ColabFold __ImportError

Downloading alphafold2_ptm weights to .: 100%|██████████| 3.47G/3.47G [00:24<00:00, 151MB/s]

ModuleNotFoundError Traceback (most recent call last)
/content/colabfold/batch.py in run(queries, result_dir, num_models, is_complex, num_recycles, recycle_early_stop_tolerance, model_order, initial_guess, num_ensemble, model_type, msa_mode, use_templates, custom_template_path, custom_template_cache_path, num_relax, relax_max_iterations, relax_tolerance, relax_stiffness, relax_max_outer_iterations, keep_existing_results, rank_by, pair_mode, pairing_strategy, data_dir, host_url, user_agent, random_seed, num_seeds, recompile_padding, zip_results, prediction_callback, save_single_representations, save_pair_representations, skip_output, jobname_prefix, save_all, save_recycles, use_dropout, use_gpu_relax, stop_at_score, dpi, max_seq, max_extra_seq, pdb_hit_file, local_pdb_path, use_cluster_profile, feature_dict_callback, calc_extra_ptm, use_probs_extra, max_template_date, max_template_hits, **kwargs)
1227 # check if TPU is available
-> 1228 from tpu_info import device
1229 if len(device.get_local_chips()) > 0:

ModuleNotFoundError: No module named 'tpu_info'

During handling of the above exception, another exception occurred:

ImportError Traceback (most recent call last)
12 frames
/usr/local/lib/python3.12/dist-packages/tensorflow/lite/python/metrics/wrapper/metrics_wrapper.py in
17 from tensorflow.compiler.mlir.lite.metrics import converter_error_data_pb2
18 from tensorflow.compiler.mlir.lite.python import wrap_converter
---> 19 from tensorflow.lite.python.metrics._pywrap_tensorflow_lite_metrics_wrapper import MetricsWrapper # pylint: disable=unused-import
20
21

ImportError: /usr/local/lib/python3.12/dist-packages/tensorflow/lite/python/metrics/_pywrap_tensorflow_lite_metrics_wrapper.so: undefined symbol: Wrapped_PyInit__pywrap_tensorflow_lite_metrics_wrapper

NOTE: If your import is failing due to a missing package, you can
manually install dependencies using either !pip or !apt.

To view examples of installing some common dependencies, click the
"Open Examples" button below.

Reproduce the failure in the Colab notebook, then inspect /content/colabfold/batch.py at the run entry point around line 1228 and the TensorFlow Lite metrics import shown in the traceback. Determine which environment or dependency condition causes the ImportError; done means the ColabFold run completes without this import failure.