zmasih

I'm running `python train.py --pipeline dali_cpu --epochs 1 --input_type tfrecord --train_file_pattern './tfrecords/train/*.tfrecord' --batch_size 16 --train_steps 1` For ` num_devices = 1` works well (no strategy is used), but for more...

@JanuszL Thank you for your answer. So, you are saying that if more than one device is available, with no explicit request, when running the code, DALI will use multiple...

I've tried ```python train.py --pipeline dali_cpu --epochs 1 --input_type tfrecord --train_file_pattern './tfrecords/train/*.tfrecord' --batch_size 16 --train_steps 1``` on a system with no GPU. I set `device_id=None` and `cpu_only=True` in `docs/examples/use_cases/tensorflow/efficientdet/pipeline/dali/efficientdet_pipeline.py` But...

Thank you @JanuszL That also didn't help. Since DALI is primarily designed for GPUs, I haven’t found much relevant guidance online. Would you have any additional resources or suggestions that...

I faced the same issue. Any updates?

> Same problem with me, it has been a week and I stuck with the dependencies issues. As far as I remember, that happened since Scenic’s older train_lib_deprecated was removed....