Datasets:
docs(dataset): record the measured record mix and the no-verification caveat
Browse files
components/google-math/README.md
CHANGED
|
@@ -22,6 +22,11 @@ school-level difficulty. Those pairs are **not research mathematics**, and the s
|
|
| 22 |
projection of the paper's curriculum rather than the paper's dataset. Every record carries
|
| 23 |
`content_origin` so consumers can separate verbatim source text from `generator_output`.
|
| 24 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 25 |
## Included sources
|
| 26 |
|
| 27 |
Each source is pinned to a commit or dataset revision, and each ships math text under Apache-2.0
|
|
@@ -90,6 +95,7 @@ The short version:
|
|
| 90 |
measured yield.
|
| 91 |
- `audit/excluded-duplicates.jsonl`, `audit/packing.json`, `audit/source-scan.json`.
|
| 92 |
- `audit/validation-evidence.json`: the independent decode and round-trip evidence described below.
|
|
|
|
| 93 |
- `statistics.json`: measured records, groups, splits, characters and packed counts, per source.
|
| 94 |
- `manifest.json`: SHA256 and byte size for every delivered file except itself and validation.json.
|
| 95 |
- `packed/`: mmla-type2-packed-v1 arrays, manifest, row attribution and per-record token counts.
|
|
@@ -163,10 +169,11 @@ boundary masks, and re-encodes every record's token stream against `data/*.jsonl
|
|
| 163 |
`all_packed_tokens_reencoded_source_shift_padding_masks_and_attribution`, `statistics_match_artifacts`).
|
| 164 |
|
| 165 |
`tools/verify_google_math.py` is the independent check against the delivered files, and its output is
|
| 166 |
-
[`audit/validation-evidence.json`](audit/validation-evidence.json). It decoded 96 packed rows
|
| 167 |
-
|
| 168 |
-
the corresponding span of `[BOS] + SentencePiece(text) + [EOS]` for its
|
| 169 |
-
(b) confirmed the decoded text equals the decoded source text modulo
|
|
|
|
| 170 |
unified diffs of a decoded row against its source record; every diff had 0 lines, whitespace identical.
|
| 171 |
It round-tripped 3 Parquet rows (indices 0, 239,658 and 479,315) back to their JSONL records with exact
|
| 172 |
match, and confirmed a `formal_lean_proof`, a `formal_lean_statement`, a `generated_qa_pair`, a
|
|
@@ -174,20 +181,34 @@ match, and confirmed a `formal_lean_proof`, a `formal_lean_statement`, a `genera
|
|
| 174 |
matching `text_sha256`. Split separation was recomputed from the data rather than trusted: 3,403 groups
|
| 175 |
in the audit and 3,403 in the data, no group spanning two splits, train/dev/test at 3,333/35/35 groups.
|
| 176 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 177 |
## Publication
|
| 178 |
|
| 179 |
-
The component is
|
| 180 |
-
[MMLA-ORG/Cayley-Math discussion #
|
| 181 |
-
on ref `refs/pr/
|
| 182 |
-
|
| 183 |
-
`main`.
|
| 184 |
|
| 185 |
A token with `write` scope or org-admin rights is required to merge it. The token used for this build can
|
| 186 |
read the repository and open pull requests, but a direct commit to `main` is rejected with
|
| 187 |
-
`403 Forbidden: You have read access but not the required permissions for this operation`. Merging
|
| 188 |
pull request, or re-running `hf upload` without `--create-pr` once such a token is available, completes
|
| 189 |
-
the submission.
|
| 190 |
-
closed, because closing them was outside this task's scope.
|
| 191 |
|
| 192 |
## Limitations
|
| 193 |
|
|
|
|
| 22 |
projection of the paper's curriculum rather than the paper's dataset. Every record carries
|
| 23 |
`content_origin` so consumers can separate verbatim source text from `generator_output`.
|
| 24 |
|
| 25 |
+
**This corpus is mostly seeded generator output, not research mathematics.** `generated_qa_pair` records
|
| 26 |
+
dominate the record count, the formal subset is a small minority, and **no Lean compilation and no proof
|
| 27 |
+
verification was performed on any subset**, so no record here has a checked proof or a checked statement.
|
| 28 |
+
`statistics.json` carries the measured counts, and the component README states them in full.
|
| 29 |
+
|
| 30 |
## Included sources
|
| 31 |
|
| 32 |
Each source is pinned to a commit or dataset revision, and each ships math text under Apache-2.0
|
|
|
|
| 95 |
measured yield.
|
| 96 |
- `audit/excluded-duplicates.jsonl`, `audit/packing.json`, `audit/source-scan.json`.
|
| 97 |
- `audit/validation-evidence.json`: the independent decode and round-trip evidence described below.
|
| 98 |
+
- `audit/publish-check.json`: the second publish gate, measured from the delivered files after the build.
|
| 99 |
- `statistics.json`: measured records, groups, splits, characters and packed counts, per source.
|
| 100 |
- `manifest.json`: SHA256 and byte size for every delivered file except itself and validation.json.
|
| 101 |
- `packed/`: mmla-type2-packed-v1 arrays, manifest, row attribution and per-record token counts.
|
|
|
|
| 169 |
`all_packed_tokens_reencoded_source_shift_padding_masks_and_attribution`, `statistics_match_artifacts`).
|
| 170 |
|
| 171 |
`tools/verify_google_math.py` is the independent check against the delivered files, and its output is
|
| 172 |
+
[`audit/validation-evidence.json`](audit/validation-evidence.json). It decoded 96 packed rows, covering
|
| 173 |
+
68 of the 70 train shards plus the single dev and test shards, and for each row (a) confirmed every row
|
| 174 |
+
slice equals the re-encoding of the corresponding span of `[BOS] + SentencePiece(text) + [EOS]` for its
|
| 175 |
+
attributed source record and (b) confirmed the decoded text equals the decoded source text modulo
|
| 176 |
+
whitespace. It recorded 8 literal
|
| 177 |
unified diffs of a decoded row against its source record; every diff had 0 lines, whitespace identical.
|
| 178 |
It round-tripped 3 Parquet rows (indices 0, 239,658 and 479,315) back to their JSONL records with exact
|
| 179 |
match, and confirmed a `formal_lean_proof`, a `formal_lean_statement`, a `generated_qa_pair`, a
|
|
|
|
| 181 |
matching `text_sha256`. Split separation was recomputed from the data rather than trusted: 3,403 groups
|
| 182 |
in the audit and 3,403 in the data, no group spanning two splits, train/dev/test at 3,333/35/35 groups.
|
| 183 |
|
| 184 |
+
[`audit/publish-check.json`](audit/publish-check.json) is a second publish gate with its own code rather
|
| 185 |
+
than this component's tools. It agrees with every total above (480,584 records, 3,403 groups, 78,364,888
|
| 186 |
+
characters, and the packed counts for all three splits), decodes 12 packed rows across all three splits
|
| 187 |
+
and compares 263 attributed segments. All 263 row slices matched their canonical token slices by exact
|
| 188 |
+
identity, and all 248 segments that cover a whole record decoded to text identical to the source record
|
| 189 |
+
under whitespace normalisation and re-encoded to the identical token sequence. The other 15 segments are
|
| 190 |
+
row-boundary fragments that begin or end inside a SentencePiece piece, so they cannot re-encode
|
| 191 |
+
fragment-by-fragment; each one's decoded text was still located verbatim inside its attributed record.
|
| 192 |
+
It matched 5 of 5 Parquet round-trips exactly and recomputed group separation from `data/*.jsonl.gz`
|
| 193 |
+
directly, finding 0 groups spanning two splits.
|
| 194 |
+
|
| 195 |
+
The record mix is deliberately stated because it is skewed: `generated_qa_pair` records are 99.31% of
|
| 196 |
+
records but 45.57% of characters, while formal Lean records are 0.63% of records and 54.22% of characters,
|
| 197 |
+
and all non-generator records together are 0.69% of records. A consumer wanting formal or prose
|
| 198 |
+
mathematics should filter on `record_type` rather than read the corpus whole.
|
| 199 |
+
|
| 200 |
## Publication
|
| 201 |
|
| 202 |
+
The component is delivered as Hugging Face dataset pull request
|
| 203 |
+
[MMLA-ORG/Cayley-Math discussion #3](https://huggingface.co/datasets/MMLA-ORG/Cayley-Math/discussions/3)
|
| 204 |
+
on ref `refs/pr/3`. It carries the whole component plus the top-level `README.md` config and Collections
|
| 205 |
+
update, and it has not been merged into `main`. Consult that discussion for the current merge state.
|
|
|
|
| 206 |
|
| 207 |
A token with `write` scope or org-admin rights is required to merge it. The token used for this build can
|
| 208 |
read the repository and open pull requests, but a direct commit to `main` is rejected with
|
| 209 |
+
`403 Forbidden: You have read access but not the required permissions for this operation`. Merging the
|
| 210 |
pull request, or re-running `hf upload` without `--create-pr` once such a token is available, completes
|
| 211 |
+
the submission.
|
|
|
|
| 212 |
|
| 213 |
## Limitations
|
| 214 |
|
components/google-math/manifest.json
CHANGED
|
@@ -13,8 +13,8 @@
|
|
| 13 |
},
|
| 14 |
{
|
| 15 |
"path": "README.md",
|
| 16 |
-
"bytes":
|
| 17 |
-
"sha256": "
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"path": "audit/excluded-duplicates.jsonl",
|
|
@@ -1263,8 +1263,8 @@
|
|
| 1263 |
},
|
| 1264 |
{
|
| 1265 |
"path": "tools/build_google_math.py",
|
| 1266 |
-
"bytes":
|
| 1267 |
-
"sha256": "
|
| 1268 |
},
|
| 1269 |
{
|
| 1270 |
"path": "tools/generate_mathematics_dataset_sample.py",
|
|
|
|
| 13 |
},
|
| 14 |
{
|
| 15 |
"path": "README.md",
|
| 16 |
+
"bytes": 20593,
|
| 17 |
+
"sha256": "b3c98cc1005b5263dbe2bad37aafba811bf69ac6f268a918bcc851fe6f83af6d"
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"path": "audit/excluded-duplicates.jsonl",
|
|
|
|
| 1263 |
},
|
| 1264 |
{
|
| 1265 |
"path": "tools/build_google_math.py",
|
| 1266 |
+
"bytes": 61871,
|
| 1267 |
+
"sha256": "8e72001f8b24f5d188c73da7f4e9e29057300e5fa279e8d370f297375f6202d4"
|
| 1268 |
},
|
| 1269 |
{
|
| 1270 |
"path": "tools/generate_mathematics_dataset_sample.py",
|
components/google-math/tools/build_google_math.py
CHANGED
|
@@ -930,6 +930,11 @@ school-level difficulty. Those pairs are **not research mathematics**, and the s
|
|
| 930 |
projection of the paper's curriculum rather than the paper's dataset. Every record carries
|
| 931 |
`content_origin` so consumers can separate verbatim source text from `generator_output`.
|
| 932 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 933 |
## Included sources
|
| 934 |
|
| 935 |
Each source is pinned to a commit or dataset revision, and each ships math text under Apache-2.0
|
|
@@ -1075,19 +1080,53 @@ match, and confirmed a `formal_lean_proof`, a `formal_lean_statement`, a `genera
|
|
| 1075 |
matching `text_sha256`. Split separation was recomputed from the data rather than trusted: 3,403 groups
|
| 1076 |
in the audit and 3,403 in the data, no group spanning two splits, train/dev/test at 3,333/35/35 groups.
|
| 1077 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1078 |
## Limitations
|
| 1079 |
|
| 1080 |
- No Lean compilation, proof checking, Comparator run or mathematical verification of any kind.
|
|
|
|
|
|
|
|
|
|
| 1081 |
- PDF prose proofs are noisy pypdf text; symbol and ligature fidelity is not guaranteed.
|
| 1082 |
- The generator subset is a seeded sample of one curriculum, not the paper's full dataset, and its
|
| 1083 |
-
questions are school-level rather than research mathematics.
|
|
|
|
|
|
|
| 1084 |
- `formal-conjectures` upstream states that some content derives from third-party sources (Wikipedia,
|
| 1085 |
MathOverflow and OEIS under CC-BY-SA 4.0). Per-file reference URLs are retained inside the Lean text; no
|
| 1086 |
separate per-file license audit was performed.
|
| 1087 |
-
- Near-duplicate detection, Phase 1 overlap analysis and editorial review were not performed.
|
|
|
|
| 1088 |
- The SentencePiece model normalises runs of spaces, so packed text decodes whitespace-normalised rather
|
| 1089 |
than byte-identically to the sources.
|
| 1090 |
-
- Only problems that upstream published are present, so several subsets are selection-biased.
|
|
|
|
|
|
|
| 1091 |
|
| 1092 |
## Statistics
|
| 1093 |
|
|
|
|
| 930 |
projection of the paper's curriculum rather than the paper's dataset. Every record carries
|
| 931 |
`content_origin` so consumers can separate verbatim source text from `generator_output`.
|
| 932 |
|
| 933 |
+
**This corpus is mostly seeded generator output, not research mathematics.** `generated_qa_pair` records
|
| 934 |
+
dominate the record count, the formal subset is a small minority, and **no Lean compilation and no proof
|
| 935 |
+
verification was performed on any subset**, so no record here has a checked proof or a checked statement.
|
| 936 |
+
`statistics.json` carries the measured counts, and the component README states them in full.
|
| 937 |
+
|
| 938 |
## Included sources
|
| 939 |
|
| 940 |
Each source is pinned to a commit or dataset revision, and each ships math text under Apache-2.0
|
|
|
|
| 1080 |
matching `text_sha256`. Split separation was recomputed from the data rather than trusted: 3,403 groups
|
| 1081 |
in the audit and 3,403 in the data, no group spanning two splits, train/dev/test at 3,333/35/35 groups.
|
| 1082 |
|
| 1083 |
+
[`audit/publish-check.json`](audit/publish-check.json) is a third check written by an independent publish
|
| 1084 |
+
gate with its own code rather than this component's tools. It agrees with every total above (480,584
|
| 1085 |
+
records, 3,403 groups, 78,364,888 characters), decodes 12 packed rows covering 248 whole-record segments
|
| 1086 |
+
with 248/248 token-identical and whitespace-identical, matches 5/5 Parquet round-trips, and reports zero
|
| 1087 |
+
groups spanning two splits.
|
| 1088 |
+
|
| 1089 |
+
The record mix is deliberately stated because it is skewed: `generated_qa_pair` records are 99.31% of
|
| 1090 |
+
records but 45.57% of characters, while formal Lean and prose records are 0.69% of records and 54.43% of
|
| 1091 |
+
characters. A consumer wanting formal or prose mathematics should filter on `record_type` rather than
|
| 1092 |
+
read the corpus whole.
|
| 1093 |
+
|
| 1094 |
+
## Publication
|
| 1095 |
+
|
| 1096 |
+
The component is published as Hugging Face dataset pull request
|
| 1097 |
+
[MMLA-ORG/Cayley-Math discussion #5](https://huggingface.co/datasets/MMLA-ORG/Cayley-Math/discussions/5),
|
| 1098 |
+
on ref `refs/pr/5`, head commit `1894ffe2dc7f7f59abfb830f91e79738b1eb11d3`. It carries the whole
|
| 1099 |
+
component plus the top-level `README.md` config and Collections update, and it has not been merged into
|
| 1100 |
+
`main`.
|
| 1101 |
+
|
| 1102 |
+
A token with `write` scope or org-admin rights is required to merge it. The token used for this build can
|
| 1103 |
+
read the repository and open pull requests, but a direct commit to `main` is rejected with
|
| 1104 |
+
`403 Forbidden: You have read access but not the required permissions for this operation`. Merging that
|
| 1105 |
+
pull request, or re-running `hf upload` without `--create-pr` once such a token is available, completes
|
| 1106 |
+
the submission. Earlier pull requests #3 and #4 are superseded by #5 and were left open rather than
|
| 1107 |
+
closed, because closing them was outside this task's scope.
|
| 1108 |
+
|
| 1109 |
## Limitations
|
| 1110 |
|
| 1111 |
- No Lean compilation, proof checking, Comparator run or mathematical verification of any kind.
|
| 1112 |
+
- The build ran in a sandbox-free subprocess only for the generator; `cryptids`, `formal-conjectures`,
|
| 1113 |
+
`formal-imo`, `formal-putnam-like`, `alphageometry`, `alphaproof-nexus` and the notebooks were never
|
| 1114 |
+
built or executed, only read as text.
|
| 1115 |
- PDF prose proofs are noisy pypdf text; symbol and ligature fidelity is not guaranteed.
|
| 1116 |
- The generator subset is a seeded sample of one curriculum, not the paper's full dataset, and its
|
| 1117 |
+
questions are school-level rather than research mathematics. One module, `algebra__linear_1d`, stopped
|
| 1118 |
+
at 3,055 of its 8,000 quota when upstream's own `assert entropy == 0` fired; nothing was repaired, and
|
| 1119 |
+
the shortfall is recorded in `audit/mathematics-dataset-generation.json`.
|
| 1120 |
- `formal-conjectures` upstream states that some content derives from third-party sources (Wikipedia,
|
| 1121 |
MathOverflow and OEIS under CC-BY-SA 4.0). Per-file reference URLs are retained inside the Lean text; no
|
| 1122 |
separate per-file license audit was performed.
|
| 1123 |
+
- Near-duplicate detection, Phase 1 overlap analysis and editorial review were not performed. 1,629
|
| 1124 |
+
exact duplicate texts were dropped and are listed in `audit/excluded-duplicates.jsonl`.
|
| 1125 |
- The SentencePiece model normalises runs of spaces, so packed text decodes whitespace-normalised rather
|
| 1126 |
than byte-identically to the sources.
|
| 1127 |
+
- Only problems that upstream published are present, so several subsets are selection-biased. `cryptids`
|
| 1128 |
+
contributes 31.6M of the 78.4M characters because a handful of Busy Beaver traces are enormous, so the
|
| 1129 |
+
corpus is dominated by machine-generated proof script rather than mathematics prose.
|
| 1130 |
|
| 1131 |
## Statistics
|
| 1132 |
|