{"library":"ogb","title":"Open Graph Benchmark (OGB)","description":"A library for downloading, preprocessing, and evaluating on the Open Graph Benchmark (OGB) datasets. Version 1.3.6 fixes Pandas 2.0 compatibility. Releases are periodic, with occasional major dataset updates.","language":"python","status":"active","last_verified":"Fri May 01","install":{"commands":["pip install ogb"],"cli":null},"imports":["from ogb.nodeproppred import PygNodePropPredDataset","from ogb.nodeproppred import Evaluator"],"auth":{"required":false,"env_vars":[]},"quickstart":{"code":"from ogb.nodeproppred import PygNodePropPredDataset\nfrom ogb.nodeproppred import Evaluator\n\ndataset = PygNodePropPredDataset(name='ogbn-arxiv')\nsplit_idx = dataset.get_idx_split()\ntrain_idx, valid_idx, test_idx = split_idx['train'], split_idx['valid'], split_idx['test']\ngraph = dataset[0]\n\n# For evaluation, use the evaluator\nevaluator = Evaluator(name='ogbn-arxiv')\n# Example: dummy predictions and labels (for illustration only)\nimport torch\ny_pred = torch.randn(len(graph.y))\ny_true = graph.y\nresult = evaluator.eval({'y_pred': y_pred, 'y_true': y_true})\nprint(result)","lang":"python","description":"Load the ogbn-arxiv dataset using PyG wrapper and run a dummy evaluation.","tag":null,"tag_description":null,"last_tested":null,"results":[]},"compatibility":null}