Omar Alsaqa

@1025KB Thank you for your reply. Not working on TFX 1.6.0, with the following error: ```python Traceback (most recent call last): File "/usr/local/lib/python3.7/dist-packages/tfx/utils/import_utils.py", line 59, in import_func_from_module user_module = importlib.import_module(module_path)...

Thanks. Waiting.

@singhniraj08 Yes it is working now with: - TensorFlow version: 2.9.0 - TFX version: 1.9.1 - TensorFlow Data Validation version: 1.9.0 - TensorFlow Transform version: 1.10.1

I am not sure if this would help but you can pass anything to Transform using `custom_config`. see [docs](https://www.tensorflow.org/tfx/api_docs/python/tfx/v1/components/Transform) ```python transform = tfx.components.Transform( examples=example_gen.outputs['examples'], schema=schema_importer.outputs['result'], materialize=False, module_file=module_file, custom_config = {'size':...

I used the same idea (if i understood right) but in the Trainer component. In this [example](https://github.com/tensorflow/tfx/blob/master/tfx/examples/penguin/experimental/penguin_utils_sklearn.py#L49) I passed the Transformed Schema as an input to `run_fn`, `_input_fn` as: See,...

try ```python training_loop = training.Loop(model, train_task, eval_tasks=[eval_task], output_dir=output_dir) ``` NOTICE: `eval_tasks=[eval_task],` instead of `eval_task=eval_task,`

The eval_tasks before the equal sign in `eval_tasks=[eval_task],` is plural.

The problem is not related to the training loop or the eval task. The problem is in the shape of data given to your model. review you data preparation code....

Submit it on [Trax-ml community](https://gitter.im/trax-ml/community).

Do you happen to have any updates on this? I could not build it with 16GB of RAM (out of memory). I had to modify `make -j` (in `dgl_sparse/build.sh`, `graphbolt/build.sh`...