I encountered an error when trying to run it on Google Colab.

[koutaro777]

I was able to complete the Setup Environment section,
but I encountered an error in the Imports and Definitions section.

↓this occured error

/usr/local/lib/python3.12/dist-packages/pydub/utils.py:300: SyntaxWarning: invalid escape sequence '\('
  m = re.match('([su]([0-9]{1,2})p?) \(([0-9]{1,2}) bit\)$', token)
/usr/local/lib/python3.12/dist-packages/pydub/utils.py:301: SyntaxWarning: invalid escape sequence '\('
  m2 = re.match('([su]([0-9]{1,2})p?)( \(default\))?$', token)
/usr/local/lib/python3.12/dist-packages/pydub/utils.py:310: SyntaxWarning: invalid escape sequence '\('
  elif re.match('(flt)p?( \(default\))?$', token):
/usr/local/lib/python3.12/dist-packages/pydub/utils.py:314: SyntaxWarning: invalid escape sequence '\('
  elif re.match('(dbl)p?( \(default\))?$', token):
---------------------------------------------------------------------------
NotFoundError                             Traceback (most recent call last)
[/tmp/ipython-input-4203749083.py](https://localhost:8080/#) in <cell line: 0>()
     12 import librosa
     13 import note_seq
---> 14 import seqio
     15 import t5
     16 import t5x

11 frames
[/usr/local/lib/python3.12/dist-packages/seqio/__init__.py](https://localhost:8080/#) in <module>
     17 # pylint:disable=wildcard-import,g-bad-import-order,g-import-not-at-top
     18 
---> 19 from seqio.dataset_providers import *
     20 from seqio import evaluation
     21 from seqio import experimental

[/usr/local/lib/python3.12/dist-packages/seqio/dataset_providers.py](https://localhost:8080/#) in <module>
     38 from packaging import version as version_lib
     39 import pyglove as pg
---> 40 from seqio import metrics as metrics_lib
     41 from seqio import preprocessors as seqio_preprocessors
     42 from seqio import task_registry_provenance_tracking

[/usr/local/lib/python3.12/dist-packages/seqio/metrics.py](https://localhost:8080/#) in <module>
     25 import jax.numpy as jnp
     26 import numpy as np
---> 27 from seqio import utils
     28 import tensorflow.compat.v2 as tf
     29 

[/usr/local/lib/python3.12/dist-packages/seqio/utils.py](https://localhost:8080/#) in <module>
     29 from absl import logging
     30 import numpy as np
---> 31 from seqio.vocabularies import Vocabulary
     32 import tensorflow.compat.v2 as tf
     33 import tensorflow_datasets as tfds

[/usr/local/lib/python3.12/dist-packages/seqio/vocabularies.py](https://localhost:8080/#) in <module>
     24 from absl import logging
     25 import tensorflow.compat.v2 as tf
---> 26 import tensorflow_text as tf_text
     27 
     28 from sentencepiece import sentencepiece_model_pb2

[/usr/local/lib/python3.12/dist-packages/tensorflow_text/__init__.py](https://localhost:8080/#) in <module>
     19 # pylint: disable=wildcard-import
     20 from tensorflow_text.core.pybinds import tflite_registrar
---> 21 from tensorflow_text.python import keras
     22 from tensorflow_text.python import metrics
     23 from tensorflow_text.python.ops import *

[/usr/local/lib/python3.12/dist-packages/tensorflow_text/python/keras/__init__.py](https://localhost:8080/#) in <module>
     19 
     20 # pylint: disable=wildcard-import
---> 21 from tensorflow_text.python.keras.layers import *
     22 
     23 # Public symbols in the "tensorflow_text.layers" package.

[/usr/local/lib/python3.12/dist-packages/tensorflow_text/python/keras/layers/__init__.py](https://localhost:8080/#) in <module>
     20 # pylint: disable=wildcard-import
     21 from tensorflow_text.python.keras.layers.todense import *
---> 22 from tensorflow_text.python.keras.layers.tokenization_layers import *
     23 
     24 # Public symbols in the "tensorflow_text.layers" package.

[/usr/local/lib/python3.12/dist-packages/tensorflow_text/python/keras/layers/tokenization_layers.py](https://localhost:8080/#) in <module>
     22 from tensorflow.python.ops import lookup_ops
     23 from tensorflow.python.ops.ragged import ragged_conversion_ops
---> 24 from tensorflow_text.python.ops import unicode_script_tokenizer
     25 from tensorflow_text.python.ops import whitespace_tokenizer
     26 from tensorflow_text.python.ops import wordpiece_tokenizer

[/usr/local/lib/python3.12/dist-packages/tensorflow_text/python/ops/__init__.py](https://localhost:8080/#) in <module>
     24 from tensorflow_text.core.pybinds.pywrap_fast_wordpiece_tokenizer_model_builder import build_fast_wordpiece_model
     25 from tensorflow_text.python.ops.bert_tokenizer import BertTokenizer
---> 26 from tensorflow_text.python.ops.boise_offset_converter import boise_tags_to_offsets
     27 from tensorflow_text.python.ops.boise_offset_converter import offsets_to_boise_tags
     28 from tensorflow_text.python.ops.byte_splitter import ByteSplitter

[/usr/local/lib/python3.12/dist-packages/tensorflow_text/python/ops/boise_offset_converter.py](https://localhost:8080/#) in <module>
     30 from tensorflow.python.framework import load_library
     31 from tensorflow.python.platform import resource_loader
---> 32 gen_boise_offset_converter = load_library.load_op_library(resource_loader.get_path_to_datafile('_boise_offset_converter.so'))
     33 
     34 

[/usr/local/lib/python3.12/dist-packages/tensorflow/python/framework/load_library.py](https://localhost:8080/#) in load_op_library(library_filename)
     52     RuntimeError: when unable to load the library or get the python wrappers.
     53   """
---> 54   lib_handle = py_tf.TF_LoadLibrary(library_filename)
     55   try:
     56     wrappers = _pywrap_python_op_gen.GetPythonWrappers(

NotFoundError: /usr/local/lib/python3.12/dist-packages/tensorflow_text/python/ops/_boise_offset_converter.so: undefined symbol: _ZN6tflite4shim23TfShapeInferenceContextC1EPN10tensorflow15shape_inference16InferenceContextE

I would appreciate it if you could resolve this error.

[martinoywa]

Did you get a fix?

[Jurangren]

I have this error too

[koutaro777]

Yes, after struggling with the errors together with ChatGPT, I finally managed to get MT3 running. I rolled back the Colab environment to the July 2025 version and used a T4 GPU. I had to edit some of the code cells slightly compared to the official version, but with this setup, I think it should run properly. Please feel free to give it a try if you’d like.

That said, I don’t think this is a fundamental solution, so I would really appreciate it if someone with more expertise could provide advice on how to resolve the issue at its root.

Music_Transcription_with_Transformers.ipynb

[DFDSDFDDS-POOP]

I also get the same error message Any fixes?

[thiagorcdl]

I was able to get it working thanks to @koutaro777 's workaround.