Upload croissant.json with huggingface_hub

croissant.json

@@ -64,11 +64,34 @@
     "neural network"
   ],
   "license": "https://creativecommons.org/licenses/by-nc-sa/4.0/",
-  "prov:wasGeneratedBy": {
-    "@type": "prov:Activity",
-    "schema:description": "Three-stage automated pipeline: (1) lightweight ONNX export capturing computational graph topology without trained weights, (2) GPU profiling on an NVIDIA T4 measuring inference runtime, peak VRAM, and peak RAM, (3) PyG feature encoding converting ONNX graphs to PyTorch Geometric Data objects. Parametric source configurations (~89% of samples) sweep hyperparameters of six architecture families to generate dense scaling curves. HuggingFace source configurations profile model variants by model type.",
-    "schema:name": "NNIRP Dataset Generation Pipeline"
-  },
+  "prov:wasDerivedFrom": [
+    {
+      "@id": "https://huggingface.co/models",
+      "schema:name": "HuggingFace Model Hub",
+      "schema:description": "Pre-trained model architectures downloaded for profiling (149 HuggingFace source configurations, case IDs 900\u20131111)."
+    },
+    {
+      "schema:name": "Parametric Architecture Families",
+      "schema:description": "Six synthetic architecture families with dense hyperparameter sweeps (41 parametric source configurations, case IDs < 900). Models are constructed programmatically with randomly initialized weights."
+    }
+  ],
+  "prov:wasGeneratedBy": [
+    {
+      "@type": "prov:Activity",
+      "schema:name": "ONNX Export",
+      "schema:description": "Lightweight ONNX export of neural network models capturing computational graph topology (operators, shapes, data types) without trained weights."
+    },
+    {
+      "@type": "prov:Activity",
+      "schema:name": "GPU Profiling",
+      "schema:description": "Inference runtime and peak memory profiling on an NVIDIA T4 GPU across multiple repetitions per model configuration."
+    },
+    {
+      "@type": "prov:Activity",
+      "schema:name": "PyG Feature Encoding",
+      "schema:description": "Conversion of ONNX computational graphs to PyTorch Geometric Data objects with structured node, edge, and graph-level features."
+    }
+  ],
   "rai:dataBiases": [
     "Parametric source configurations account for ~22% of source configurations but ~89% of samples due to dense hyperparameter sweeps, creating a significant sample-mass imbalance between parametric and HuggingFace-derived data.",
     "All profiling was performed on a single GPU type (NVIDIA T4). Runtime predictions may not transfer to other hardware without re-profiling.",
@@ -91,7 +114,7 @@
   ],
   "rai:hasSyntheticData": true,
   "rai:personalSensitiveInformation": "The dataset contains no personal or sensitive information. All data consists of computational graph topologies, numerical profiling statistics, and engineered features derived from neural network architectures.",
-  "url": "https://huggingface.co/datasets/nnirp/NNIRP-dataset.",
+  "url": "https://huggingface.co/datasets/nnirp/NNIRP-dataset",
   "version": "1.0.0",
   "distribution": [
     {
@@ -99,7 +122,7 @@
       "@id": "hf-repository",
       "name": "hf-repository",
       "description": "HuggingFace dataset repository",
-      "contentUrl": "https://huggingface.co/datasets/nnirp/NNIRP-dataset./tree/main",
+      "contentUrl": "https://huggingface.co/datasets/nnirp/NNIRP-dataset/tree/main",
       "encodingFormat": "application/x-git+directory",
       "sha256": "0000000000000000000000000000000000000000000000000000000000000000"
     },
@@ -108,7 +131,7 @@
       "@id": "splits-manifest",
       "name": "splits-manifest",
       "description": "Canonical cluster-atomic train/val/test split definition. Maps each split to a list of source configuration IDs.",
-      "contentUrl": "https://huggingface.co/datasets/nnirp/NNIRP-dataset./resolve/main/manifests/splits.json",
+      "contentUrl": "https://huggingface.co/datasets/nnirp/NNIRP-dataset/resolve/main/manifests/splits.json",
       "encodingFormat": "application/json",
       "sha256": "15532da4e04a8acb5e7ab68cb698352555e293d3a9fbd6cb39a44ad364ad5e96"
     },
@@ -117,7 +140,7 @@
       "@id": "clusters-manifest",
       "name": "clusters-manifest",
       "description": "Cluster taxonomy mapping each cluster to its sub-family, family, and constituent source configuration IDs.",
-      "contentUrl": "https://huggingface.co/datasets/nnirp/NNIRP-dataset./resolve/main/manifests/clusters.json",
+      "contentUrl": "https://huggingface.co/datasets/nnirp/NNIRP-dataset/resolve/main/manifests/clusters.json",
       "encodingFormat": "application/json",
       "sha256": "1cefd174473b50ad921ed539f9704588bcf0ac91ead04709bd245c47dcf8595e"
     },