Errors trying to run Efficientdet.ipynb (spec_train.yaml compatibility)

Accelerated Computing Intelligent Video Analytics TAO Toolkit
1

I’m trying to run notebooks/tao_launcher_starter_kit/efficientdet_tf2 which I’ve cloned from github.

The notepad fails at the step ‘convert training data to TFRecords’ where it appears that spec_train.yaml isn’t compatible with the version of TAO toolkit I’m running (4.0.1)

# convert training data to TFRecords
!sed -i "s|DATADIR|$DATA_DOWNLOAD_DIR|g" $LOCAL_SPECS_DIR/spec_train_fixed.yaml
!tao efficientdet_tf2 dataset_convert -e $SPECS_DIR/spec_train_fixed.yaml \
    dataset_convert.results_dir=$DATA_DOWNLOAD_DIR

2025-12-12 16:42:16,823 [INFO] root: Registry: ['nvcr.io']
2025-12-12 16:42:16,906 [INFO] tlt.components.instance_handler.local_instance: Running command in container: nvcr.io/nvidia/tao/tao-toolkit:4.0.0-tf2.9.1
2025-12-12 16:42:22.322377: I tensorflow/core/util/util.cc:169] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`.
[1765557744.469493] [test-Precision-7780:34   :f]        vfs_fuse.c:424  UCX  WARN  failed to connect to vfs socket '��': Invalid argument
2025-12-12 16:42:24,707 [WARNING] matplotlib: Matplotlib created a temporary config/cache directory at /tmp/matplotlib-bto3ikrj because the default path (/.config/matplotlib) is not a writable directory; it is highly recommended to set the MPLCONFIGDIR environment variable to a writable directory, in particular to speed up the import of Matplotlib and to better support multiprocessing.
2025-12-12 16:42:24,953 [INFO] matplotlib.font_manager: generated new fontManager
<frozen cv.efficientdet.scripts.dataset_convert>:340: UserWarning: 
'spec_train_fixed.yaml' is validated against ConfigStore schema with the same name.
This behavior is deprecated in Hydra 1.1 and will be removed in Hydra 1.2.
See https://hydra.cc/docs/next/upgrades/1.0_to_1.1/automatic_schema_matching for migration instructions.
Error merging 'spec_train_fixed.yaml' with schema
Key 'train_tfrecords' not in 'TrainConfig'
    full_key: train.train_tfrecords
    reference_type=TrainConfig
    object_type=TrainConfig

This is running in Python 3.6.9 on Ubuntu 22 with RTX 5000. I’m trying to run the notepad with no modification.

2

Could you please docker pull the nvcr.io/nvidia/tao/tao-toolkit:6.0.0-tf2 and then docker run into it?
The corresponding notebook and spec files can be found in tao_tutorials/notebooks/tao_launcher_starter_kit/efficientdet_tf2 at main · NVIDIA/tao_tutorials · GitHub.

3

There is no update from you for a period, assuming this is not an issue anymore. Hence we are closing this topic. If need further support, please open a new one. Thanks.