pandas_dedupe: "AttributeError: 'float' object has no attribute 'keys'"

I'm trying to run a dedupe with pandas_dedupe but when running it it gives me this error "AttributeError: 'float' object has no attribute 'keys'"

import pandas as pd
from pandas_dedupe import dedupe_dataframe
df = pd.read_csv('/home/biminsal/Documentos/sampleDataDos.csv', sep=";")
dfinal = dedupe_dataframe(df, ['id', 'nombre_uno','apellido_uno'], canonicalize=True, sample_size=1)

Data sample:

    id  nombre_uno  apellido_uno
0   a001    Karlo   Perez
1   a002    Carlos  Perez
2   a003    Juan    Gomez
3   b001    Carlos  Perez
4   b002    Juan    Gomez

Error:

AttributeError                            Traceback (most recent call last)
...
File ~/.local/lib/python3.9/site-packages/dedupe/labeler.py:203, in BlockLearner._sample_indices(self, sample_size)
    201     sample_ids = ((keys[i][0], keys[i][1]) for i in sample_indices)
    202 else:
--> 203     sample_ids = weight.keys()
    205 return sample_ids

AttributeError: 'float' object has no attribute 'keys'

I have tried to find information and have applied several possible fixes, but the same problem remains. Thanks so much for any help.