  },
+  "vision_docmatix": {
+   "tasks": 2,
+   "top_task": "vision",
+   "top_share": 0.883
+  },
+  "vision_forge": {
+   "tasks": 1,
+   "top_task": "vision",
+   "top_share": 1.0
+  },
+  "plot2code": {
+   "tasks": 2,
+   "top_task": "vision",
+   "top_share": 0.971
+  },
+  "sophia_svg": {
+   "tasks": 1,
+   "top_task": "text2svg",
+   "top_share": 1.0
+  },
+  "icons_forge": {
+   "tasks": 9,
+   "top_task": "tabler-outline",
+   "top_share": 0.415
+  },
+  "re_synth_forge": {
+   "tasks": 2937,
+   "top_task": "rle",
+   "top_share": 0.022
+  },
+  "cad_synth_forge": {
+   "tasks": 23,
+   "top_task": "plate_holes",
+   "top_share": 0.096
+  },
+  "cad_evo_forge": {
+   "tasks": 298,
+   "top_task": "box_stack_pocket_boss",
+   "top_share": 0.146
+  },
+  "longctx_forge": {
+   "tasks": 14,
+   "top_task": "utility library header rebuild",
+   "top_share": 0.37
+  }
+ },
+ "license_note": "per-row license in meta.license; carried rows keep their v1.0/v1.1 vetting; carry-only sources: agenticsft/demyagent (license unverified upstream), cadforge_3d (upstream dark).",
+ "write_wall_s": 346.8,
+ "pipeline": "carry vetted v1.1 shards -> top-up fetch (rows+parquet bulk, offset-cursor resume) -> per-row gates (shape, non_latin, think-balance, dup, license, no_image) -> ONE global dedup index seeded with the carried corpus -> real gpt2 BPE per row -> per-slice pareto (strict) -> domain token-share caps -> deterministic interleave.",
+ "card_rev": "2-polished"
+}

README.md ADDED Viewed

	@@ -0,0 +1,195 @@

+---
+license: cc-by-4.0
+task_categories:
+- question-answering
+- text-generation
+language:
+- en
+size_categories:
+- 100K<n<1M
+tags:
+- sft
+- instruction-tuning
+- agentic
+- tool-calling
+- reverse-engineering
+- cad
+- cadquery
+- svg
+- long-context
+- code
+- math
+- science
+- synthetic
+- pareto-pruned
+pretty_name: filterme-sft v1.7
+source_datasets: sft-forge
+---
+# filterme-sft · v1.7
+> **Current release: `v1.7`** — the repo root always holds the latest release: finals (`train.jsonl.gz`, `manifest.jsonl.gz`), per-source provenance pools (`shards/`), build state and the QA report. One clean base — no historical build trees.
+| rows | real gpt2 tokens | sources | pareto pruned | QA |
+|---:|---:|---:|---:|---|
+| 165,545 | 2,004,534,369 | 30 | 13,622 | **all gates green** |
+One pareto-pruned, token-rebalanced SFT mixture built by **sft-forge**: ingest → tag → gate → ONE global dedup → strict per-slice pareto (quality vs real gpt2 token cost) → domain token-share caps → deterministic interleave. Every row carries fine-grained tags and a true BPE token count — filter, never re-split. The name is the doctrine: **filtered, not scraped**.
+## Version history
+| version | rows | gpt2 tokens | sources | highlights |
+|---|---:|---:|---:|---|
+| v0.3 | 1 | 123 | 1 | smoke fixture. Briefly the public card by accident (a stale upload_folder clobber); retired for good in this push. |
+| v1.0 | — | — | — | first ensemble: multi-source ingest, ONE global dedup index, per-slice strict pareto. |
+| v1.1 | — | — | — | scale-up + license-purity gates (Apache/MIT/CC-BY-4.0/ISC) on every forge lane. |
+| v1.2 | 149,358 | 1,683,409,624 | 24 | first full publication audit (0 diffs vs card); incremental per-source shard pushes. |
+| v1.3 | 128,008 | ~1,452,443,000 | 29 | lanes: sophia_svg, icons_forge, re_synth_forge, cad_synth_forge, longctx_forge. |
+| v1.4 | 130,797 | 1,352,839,219 | 29 | caps reshuffle + generator diversity v2 (RE share 0.1% → 13.7%); concision envelopes. |
+| v1.5 | 139,707 | 1,452,971,107 | 30 | cad_evo exec-verified CAD evolution sessions; longctx monster recovery; quota_keep rewrite (normalized caps, tolerance exit, input-robust). |
+| v1.6 | 137,956 | 1,462,487,838 | 30 | lingctx growth (128 mega sessions, 199.8M gpt2 kept) + cad_evo tighter-tolerance replacement (0.25% / 0.10 mm verify band, 4-7-step arcs, 41% repair loops, 2,632 kept); caps rebalanced to keep the monsters in-mix. |
+| v1.7 ← **current** | 165,545 | 2,004,534,369 | 30 | the 2B release on the clean filterme-sft base: caps re-solved to convert the pool's unused 1.5B tokens (+542M vs v1.6); longctx lane grows again (241 mega sessions in-final, 374M gpt2); cad_evo +1,620 tight-band sessions; fresh repo, no legacy build trees. |
+## What's new in v1.7
+- **the 2B conversion** — the v1.6 pool held ~3.1B real gpt2 tokens but the releases only consumed 1.46B (quota was cap-constrained, not pool-constrained). v1.7 re-solved the caps against the exact post-pareto pool (`scripts/design_caps_v17.py`: bisection on the elastic-family scale, validated with the production `quota_keep`) and converted the headroom: **2,004,534,369 real gpt2 tokens (+542M vs v1.6)** with every domain family still inside its tolerance band (max violation 1.25pp).
+- **longctx grows again** — +150 fresh mega sessions (seed base 40000+) join the 301 carried; **241 survive in-final (374M gpt2 tokens)** — hundred-class turn counts over REAL executed state: tools run, files mutate, every citation checked against prior tool output. These are the rows that teach long-horizon coherence.
+- **cad_evo tight-band growth** — +1,620 new evolution sessions at the 0.25% / 0.10 mm verify band (41% real repair loops, 204+ arc families, 12/12 sampled re-exec inside the band). CAD and 3D keep their own slice of the mix with exec-verified only.
+- **clean base, new name** — this release lives at **IO-AIR/filterme-sft**: a fresh repo with root-level files only (`shards/`, `state/`, cards, finals). No `build/<version>/` history trees, no smoke leftovers, no stale cards — the accumulated 241-file legacy of sft-forge-ensemble stays behind.
+## Token mix by domain (real gpt2, kept rows)
+| domain | gpt2 tokens | share | |
+|---|---:|---:|---|
+| reasoning_math | 381,589,384 | 19.04% | `█████···················` |
+| agentic_swe | 359,611,343 | 17.94% | `████····················` |
+| reasoning_sci | 294,581,048 | 14.70% | `████····················` |
+| reverse_engineering | 284,560,356 | 14.20% | `███·····················` |
+| code_agent | 220,638,595 | 11.01% | `███·····················` |
+| agentic_tool | 219,243,982 | 10.94% | `███·····················` |
+| general_sft | 198,958,112 | 9.93% | `██······················` |
+| vision_svg | 40,415,056 | 2.02% | `························` |
+| cad_3d | 4,103,436 | 0.21% | `························` |
+| style_house | 325,961 | 0.02% | `························` |
+| vision_image | 294,309 | 0.01% | `························` |
+| vision_plot | 212,787 | 0.01% | `························` |
+Deep agentic trajectories dominate raw token share by design — they are the rows small models cannot synthesize. Rebalance a training mix by dropping or down-weighting tagged rows via the tag filter, never by re-splitting.
+## Sources (30)
+Kept rows/tokens sum exactly to the release totals; pool rows are the provenance shard sizes (`build/<version>/shards/`) before global dedup, pareto and quota.
+| source | kept rows | kept gpt2 | pool rows |
+|---|---:|---:|---:|
+| longctx_forge | 241 | 386,971,361 | 451 |
+| openmath_cot | 47,510 | 381,589,384 | 66,787 |
+| opensci | 49,751 | 294,581,048 | 65,779 |
+| toolmind | 13,139 | 171,025,036 | 13,351 |
+| ultradata_code | 2,764 | 163,636,053 | 4,200 |
+| ollvm_re | 704 | 115,095,930 | 704 |
+| nemotron_swe | 1,694 | 95,892,622 | 4,702 |
+| swesmith | 2,880 | 94,358,439 | 5,174 |
+| coderforge | 948 | 78,556,491 | 2,184 |
+| pyromind_long | 987 | 58,427,013 | 1,000 |
+| ultradata_search | 1,306 | 39,506,375 | 2,576 |
+| sophia_svg | 15,948 | 38,153,242 | 15,948 |
+| ultradata_general | 1,338 | 33,996,539 | 2,692 |
+| pyromind_short | 1,846 | 14,409,738 | 1,932 |
+| re_synth_forge | 5,100 | 12,075,013 | 6,059 |
+| ultradata_tooluse | 1,226 | 8,740,063 | 2,095 |
+| agenticsft | 500 | 8,115,596 | 500 |
+| cad_evo_forge | 3,593 | 3,389,137 | 5,820 |
+| icons_forge | 6,354 | 2,261,814 | 6,354 |
+| nemotron_cuda | 88 | 847,831 | 2,276 |
+| mcp_atlas | 31 | 729,868 | 462 |
+| cad_synth_forge | 3,204 | 709,604 | 3,204 |
+| demyagent | 228 | 628,420 | 500 |
+| vision_docmatix | 1,617 | 225,903 | 1,617 |
+| plot2code | 379 | 212,787 | 379 |
+| multiturn | 177 | 177,528 | 192 |
+| unified | 758 | 148,433 | 758 |
+| vision_forge | 1,232 | 68,406 | 1,232 |
+| cadforge_cq | 1 | 2,667 | 3 |
+| cadforge_3d | 1 | 2,028 | 14 |
+| **total** | **165,545** | **2,004,534,369** | 218,945 |
+## Pipeline
+1. **carry** vetted shards from the previous release (byte-faithful, verified against `pushed_sources.json`)
+2. **top-up fetch** — rows + parquet bulk with offset-cursor resume
+3. **per-row gates** — shape, non-Latin, think-balance, dup, license, no_image
+4. **ONE global dedup index** seeded with the carried corpus (Jaccard 0.65)
+5. **real gpt2 BPE count per row** — no estimates anywhere
+6. **strict per-slice pareto** — quality vs token cost; equal-correctness rows drop the bloated one first (concision envelopes)
+7. **domain token-share caps** — normalized-caps cascade, tolerance exit (`quota_keep`; see `quota_meta` in ENSEMBLE_CARD.json)
+8. **deterministic interleave** — same seed, same bytes, reproducible
+## QA gates (all green)
+| gate | scope | result |
+|---|---|---|
+| JSON parse | all 165,545 rows | 0 bad |
+| forbidden keys / bad ids | all rows | 0 / 0 |
+| SVG XML parse + viewbox + forbidden attrs | 22,302 vision_svg | 0 fail · 0 missing · 0 forbidden |
+| asm presence + length floor + session depth | 5,101 RE | 0 no-asm · 0 thin · 0 short |
+| cadquery + `result` contract | 3,204 CAD | 0 no-cq · 0 no-result |
+| longctx turn/size floors | 241 sessions | 0 short · 0 small |
+| domain caps (normalized, tolerance exit) | 12 families | max violation 0.0114 ≤ 0.0125 in 4 rounds |
+## License ledger (per-row `meta.license`)
+| license | rows |
+|---|---:|
+| cc-by-4.0 | 112,925 |
+| apache-2.0 | 41,533 |
+| mit | 8,952 |
+| isc | 925 |
+| cc-by-4.0 (per-row) | 223 |
+| apache-2.0/mit (per-row) | 52 |
+| house-generated (no upstream license) | 935 |
+| **total** | **165,545** |
+License doctrine: only Apache/MIT/CC-BY-4.0/ISC sources are adopted; sources governed by individual model providers' terms are excluded outright. Rows keep their per-row license in `meta.license`; carried rows keep their original vetting.
+## Files
+- `train.jsonl.gz` — the mixture (interleaved; shuffle again before packing)
+- `manifest.jsonl.gz` — per-row index: id, source, domain, turns, gpt2_tokens, tags
+- `shards/*.jsonl.gz` — per-source provenance pools
+- `state/` — progress, pushed_sources, qa_report
+- `images.tar.gz` — PNGs referenced by vision rows (`meta.images`)
+## Format
+```json
+{"id", "tags", "messages" (assistant turns loss-mask-ready via `trainable`), "approx_tokens", "gpt2_tokens", "meta"}
+```
+## Tag vocabulary (filter, never re-split)
+`src:<source>` · `domain:<fine>` (agentic_general, agentic_search, agentic_swe, agentic_tool, cad_3d, code_agent, general_sft, reasoning_math, reasoning_sci, reverse_engineering, style_house, vision_image, vision_plot, vision_svg) · `turns:<mono|short|mid|deep|mega>` · `tok:<1k|1k-5k|5k-20k|20k-50k|50k+>` · `tools` + `toolcalls:<1-3|4-10|10+>` · `cot` · `multimodal` · `verified`
+## Usage
+```python
+from datasets import load_dataset
+ds = load_dataset("IO-AIR/filterme-sft", split="train")  # or streaming=True
+# row: {id, tags, messages, trainable, approx_tokens, gpt2_tokens, meta}
+# loss masking: train only where trainable is True (assistant turns)
+# svg_rows = [r for r in ds if "domain:vision_svg" in r["tags"]]
+# mega    = [r for r in ds if "turns:mega" in r["tags"]]
+```
+## Related
+- **[IO-AIR/arenamax](https://huggingface.co/datasets/IO-AIR/arenamax)** — companion human-preference dataset (arena winners + rating-best-of-N), published separately with SFT and DPO configs.
+## Honest notes
+- Carried rows keep their earlier-release vetting; `meta.license` is per-row.
+- Carry-only sources: agenticsft / demyagent (upstream license unverified), cadforge_3d (upstream dark since v1.0 — the 14 verified rows are all that exist).
+- Vision rows' text token share is small by nature: the pixels live in `images.tar.gz`, outside any text token count. Every forge-rendered vision answer is program-verified upstream.
+- Every generated CAD row was volume/bbox-verified against an analytic exact answer; every RE row carries real asm from gcc/objdump.

qa_report.json ADDED Viewed

	@@ -0,0 +1,322 @@

+{
+ "root": "data/bigbuild/v1.7",
+ "gates": {
+  "structural": {
+   "rows": 165545,
+   "bad_json": 0,
+   "forbidden_keys": 0,
+   "bad_ids": 0
+  },
+  "domains": {
+   "reasoning_math": {
+    "rows": 47510,
+    "gpt2_tokens": 381589384,
+    "share": 0.19036
+   },
+   "agentic_swe": {
+    "rows": 4883,
+    "gpt2_tokens": 359611343,
+    "share": 0.1794
+   },
+   "reasoning_sci": {
+    "rows": 49751,
+    "gpt2_tokens": 294581048,
+    "share": 0.14696
+   },
+   "reverse_engineering": {
+    "rows": 5875,
+    "gpt2_tokens": 284560356,
+    "share": 0.14196
+   },
+   "code_agent": {
+    "rows": 3482,
+    "gpt2_tokens": 220638595,
+    "share": 0.11007
+   },
+   "agentic_tool": {
+    "rows": 16069,
+    "gpt2_tokens": 219243982,
+    "share": 0.10937
+   },
+   "general_sft": {
+    "rows": 4711,
+    "gpt2_tokens": 198958112,
+    "share": 0.09925
+   },
+   "vision_svg": {
+    "rows": 22302,
+    "gpt2_tokens": 40415056,
+    "share": 0.02016
+   },
+   "cad_3d": {
+    "rows": 6799,
+    "gpt2_tokens": 4103436,
+    "share": 0.00205
+   },
+   "style_house": {
+    "rows": 935,
+    "gpt2_tokens": 325961,
+    "share": 0.00016
+   },
+   "vision_image": {
+    "rows": 2849,
+    "gpt2_tokens": 294309,
+    "share": 0.00015
+   },
+   "vision_plot": {
+    "rows": 379,
+    "gpt2_tokens": 212787,
+    "share": 0.00011
+   }
+  },
+  "licenses": {
+   "-": 935,
+   "cc-by-4.0": 112925,
+   "apache-2.0": 41533,
+   "mit": 8952,
+   "isc": 925,
+   "cc-by-4.0 (per-row)": 223,
+   "apache-2.0/mit (per-row)": 52
+  },
+  "svg": {
+   "checked": 22302,
+   "xml_fail": 0,
+   "no_viewbox": 0,
+   "thin": 117,
+   "forbidden": 0,
+   "min_elements": 2
+  },
+  "reverse_engineering": {
+   "checked": 5101,
+   "no_asm": 0,
+   "thin_asm": 0,
+   "short_session": 0
+  },
+  "cad": {
+   "checked": 3204,
+   "no_cq": 0,
+   "no_result": 0
+  },
+  "longctx": {
+   "checked": 241,
+   "short_turns": 0,
+   "small": 0
+  },
+  "concision": {
+   "style_house": {
+    "p50": 168,
+    "p95": 1258,
+    "n": 935
+   },
+   "reasoning_sci": {
+    "p50": 3755,
+    "p95": 20473,
+    "n": 49751
+   },
+   "reverse_engineering": {
+    "p50": 2373,
+    "p95": 70949,
+    "n": 5875
+   },
+   "reasoning_math": {
+    "p50": 7332,
+    "p95": 16579,
+    "n": 47510
+   },
+   "agentic_tool": {
+    "p50": 2984,
+    "p95": 78328,
+    "n": 16069
+   },
+   "code_agent": {
+    "p50": 57727,
+    "p95": 129890,
+    "n": 3482
+   },
+   "vision_plot": {
+    "p50": 449,
+    "p95": 1289,
+    "n": 379
+   },
+   "cad_3d": {
+    "p50": 749,
+    "p95": 1092,
+    "n": 6799
+   },
+   "general_sft": {
+    "p50": 21625,
+    "p95": 75135,
+    "n": 4711
+   },
+   "vision_image": {
+    "p50": 67,
+    "p95": 272,
+    "n": 2849
+   },
+   "agentic_swe": {
+    "p50": 35073,
+    "p95": 122673,
+    "n": 4883
+   },
+   "vision_svg": {
+    "p50": 1693,
+    "p95": 4192,
+    "n": 22302
+   }
+  },
+  "diversity": {
+   "unified": {
+    "tasks": 1,
+    "top_task": "style",
+    "top_share": 1.0
+   },
+   "opensci": {
+    "tasks": 2,
+    "top_task": "science",
+    "top_share": 0.574
+   },
+   "re_synth_forge": {
+    "tasks": 2917,
+    "top_task": "rle",
+    "top_share": 0.022
+   },
+   "openmath_cot": {
+    "tasks": 6,
+    "top_task": "math",
+    "top_share": 0.55
+   },
+   "ultradata_tooluse": {
+    "tasks": 1,
+    "top_task": "tooluse",
+    "top_share": 1.0
+   },
+   "ultradata_code": {
+    "tasks": 2685,
+    "top_task": "coding",
+    "top_share": 0.029
+   },
+   "plot2code": {
+    "tasks": 2,
+    "top_task": "vision",
+    "top_share": 0.971
+   },
+   "cad_synth_forge": {
+    "tasks": 23,
+    "top_task": "plate_holes",
+    "top_share": 0.096
+   },
+   "ultradata_general": {
+    "tasks": 1235,
+    "top_task": "general",
+    "top_share": 0.078
+   },
+   "vision_docmatix": {
+    "tasks": 2,
+    "top_task": "vision",
+    "top_share": 0.883
+   },
+   "swesmith": {
+    "tasks": 1293,
+    "top_task": "swe",
+    "top_share": 0.415
+   },
+   "icons_forge": {
+    "tasks": 9,
+    "top_task": "tabler-outline",
+    "top_share": 0.415
+   },
+   "sophia_svg": {
+    "tasks": 1,
+    "top_task": "text2svg",
+    "top_share": 1.0
+   },
+   "pyromind_short": {
+    "tasks": 2,
+    "top_task": "tooluse",
+    "top_share": 0.567
+   },
+   "cad_evo_forge": {
+    "tasks": 298,
+    "top_task": "box_stack_pocket_boss",
+    "top_share": 0.146
+   },
+   "toolmind": {
+    "tasks": 2,
+    "top_task": "tooluse",
+    "top_share": 0.543
+   },
+   "ultradata_search": {
+    "tasks": 1306,
+    "top_task": "Search_Agent_002436",
+    "top_share": 0.001
+   },
+   "vision_forge": {
+    "tasks": 1,
+    "top_task": "vision",
+    "top_share": 1.0
+   },
+   "coderforge": {
+    "tasks": 719,
+    "top_task": "swe",
+    "top_share": 0.243
+   },
+   "agenticsft": {
+    "tasks": 1,
+    "top_task": "agent",
+    "top_share": 1.0
+   },
+   "nemotron_swe": {
+    "tasks": 1560,
+    "top_task": "swe",
+    "top_share": 0.08
+   },
+   "demyagent": {
+    "tasks": 1,
+    "top_task": "agent",
+    "top_share": 1.0
+   },
+   "ollvm_re": {
+    "tasks": 2,
+    "top_task": "re",
+    "top_share": 0.999
+   },
+   "pyromind_long": {
+    "tasks": 1,
+    "top_task": "tooluse",
+    "top_share": 1.0
+   },
+   "longctx_forge": {
+    "tasks": 14,
+    "top_task": "pycore v1.2 regression fix",
+    "top_share": 0.328
+   },
+   "multiturn": {
+    "tasks": 1,
+    "top_task": "style",
+    "top_share": 1.0
+   },
+   "nemotron_cuda": {
+    "tasks": 1,
+    "top_task": "cuda",
+    "top_share": 1.0
+   },
+   "mcp_atlas": {
+    "tasks": 1,
+    "top_task": "mcp",
+    "top_share": 1.0
+   },
+   "cadforge_3d": {
+    "tasks": 1,
+    "top_task": "threedee",
+    "top_share": 1.0
+   },
+   "cadforge_cq": {
+    "tasks": 1,
+    "top_task": "cadquery",
+    "top_share": 1.0
+   }
+  }
+ },
+ "rows": 165545,
+ "gpt2_tokens_total": 2004534369
+}