set index array dtype to int64

[bzrry]

i had the crash below due to indices being floats rather than ints. initializing the empty array to ints solved the issue.

Seed set to 12
/mnt/vcc-data/competition_support_set/{competition_train,k562_gwps,rpe1,jurkat,k562,hepg2}.h5
Dataset path does not exist: /mnt/vcc-data/competition_support_set/{competition_train,k562_gwps,rpe1,jurkat,k562,hepg2}.h5
Processed competition_train: 221273 train, 0 val, 0 test
Processing replogle_h1:  17%|███▌                 | 1/6 [00:00<00:01,  2.65it/s]No cell barcode information found in /mnt/vcc-data/competition_support_set/k562_gwps.h5. Generating generic barcodes.
Processed k562_gwps: 111605 train, 0 val, 0 test
Processing replogle_h1:  33%|███████              | 2/6 [00:00<00:00,  4.46it/s]No cell barcode information found in /mnt/vcc-data/competition_support_set/rpe1.h5. Generating generic barcodes.
Processed rpe1: 22317 train, 0 val, 0 test
Processing replogle_h1:  33%|███████              | 2/6 [00:00<00:00,  4.46it/s]No cell barcode information found in /mnt/vcc-data/competition_support_set/jurkat.h5. Generating generic barcodes.
Processed jurkat: 21412 train, 0 val, 0 test
Processing replogle_h1:  33%|███████              | 2/6 [00:00<00:00,  4.46it/s]No cell barcode information found in /mnt/vcc-data/competition_support_set/k562.h5. Generating generic barcodes.
Processed k562: 18465 train, 0 val, 0 test
Processing replogle_h1:  33%|███████              | 2/6 [00:00<00:00,  4.46it/s]No cell barcode information found in /mnt/vcc-data/competition_support_set/hepg2.h5. Generating generic barcodes.
Processed hepg2: 4976 train, 0 val, 9386 test
Processing replogle_h1: 100%|█████████████████████| 6/6 [00:00<00:00, 10.06it/s]
Traceback (most recent call last):
  File "/root/state/.venv/bin/state", line 8, in <module>
    sys.exit(main())
             ^^^^^^
  File "/root/state/src/state/__main__.py", line 120, in main
    run_tx_train(cfg)
  File "/root/state/src/state/_cli/_tx/_train.py", line 123, in run_tx_train
    dl = data_module.train_dataloader()
         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/state/.venv/lib/python3.11/site-packages/cell_load/data_modules/perturbation_dataloader.py", line 338, in train_dataloader
    return self._create_dataloader(self.train_datasets, test=test)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/state/.venv/lib/python3.11/site-packages/cell_load/data_modules/perturbation_dataloader.py", line 372, in _create_dataloader
    sampler = PerturbationBatchSampler(
              ^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/state/.venv/lib/python3.11/site-packages/cell_load/data_modules/samplers.py", line 83, in __init__
    self.sentences = self._create_sentences()
                     ^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/state/.venv/lib/python3.11/site-packages/cell_load/data_modules/samplers.py", line 257, in _create_sentences
    subset_batches = self._process_subset(global_offset, subset)
                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/root/state/.venv/lib/python3.11/site-packages/cell_load/data_modules/samplers.py", line 207, in _process_subset
    cell_codes = cache.cell_type_codes[indices]
                 ~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^
IndexError: arrays used as indices must be of integer (or boolean) type

[abhinadduri]

thanks for fixing @bzrry !