Protocol esm1b: AttributeError: 'dict' object has no attribute 'startswith'

[dvonkan]

Metadata

key	value
version	0.2.3
cuda	True

Parameter

key	value
type	embed
protocol	esm1b
reduce	False
discard_per_amino_acid_embeddings	False

Traceback

Traceback (most recent call last):
  File "**(directory)**.cache/pypoetry/virtualenvs/biotrainer-8s_lNhFR-py3.10/lib/python3.10/site-packages/bio_embeddings/utilities/pipeline.py", line 322, in execute_pipeline_from_config
    stage_output_parameters = stage_runnable(**stage_parameters)
  File "**(directory)**.cache/pypoetry/virtualenvs/biotrainer-8s_lNhFR-py3.10/lib/python3.10/site-packages/bio_embeddings/embed/pipeline.py", line 377, in run
    embedder: EmbedderInterface = embedder_class(**result_kwargs)
  File "**(directory)**.cache/pypoetry/virtualenvs/biotrainer-8s_lNhFR-py3.10/lib/python3.10/site-packages/bio_embeddings/embed/esm_embedder.py", line 39, in __init__
    model, alphabet = load_model_and_alphabet_local(self._options["model_file"])
  File "**(directory)**.cache/pypoetry/virtualenvs/biotrainer-8s_lNhFR-py3.10/lib/python3.10/site-packages/bio_embeddings/embed/esm_embedder.py", line 23, in load_model_and_alphabet_local
    return load_model_and_alphabet_core(model_data, None)
  File "**(directory)**.cache/pypoetry/virtualenvs/biotrainer-8s_lNhFR-py3.10/lib/python3.10/site-packages/esm/pretrained.py", line 190, in load_model_and_alphabet_core
    if model_name.startswith("esm2"):
AttributeError: 'dict' object has no attribute 'startswith'

More info