[bug?] [question] Confused about the params size of hash grid

[JackFishxxx]

Thank you for the great work on this project! I am currently using the code and have some questions regarding the storage size calculation of the hash grid.

I noticed some discrepancies in the expected and actual storage sizes. Could you please clarify how the storage size for the hash grid is calculated in your implementation? Specifically:

>>>hash_grid_config = {
    "otype": "Grid",
    "type": "Dense",
    "n_levels": 8,
    "n_features_per_level": 8,
    "base_resolution": 2
}
>>>hash_grid = tcnn.Encoding(
    n_input_dims=2,
    encoding_config=hash_grid_config
)
>>>hash_grid.state_dict()["params"].shape
>>>torch.Size([699072])

the param size is correct here: (256**2+128**2+64**2+32**2+16**2+8**2+4**2+2**2)*8 = 699040, since the coarest level with 2**2 features would be aligned with 8, so the final param size should be: (256**2+128**2+64**2+32**2+16**2+8**2+4**2+8)*8 = 699072

Another case is as following:

>>>hash_grid_config = {
    "otype": "Grid",
    "type": "Dense",
    "n_levels": 1,
    "n_features_per_level": 8,
    "base_resolution": 1
}
>>>hash_grid = tcnn.Encoding(
    n_input_dims=2,
    encoding_config=hash_grid_config
)
>>>hash_grid.state_dict()["params"].shape
>>>torch.Size([64])

the param size is: (1)*8=8, but the number of features should be aligned with 8, so the final param is 8*8=64

However, the following case seems to be incorrect:

>>>hash_grid_config = {
    "otype": "Grid",
    "type": "Dense",
    "n_levels": 2,
    "n_features_per_level": 8,
    "base_resolution": 2
}
>>>hash_grid = tcnn.Encoding(
    n_input_dims=2,
    encoding_config=hash_grid_config
)
>>>hash_grid.state_dict()["params"].shape
>>>torch.Size([524352])

The correct number of params should be: (2**2+4**2)*8=160, while the 2**2 would be aligned with 8, so the final param size would be: (8+4**2)*8=192, which is different from the real results.

It really confused me. Did I miss something? I appreciate any insight you can provide. Thanks in advance for your assistance!