pyarrow.lib.ArrowInvalid: Needed to copy 1 chunks with 78694 nulls, but zero_copy_only was True

[libingbuaa]

Errors when using https://huggingface.co/datasets/TAAC2025/TencentGR-1M as train data:

all the rows: 1001845
Loading Seqs: 0%| | 0/1001845 [00:00<?, ? rows/s]
Traceback (most recent call last):
File "/Users/Documents/code/TAAC_2025/baseline_2025/main.py", line 138, in
dataset = MyDataset(data_path, args)
File "/Users/Documents/code/TAAC_2025/baseline_2025/dataset.py", line 115, in init
self.full_events, self.user_indices = load_seq_as_list(self.data_dir / "seq")
File "/Users/Documents/code/TAAC_2025/baseline_2025/dataset.py", line 718, in load_seq_as_list
batch_events['action_type'] = flat_structs.field('action_type').to_numpy()
File "pyarrow/array.pxi", line 1787, in pyarrow.lib.Array.to_numpy
File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
pyarrow.lib.ArrowInvalid: Needed to copy 1 chunks with 78694 nulls, but zero_copy_only was True