DIsable to predict properties by fairseq-generate

After training on pcqm4mv2, I wanted to use checkpoint to predict properties. So I used
fairseq-generate $data_binarization --task graph_prediction --path ckpts/checkpoint_best.pt --num-classes 1
The bug was shown as:
Using backend: pytorch Traceback (most recent call last): File "/home/nv3090x2/下载/ENTER/envs/graphormer/bin/fairseq-generate", line 8, in <module> sys.exit(cli_main()) File "/home/nv3090x2/下载/ENTER/lib/python3.9/site-packages/fairseq_cli/generate.py", line 413, in cli_main main(args) File "/home/nv3090x2/下载/ENTER/lib/python3.9/site-packages/fairseq_cli/generate.py", line 50, in main return _main(cfg, sys.stdout) File "/home/nv3090x2/下载/ENTER/lib/python3.9/site-packages/fairseq_cli/generate.py", line 83, in _main task = tasks.setup_task(cfg.task) File "/home/nv3090x2/下载/ENTER/lib/python3.9/site-packages/fairseq/tasks/__init__.py", line 46, in setup_task return task.setup_task(cfg, **kwargs) File "/home/nv3090x2/下载/ENTER/lib/python3.9/site-packages/fairseq/tasks/graph_prediction.py", line 179, in setup_task return cls(cfg) File "/home/nv3090x2/下载/ENTER/lib/python3.9/site-packages/fairseq/tasks/graph_prediction.py", line 155, in __init__ self.dm = GraphormerDataset( File "/home/nv3090x2/下载/ENTER/lib/python3.9/site-packages/fairseq/data/dataset.py", line 84, in __init__ self.dataset = PYGDatasetLookupTable.GetPYGDataset(dataset_spec, seed=seed) File "/home/nv3090x2/下载/ENTER/lib/python3.9/site-packages/fairseq/data/pyg_datasets/pyg_dataset_lookup_table.py", line 102, in GetPYGDataset raise ValueError(f"Unknown dataset name {name} for pyg source.") ValueError: Unknown dataset name pcqm4m for pyg source.
I really hope you can help me with the bug. Thank you very much.

Reproduce the reported fairseq-generate command and traceback first. Trace setup through fairseq_cli/generate.py, fairseq/tasks/graph_prediction.py, and fairseq/data/pyg_datasets/pyg_dataset_lookup_table.py, focusing on why pcqm4m is rejected. Done means the documented prediction command no longer raises this dataset-name error and property prediction can proceed.