RESOLVI: using num_samples > 1 in get_normalized_expression fails

[abs51295]

I am trying to get normalized expression from resolVI model. If I pass num_samples > 1 it fails.

resolvae = RESOLVI.load(args.scvi_output_dir, adata=adata)

adata.layers['generated_counts'] = scp.csr_matrix(resolvae.get_normalized_expression(batch_size=5000, n_samples=30, library_size=10000))

Traceback (most recent call last):
  File "/common/knottsilab/MET_analysis/MET_analysis_2025/neoadjuvant_cagra_outputs/run_resolVI_downstream.py", line 91, in <module>
    adata.layers['generated_counts'] = scp.csr_matrix(resolvae.get_normalized_expression(batch_size=5000, n_samples=3, library_size=10000))
                                                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/shaha4/miniforge3/envs/rapids-24.12-stable/lib/python3.11/site-packages/torch/utils/_contextlib.py", line 116, in decorate_context
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "/home/shaha4/miniforge3/envs/rapids-24.12-stable/lib/python3.11/site-packages/scvi/external/resolvi/_utils.py", line 341, in get_normalized_expression
    return pd.DataFrame(
           ^^^^^^^^^^^^^
  File "/home/shaha4/miniforge3/envs/rapids-24.12-stable/lib/python3.11/site-packages/pandas/core/frame.py", line 827, in __init__
    mgr = ndarray_to_mgr(
          ^^^^^^^^^^^^^^^
  File "/home/shaha4/miniforge3/envs/rapids-24.12-stable/lib/python3.11/site-packages/pandas/core/internals/construction.py", line 336, in ndarray_to_mgr
    _check_values_indices_shape_match(values, index, columns)
  File "/home/shaha4/miniforge3/envs/rapids-24.12-stable/lib/python3.11/site-packages/pandas/core/internals/construction.py", line 420, in _check_values_indices_shape_match
    raise ValueError(f"Shape of passed values is {passed}, indices imply {implied}")
ValueError: Shape of passed values is (34338927, 380), indices imply (11446309, 380)

Versions:

1.3.0

[abs51295]

also, this might sound like a stupid question but is using this method equivalent to getting generated counts from sample_posterior?

 sample_quantile = 'post_sample_q25'

 samples_corr = resolvae.sample_posterior(
                   model=resolvae.module.model_corrected,
                   return_sites=['px_rate', 'obs'],
                   num_samples=30, return_samples=False, batch_size=6000)

samples_corr = pd.DataFrame(samples_corr).T

 adata.layers["generated_expression"] = scp.csr_matrix(samples_corr.loc[sample_quantile, 'obs'])```