Update Croissant files
Browse files
metadata/CROISSANT_VALIDATION_NOTES.md
ADDED
|
@@ -0,0 +1,61 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# Temporal Twins Croissant Validation Notes
|
| 2 |
+
|
| 3 |
+
## 1. How to Validate
|
| 4 |
+
|
| 5 |
+
Use the official MLCommons Croissant tooling after the dataset release files are hosted.
|
| 6 |
+
|
| 7 |
+
1. Confirm the hosted dataset and code repository URLs in `metadata/temporal_twins_croissant.json` are correct for the current release.
|
| 8 |
+
2. Validate the file with the official Croissant validator from the MLCommons Croissant project. If you use the web validator, upload the final JSON-LD file or point it at the hosted Croissant URL.
|
| 9 |
+
3. As a local smoke check, you can also load the JSON-LD with a JSON parser before running the full validator:
|
| 10 |
+
|
| 11 |
+
```bash
|
| 12 |
+
python3 - <<'PY'
|
| 13 |
+
import json
|
| 14 |
+
from pathlib import Path
|
| 15 |
+
path = Path("metadata/temporal_twins_croissant.json")
|
| 16 |
+
with path.open() as f:
|
| 17 |
+
json.load(f)
|
| 18 |
+
print("JSON parse OK")
|
| 19 |
+
PY
|
| 20 |
+
```
|
| 21 |
+
|
| 22 |
+
4. After JSON parsing succeeds, run the official Croissant validation step and confirm the record sets, fields, and distribution references resolve correctly.
|
| 23 |
+
|
| 24 |
+
## 2. Hosted URLs and Remaining Placeholders
|
| 25 |
+
|
| 26 |
+
Dataset-side URLs now resolve to:
|
| 27 |
+
|
| 28 |
+
- Dataset URL: `https://huggingface.co/datasets/temporal-twins-benchmark/temporal-twins`
|
| 29 |
+
- Croissant metadata URL: `https://huggingface.co/datasets/temporal-twins-benchmark/temporal-twins/raw/main/metadata/temporal_twins_croissant.json`
|
| 30 |
+
- Croissant metadata browser page: `https://huggingface.co/datasets/temporal-twins-benchmark/temporal-twins/blob/main/metadata/temporal_twins_croissant.json`
|
| 31 |
+
- Data URL: `https://huggingface.co/datasets/temporal-twins-benchmark/temporal-twins/tree/main/data`
|
| 32 |
+
- Results URL: `https://huggingface.co/datasets/temporal-twins-benchmark/temporal-twins/tree/main/results`
|
| 33 |
+
- Configs URL: `https://huggingface.co/datasets/temporal-twins-benchmark/temporal-twins/tree/main/configs`
|
| 34 |
+
- Metadata URL: `https://huggingface.co/datasets/temporal-twins-benchmark/temporal-twins/tree/main/metadata`
|
| 35 |
+
- Release landing URL: `https://huggingface.co/datasets/temporal-twins-benchmark/temporal-twins`
|
| 36 |
+
|
| 37 |
+
Code repository URL:
|
| 38 |
+
|
| 39 |
+
- `https://huggingface.co/temporal-twins-benchmark/temporal-twins-code`
|
| 40 |
+
|
| 41 |
+
Paper URL status:
|
| 42 |
+
|
| 43 |
+
- Not available during double-blind review; to be added after publication.
|
| 44 |
+
|
| 45 |
+
## 3. Release Checklist
|
| 46 |
+
|
| 47 |
+
- Dataset URL is accessible to reviewers.
|
| 48 |
+
- Croissant file validates with the official MLCommons Croissant validator.
|
| 49 |
+
- Distribution URLs resolve to the intended hosted artifacts.
|
| 50 |
+
- Record-set columns match the actual hosted files.
|
| 51 |
+
- RAI fields are present.
|
| 52 |
+
- Dataset license is present (`CC-BY-4.0`).
|
| 53 |
+
- Code repository license is present (`Apache-2.0`).
|
| 54 |
+
|
| 55 |
+
## 4. Packaging Notes
|
| 56 |
+
|
| 57 |
+
- The Croissant file describes four dataset slices: `oracle_calib`, `easy`, `medium`, and `hard`.
|
| 58 |
+
- It assumes deterministic release seeds `0, 1, 2, 3, 4`.
|
| 59 |
+
- It assumes paper-suite configuration `num_users=350`, `simulation_days=45`, `fast_mode=false`, and `n_checkpoints=8`.
|
| 60 |
+
- The `matched_prefix_examples` record set uses the release-facing column name `matched_local_event_idx`.
|
| 61 |
+
- If the final hosted matched-pairs files keep the internal pipeline column name `eval_local_event_idx` instead, either rename that column in the export or update the Croissant metadata so the record-set field names match the hosted files exactly.
|
metadata/temporal_twins_croissant.json
CHANGED
|
@@ -1,5 +1,6 @@
|
|
| 1 |
{
|
| 2 |
"@context": {
|
|
|
|
| 3 |
"@vocab": "https://schema.org/",
|
| 4 |
"sc": "https://schema.org/",
|
| 5 |
"cr": "http://mlcommons.org/croissant/",
|
|
@@ -51,79 +52,42 @@
|
|
| 51 |
"reproducible benchmark"
|
| 52 |
],
|
| 53 |
"distribution": [
|
| 54 |
-
{
|
| 55 |
-
"@id": "transactions-archive",
|
| 56 |
-
"@type": "cr:FileObject",
|
| 57 |
-
"name": "Transactions archive",
|
| 58 |
-
"description": "Hosted archive containing synthetic transaction files for oracle_calib, easy, medium, and hard across seeds 0 through 4.",
|
| 59 |
-
"contentUrl": "https://huggingface.co/datasets/temporal-twins-benchmark/temporal-twins/tree/main/data",
|
| 60 |
-
"encodingFormat": "application/zip"
|
| 61 |
-
},
|
| 62 |
-
{
|
| 63 |
-
"@id": "matched-prefix-archive",
|
| 64 |
-
"@type": "cr:FileObject",
|
| 65 |
-
"name": "Matched-prefix examples archive",
|
| 66 |
-
"description": "Hosted release archive containing matched-prefix fraud/benign evaluation examples under release/data/*/seed_*/matched_pairs.parquet.",
|
| 67 |
-
"contentUrl": "https://huggingface.co/datasets/temporal-twins-benchmark/temporal-twins",
|
| 68 |
-
"encodingFormat": "application/zip"
|
| 69 |
-
},
|
| 70 |
-
{
|
| 71 |
-
"@id": "configs-archive",
|
| 72 |
-
"@type": "cr:FileObject",
|
| 73 |
-
"name": "Configs archive",
|
| 74 |
-
"description": "Hosted release archive containing benchmark configuration files under release/configs/*.yaml.",
|
| 75 |
-
"contentUrl": "https://huggingface.co/datasets/temporal-twins-benchmark/temporal-twins",
|
| 76 |
-
"encodingFormat": "application/zip"
|
| 77 |
-
},
|
| 78 |
-
{
|
| 79 |
-
"@id": "results-archive",
|
| 80 |
-
"@type": "cr:FileObject",
|
| 81 |
-
"name": "Results archive",
|
| 82 |
-
"description": "Hosted release archive containing the deterministic 5-seed paper-suite outputs under release/results/.",
|
| 83 |
-
"contentUrl": "https://huggingface.co/datasets/temporal-twins-benchmark/temporal-twins",
|
| 84 |
-
"encodingFormat": "application/zip"
|
| 85 |
-
},
|
| 86 |
{
|
| 87 |
"@id": "metadata-files",
|
| 88 |
"@type": "cr:FileSet",
|
| 89 |
"name": "Metadata files",
|
| 90 |
-
"description": "Metadata payload for the
|
| 91 |
-
"
|
| 92 |
-
"@id": "results-archive"
|
| 93 |
-
},
|
| 94 |
-
"includes": "release/metadata/*"
|
| 95 |
},
|
| 96 |
{
|
| 97 |
"@id": "transactions-files",
|
| 98 |
"@type": "cr:FileSet",
|
| 99 |
"name": "Synthetic transactions parquet files",
|
| 100 |
"description": "Expected synthetic transaction files for benchmark modes oracle_calib, easy, medium, and hard across seeds 0 through 4.",
|
| 101 |
-
"
|
| 102 |
-
"@id": "transactions-archive"
|
| 103 |
-
},
|
| 104 |
-
"includes": "release/data/*/seed_*/transactions.parquet",
|
| 105 |
"encodingFormat": "application/x-parquet"
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"@id": "matched-prefix-files",
|
| 109 |
"@type": "cr:FileSet",
|
| 110 |
"name": "Matched-prefix example parquet files",
|
| 111 |
-
"description": "Expected matched-prefix benchmark examples for the release. Each file contains fraud and benign twin examples evaluated at the same local prefix index.",
|
| 112 |
-
"
|
| 113 |
-
"@id": "matched-prefix-archive"
|
| 114 |
-
},
|
| 115 |
-
"includes": "release/data/*/seed_*/matched_pairs.parquet",
|
| 116 |
"encodingFormat": "application/x-parquet"
|
| 117 |
},
|
| 118 |
{
|
| 119 |
"@id": "config-files",
|
| 120 |
"@type": "cr:FileSet",
|
| 121 |
"name": "Benchmark config files",
|
| 122 |
-
"description": "YAML configuration files for the
|
| 123 |
-
"
|
| 124 |
-
|
| 125 |
-
|
| 126 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
| 127 |
},
|
| 128 |
{
|
| 129 |
"@id": "paper-suite-runs-csv",
|
|
@@ -131,9 +95,6 @@
|
|
| 131 |
"name": "Per-run paper-suite results",
|
| 132 |
"description": "Per-run deterministic results for the final 5-seed paper-scale suite.",
|
| 133 |
"contentUrl": "https://huggingface.co/datasets/temporal-twins-benchmark/temporal-twins/raw/main/results/paper_suite_runs.csv",
|
| 134 |
-
"containedIn": {
|
| 135 |
-
"@id": "results-archive"
|
| 136 |
-
},
|
| 137 |
"encodingFormat": "text/csv"
|
| 138 |
},
|
| 139 |
{
|
|
@@ -142,9 +103,6 @@
|
|
| 142 |
"name": "Paper-suite summary results",
|
| 143 |
"description": "Mean and standard deviation summary of the deterministic 5-seed paper suite.",
|
| 144 |
"contentUrl": "https://huggingface.co/datasets/temporal-twins-benchmark/temporal-twins/raw/main/results/paper_suite_summary.csv",
|
| 145 |
-
"containedIn": {
|
| 146 |
-
"@id": "results-archive"
|
| 147 |
-
},
|
| 148 |
"encodingFormat": "text/csv"
|
| 149 |
},
|
| 150 |
{
|
|
@@ -153,9 +111,6 @@
|
|
| 153 |
"name": "Paper-suite runtime summary",
|
| 154 |
"description": "Runtime and StaticGNN evaluation diagnostics for the final paper suite.",
|
| 155 |
"contentUrl": "https://huggingface.co/datasets/temporal-twins-benchmark/temporal-twins/raw/main/results/paper_suite_runtime.csv",
|
| 156 |
-
"containedIn": {
|
| 157 |
-
"@id": "results-archive"
|
| 158 |
-
},
|
| 159 |
"encodingFormat": "text/csv"
|
| 160 |
},
|
| 161 |
{
|
|
@@ -164,9 +119,6 @@
|
|
| 164 |
"name": "Paper-suite failed gate checks",
|
| 165 |
"description": "Gate-check and advisory-check outcomes for each run in the final paper suite.",
|
| 166 |
"contentUrl": "https://huggingface.co/datasets/temporal-twins-benchmark/temporal-twins/raw/main/results/paper_suite_failed_checks.csv",
|
| 167 |
-
"containedIn": {
|
| 168 |
-
"@id": "results-archive"
|
| 169 |
-
},
|
| 170 |
"encodingFormat": "text/csv"
|
| 171 |
},
|
| 172 |
{
|
|
@@ -175,9 +127,6 @@
|
|
| 175 |
"name": "Temporal Twins Croissant metadata",
|
| 176 |
"description": "MLCommons Croissant 1.1 metadata for the full Temporal Twins benchmark collection.",
|
| 177 |
"contentUrl": "https://huggingface.co/datasets/temporal-twins-benchmark/temporal-twins/raw/main/metadata/temporal_twins_croissant.json",
|
| 178 |
-
"containedIn": {
|
| 179 |
-
"@id": "metadata-files"
|
| 180 |
-
},
|
| 181 |
"encodingFormat": "application/ld+json"
|
| 182 |
}
|
| 183 |
],
|
|
|
|
| 1 |
{
|
| 2 |
"@context": {
|
| 3 |
+
"@language": "en",
|
| 4 |
"@vocab": "https://schema.org/",
|
| 5 |
"sc": "https://schema.org/",
|
| 6 |
"cr": "http://mlcommons.org/croissant/",
|
|
|
|
| 52 |
"reproducible benchmark"
|
| 53 |
],
|
| 54 |
"distribution": [
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 55 |
{
|
| 56 |
"@id": "metadata-files",
|
| 57 |
"@type": "cr:FileSet",
|
| 58 |
"name": "Metadata files",
|
| 59 |
+
"description": "Metadata payload for the hosted release, including this Croissant file and companion notes.",
|
| 60 |
+
"includes": "metadata/*"
|
|
|
|
|
|
|
|
|
|
| 61 |
},
|
| 62 |
{
|
| 63 |
"@id": "transactions-files",
|
| 64 |
"@type": "cr:FileSet",
|
| 65 |
"name": "Synthetic transactions parquet files",
|
| 66 |
"description": "Expected synthetic transaction files for benchmark modes oracle_calib, easy, medium, and hard across seeds 0 through 4.",
|
| 67 |
+
"includes": "data/*/seed_*/transactions.parquet",
|
|
|
|
|
|
|
|
|
|
| 68 |
"encodingFormat": "application/x-parquet"
|
| 69 |
},
|
| 70 |
{
|
| 71 |
"@id": "matched-prefix-files",
|
| 72 |
"@type": "cr:FileSet",
|
| 73 |
"name": "Matched-prefix example parquet files",
|
| 74 |
+
"description": "Expected matched-prefix benchmark examples for the hosted release. Each file contains fraud and benign twin examples evaluated at the same local prefix index.",
|
| 75 |
+
"includes": "data/*/seed_*/matched_pairs.parquet",
|
|
|
|
|
|
|
|
|
|
| 76 |
"encodingFormat": "application/x-parquet"
|
| 77 |
},
|
| 78 |
{
|
| 79 |
"@id": "config-files",
|
| 80 |
"@type": "cr:FileSet",
|
| 81 |
"name": "Benchmark config files",
|
| 82 |
+
"description": "YAML configuration files for the hosted release.",
|
| 83 |
+
"includes": "configs/*.yaml"
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"@id": "results-files",
|
| 87 |
+
"@type": "cr:FileSet",
|
| 88 |
+
"name": "Results files",
|
| 89 |
+
"description": "Hosted result summaries and diagnostics for the deterministic paper suite.",
|
| 90 |
+
"includes": "results/*"
|
| 91 |
},
|
| 92 |
{
|
| 93 |
"@id": "paper-suite-runs-csv",
|
|
|
|
| 95 |
"name": "Per-run paper-suite results",
|
| 96 |
"description": "Per-run deterministic results for the final 5-seed paper-scale suite.",
|
| 97 |
"contentUrl": "https://huggingface.co/datasets/temporal-twins-benchmark/temporal-twins/raw/main/results/paper_suite_runs.csv",
|
|
|
|
|
|
|
|
|
|
| 98 |
"encodingFormat": "text/csv"
|
| 99 |
},
|
| 100 |
{
|
|
|
|
| 103 |
"name": "Paper-suite summary results",
|
| 104 |
"description": "Mean and standard deviation summary of the deterministic 5-seed paper suite.",
|
| 105 |
"contentUrl": "https://huggingface.co/datasets/temporal-twins-benchmark/temporal-twins/raw/main/results/paper_suite_summary.csv",
|
|
|
|
|
|
|
|
|
|
| 106 |
"encodingFormat": "text/csv"
|
| 107 |
},
|
| 108 |
{
|
|
|
|
| 111 |
"name": "Paper-suite runtime summary",
|
| 112 |
"description": "Runtime and StaticGNN evaluation diagnostics for the final paper suite.",
|
| 113 |
"contentUrl": "https://huggingface.co/datasets/temporal-twins-benchmark/temporal-twins/raw/main/results/paper_suite_runtime.csv",
|
|
|
|
|
|
|
|
|
|
| 114 |
"encodingFormat": "text/csv"
|
| 115 |
},
|
| 116 |
{
|
|
|
|
| 119 |
"name": "Paper-suite failed gate checks",
|
| 120 |
"description": "Gate-check and advisory-check outcomes for each run in the final paper suite.",
|
| 121 |
"contentUrl": "https://huggingface.co/datasets/temporal-twins-benchmark/temporal-twins/raw/main/results/paper_suite_failed_checks.csv",
|
|
|
|
|
|
|
|
|
|
| 122 |
"encodingFormat": "text/csv"
|
| 123 |
},
|
| 124 |
{
|
|
|
|
| 127 |
"name": "Temporal Twins Croissant metadata",
|
| 128 |
"description": "MLCommons Croissant 1.1 metadata for the full Temporal Twins benchmark collection.",
|
| 129 |
"contentUrl": "https://huggingface.co/datasets/temporal-twins-benchmark/temporal-twins/raw/main/metadata/temporal_twins_croissant.json",
|
|
|
|
|
|
|
|
|
|
| 130 |
"encodingFormat": "application/ld+json"
|
| 131 |
}
|
| 132 |
],
|