anonymous-submission042 commited on
Commit
d4c3c2d
·
verified ·
1 Parent(s): 457e886

v1.1: drop question_zh and task_id; add hasSyntheticData + provenance fields

Browse files
README.md CHANGED
@@ -1,7 +1,6 @@
1
  ---
2
  language:
3
  - en
4
- - zh
5
  license:
6
  - cc-by-4.0
7
  - cc-by-nc-sa-4.0
@@ -74,10 +73,8 @@ Each row has the following fields:
74
  | `item_id` | string | Source schematic identifier |
75
  | `source` | string | Source class (`qucs` / `kicad` / `myriadrf` / `m17` / `oresat`) |
76
  | `level` | string | One of `Component Understanding` / `Structural Understanding` / `Functional Understanding` / `Dynamic Reasoning` |
77
- | `task_id` | string | Internal task type (e.g. `existence`, `pin_to_net`, `counterfactual_plot_choice`); useful for fine-grained analysis |
78
  | `category` | string | Coarse-grained tag |
79
  | `question` | string | English prompt (what models are evaluated on) |
80
- | `question_zh` | string | Chinese translation (parallel; not used for evaluation) |
81
  | `image` | PIL.Image | Primary schematic rendering (`image.png`) |
82
  | `context_images` | list of `{caption, image}` | Auxiliary context images (Dynamic Reasoning only — schematic plus baseline / variant simulation plots) |
83
  | `options` | list of `{label, text, image}` | Multi-choice options (Dynamic Reasoning only). Some options have only `text`, others have both `text` and `image`. |
@@ -121,7 +118,7 @@ For redistribution that requires fully permissive licensing, use only the `permi
121
 
122
  1. **Source-class size imbalance**: question counts per source class span 40–974; per-source claims should be reported with N.
123
  2. **Dynamic Reasoning scope**: only one source class has the simulation-grounded subset (55 questions). This dimension is reported as a small stress test, not the main result.
124
- 3. **Language**: questions are evaluated in English; Chinese parallel translations are provided for reference only.
125
  4. **Single-image protocol**: each question is paired with one primary schematic image (Dynamic Reasoning rows additionally provide context plots / option plots).
126
  5. **Anonymized release**: this submission account is for double-blind peer review. The dataset will be transferred to the official maintainer account upon acceptance.
127
 
 
1
  ---
2
  language:
3
  - en
 
4
  license:
5
  - cc-by-4.0
6
  - cc-by-nc-sa-4.0
 
73
  | `item_id` | string | Source schematic identifier |
74
  | `source` | string | Source class (`qucs` / `kicad` / `myriadrf` / `m17` / `oresat`) |
75
  | `level` | string | One of `Component Understanding` / `Structural Understanding` / `Functional Understanding` / `Dynamic Reasoning` |
 
76
  | `category` | string | Coarse-grained tag |
77
  | `question` | string | English prompt (what models are evaluated on) |
 
78
  | `image` | PIL.Image | Primary schematic rendering (`image.png`) |
79
  | `context_images` | list of `{caption, image}` | Auxiliary context images (Dynamic Reasoning only — schematic plus baseline / variant simulation plots) |
80
  | `options` | list of `{label, text, image}` | Multi-choice options (Dynamic Reasoning only). Some options have only `text`, others have both `text` and `image`. |
 
118
 
119
  1. **Source-class size imbalance**: question counts per source class span 40–974; per-source claims should be reported with N.
120
  2. **Dynamic Reasoning scope**: only one source class has the simulation-grounded subset (55 questions). This dimension is reported as a small stress test, not the main result.
121
+ 3. **Language**: questions are evaluated in English. (A Chinese parallel set was used internally during construction for human review but is not part of the released schema.)
122
  4. **Single-image protocol**: each question is paired with one primary schematic image (Dynamic Reasoning rows additionally provide context plots / option plots).
123
  5. **Anonymized release**: this submission account is for double-blind peer review. The dataset will be transferred to the official maintainer account upon acceptance.
124
 
croissant.json CHANGED
@@ -42,7 +42,10 @@
42
  "separator": "cr:separator",
43
  "source": "cr:source",
44
  "subField": "cr:subField",
45
- "transform": "cr:transform"
 
 
 
46
  },
47
  "@type": "sc:Dataset",
48
  "conformsTo": "http://mlcommons.org/croissant/1.0",
@@ -76,7 +79,7 @@
76
  "contentUrl": "data/permissive/test-00000-of-00001.parquet",
77
  "encodingFormat": "application/vnd.apache.parquet",
78
  "license": "https://creativecommons.org/licenses/by/4.0/",
79
- "sha256": "4dae49e1fae73019ec4e31b990f5c8622fa48f8ababd66f6a0aa04af9f44ca62"
80
  },
81
  {
82
  "@type": "cr:FileObject",
@@ -86,7 +89,7 @@
86
  "contentUrl": "data/nc_allowed/test-00000-of-00001.parquet",
87
  "encodingFormat": "application/vnd.apache.parquet",
88
  "license": "https://creativecommons.org/licenses/by-nc-sa/4.0/",
89
- "sha256": "42bbecbc878815e7f7c5cdda2bbeeceee3ecf9791302b424df8f5997e45e869f"
90
  }
91
  ],
92
  "recordSet": [
@@ -156,21 +159,6 @@
156
  }
157
  }
158
  },
159
- {
160
- "@type": "cr:Field",
161
- "@id": "permissive/task_id",
162
- "name": "task_id",
163
- "description": "Internal task type (e.g. existence, pin_to_net, counterfactual_plot_choice)",
164
- "dataType": "sc:Text",
165
- "source": {
166
- "fileObject": {
167
- "@id": "permissive-parquet"
168
- },
169
- "extract": {
170
- "column": "task_id"
171
- }
172
- }
173
- },
174
  {
175
  "@type": "cr:Field",
176
  "@id": "permissive/category",
@@ -201,21 +189,6 @@
201
  }
202
  }
203
  },
204
- {
205
- "@type": "cr:Field",
206
- "@id": "permissive/question_zh",
207
- "name": "question_zh",
208
- "description": "Chinese parallel translation",
209
- "dataType": "sc:Text",
210
- "source": {
211
- "fileObject": {
212
- "@id": "permissive-parquet"
213
- },
214
- "extract": {
215
- "column": "question_zh"
216
- }
217
- }
218
- },
219
  {
220
  "@type": "cr:Field",
221
  "@id": "permissive/image",
@@ -407,21 +380,6 @@
407
  }
408
  }
409
  },
410
- {
411
- "@type": "cr:Field",
412
- "@id": "nc_allowed/task_id",
413
- "name": "task_id",
414
- "description": "Internal task type (e.g. existence, pin_to_net, counterfactual_plot_choice)",
415
- "dataType": "sc:Text",
416
- "source": {
417
- "fileObject": {
418
- "@id": "nc_allowed-parquet"
419
- },
420
- "extract": {
421
- "column": "task_id"
422
- }
423
- }
424
- },
425
  {
426
  "@type": "cr:Field",
427
  "@id": "nc_allowed/category",
@@ -452,21 +410,6 @@
452
  }
453
  }
454
  },
455
- {
456
- "@type": "cr:Field",
457
- "@id": "nc_allowed/question_zh",
458
- "name": "question_zh",
459
- "description": "Chinese parallel translation",
460
- "dataType": "sc:Text",
461
- "source": {
462
- "fileObject": {
463
- "@id": "nc_allowed-parquet"
464
- },
465
- "extract": {
466
- "column": "question_zh"
467
- }
468
- }
469
- },
470
  {
471
  "@type": "cr:Field",
472
  "@id": "nc_allowed/image",
@@ -611,5 +554,42 @@
611
  "rai:dataBiases": "Source-class bias: the largest source class contributes ~41% of questions, which can dominate micro accuracy. We provide a `permissive` config that excludes the NonCommercial source and recommend reporting both micro and source-balanced macro accuracy. Domain bias: schematics are biased toward open-source SDR / amateur radio / satellite hardware; commercial proprietary RF designs are not represented.",
612
  "rai:dataReleaseMaintenancePlan": "Versioned releases under semver. Discussion thread is open for gold-answer corrections and parser / scorer disputes. After acceptance, this anonymous repository will be transferred to a maintainer account.",
613
  "rai:personalSensitiveInformation": "None. The dataset contains schematic images from open-source hardware projects, with no personally identifiable information, biometric data, or sensitive content.",
614
- "rai:dataSocialImpact": "Positive: enables systematic evaluation of multimodal LLMs on RF engineering tasks, supporting safer deployment in EE / RF design workflows. Risk: LLMs that perform well on this benchmark are not certified for production EE design. Use as evaluation tool, not deployment certification."
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
615
  }
 
42
  "separator": "cr:separator",
43
  "source": "cr:source",
44
  "subField": "cr:subField",
45
+ "transform": "cr:transform",
46
+ "prov": "http://www.w3.org/ns/prov#",
47
+ "wasDerivedFrom": "prov:wasDerivedFrom",
48
+ "wasGeneratedBy": "prov:wasGeneratedBy"
49
  },
50
  "@type": "sc:Dataset",
51
  "conformsTo": "http://mlcommons.org/croissant/1.0",
 
79
  "contentUrl": "data/permissive/test-00000-of-00001.parquet",
80
  "encodingFormat": "application/vnd.apache.parquet",
81
  "license": "https://creativecommons.org/licenses/by/4.0/",
82
+ "sha256": "8bb150251f50574342264ae3a973aff0d7fa74f8d6a9192bbd20ea7df468d866"
83
  },
84
  {
85
  "@type": "cr:FileObject",
 
89
  "contentUrl": "data/nc_allowed/test-00000-of-00001.parquet",
90
  "encodingFormat": "application/vnd.apache.parquet",
91
  "license": "https://creativecommons.org/licenses/by-nc-sa/4.0/",
92
+ "sha256": "4b1f90bf3436d76c1d95792e536f64fd0723bd5a1b97e52c0c923b9b9703d9c2"
93
  }
94
  ],
95
  "recordSet": [
 
159
  }
160
  }
161
  },
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
162
  {
163
  "@type": "cr:Field",
164
  "@id": "permissive/category",
 
189
  }
190
  }
191
  },
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
192
  {
193
  "@type": "cr:Field",
194
  "@id": "permissive/image",
 
380
  }
381
  }
382
  },
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
383
  {
384
  "@type": "cr:Field",
385
  "@id": "nc_allowed/category",
 
410
  }
411
  }
412
  },
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
413
  {
414
  "@type": "cr:Field",
415
  "@id": "nc_allowed/image",
 
554
  "rai:dataBiases": "Source-class bias: the largest source class contributes ~41% of questions, which can dominate micro accuracy. We provide a `permissive` config that excludes the NonCommercial source and recommend reporting both micro and source-balanced macro accuracy. Domain bias: schematics are biased toward open-source SDR / amateur radio / satellite hardware; commercial proprietary RF designs are not represented.",
555
  "rai:dataReleaseMaintenancePlan": "Versioned releases under semver. Discussion thread is open for gold-answer corrections and parser / scorer disputes. After acceptance, this anonymous repository will be transferred to a maintainer account.",
556
  "rai:personalSensitiveInformation": "None. The dataset contains schematic images from open-source hardware projects, with no personally identifiable information, biometric data, or sensitive content.",
557
+ "rai:dataSocialImpact": "Positive: enables systematic evaluation of multimodal LLMs on RF engineering tasks, supporting safer deployment in EE / RF design workflows. Risk: LLMs that perform well on this benchmark are not certified for production EE design. Use as evaluation tool, not deployment certification.",
558
+ "rai:hasSyntheticData": false,
559
+ "prov:wasDerivedFrom": [
560
+ {
561
+ "@type": "sc:Dataset",
562
+ "name": "Qucs-S example library (RF subset)",
563
+ "description": "Educational RF circuit examples shipped with Qucs-S; rendered to PNG via Qucs native renderer.",
564
+ "url": "https://github.com/ra3xdh/qucs_s"
565
+ },
566
+ {
567
+ "@type": "sc:Dataset",
568
+ "name": "KiCad public RF schematic projects",
569
+ "description": "Curated set of public GitHub KiCad projects with RF subsystems; rendered to PNG via KiCad CLI."
570
+ },
571
+ {
572
+ "@type": "sc:Dataset",
573
+ "name": "MyriadRF / LimeSDR hardware schematics",
574
+ "description": "Professional SDR board schematics released by Lime Microsystems and the MyriadRF community.",
575
+ "url": "https://github.com/myriadrf"
576
+ },
577
+ {
578
+ "@type": "sc:Dataset",
579
+ "name": "M17 Project digital radio hardware",
580
+ "description": "Open-source digital radio community hardware schematics (CC-BY-NC-SA-4.0 — used in nc_allowed config only).",
581
+ "url": "https://github.com/M17-Project"
582
+ },
583
+ {
584
+ "@type": "sc:Dataset",
585
+ "name": "OreSat satellite GPS module",
586
+ "description": "OreSat space hardware schematics covering RF / GPS subsystems.",
587
+ "url": "https://github.com/oresat"
588
+ }
589
+ ],
590
+ "prov:wasGeneratedBy": {
591
+ "@type": "prov:Activity",
592
+ "name": "RFSchemBench construction pipeline",
593
+ "description": "(1) Schematic source files from open-source repos rendered to PNG via KiCad CLI / Qucs native; (2) Page-level RF-relevance gate (Qwen3.6, used only as filter, not for gold answers); (3) Domain-expert-authored Python rules (`qa_pipelines/<source>/build_tier{1..4}_question_bank.py`) deterministically extract structured ground truth from KiCad netlist / Qucs schematic graph / ngspice simulation outputs and emit (question, gold) tuples; (4) Iterative rule refinement: experts review samples and patch rules upon discovering systematic errors."
594
+ }
595
  }
data/nc_allowed/test-00000-of-00001.parquet CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:42bbecbc878815e7f7c5cdda2bbeeceee3ecf9791302b424df8f5997e45e869f
3
- size 563110293
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4b1f90bf3436d76c1d95792e536f64fd0723bd5a1b97e52c0c923b9b9703d9c2
3
+ size 563068439
data/permissive/test-00000-of-00001.parquet CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:4dae49e1fae73019ec4e31b990f5c8622fa48f8ababd66f6a0aa04af9f44ca62
3
- size 541936367
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8bb150251f50574342264ae3a973aff0d7fa74f8d6a9192bbd20ea7df468d866
3
+ size 541895731