CoreML PyTorch Conversion More Samples?

Question

Created Oct ’23

Replies 1

Boosts 0

Participants 1

I'm trying to convert a PyTorch forward Transformer model to CoreML but am running into several issues, like these errors:

"For mlprogram, inputs with infinite upper_bound is not allowed. Please set upper.

bound" 570 • to a positive value in "RangeDim)" for the "inputs" param in ct.convert().'

raise NotImplementedError ( 259 "inplace_ops pass doesn't yet support append op inside conditional"

Are there any more samples besides https://developer.apple.com/videos/play/tech-talks/10154

The sample in that video an imageType is used as input but in my model text is the input (and the output).

I also get warned that converting "torch script" is experimental but in the video it says it a torch script is required to convert (though I know the video is a few years old).

Boost

Answer 1

Macho Man Randy Savage OP

Nov ’23

I tried converting a pre-trained Tensorflow model to and it also spits out the error:

"For mlprogram, inputs with infinite upper_bound is not allowed. Please set upper_bound to a positive value in "RangeDim()" for the "inputs" param in st.convert().

Code:

model = TFAutoModel.from_pretrained("pretrainedmodename_here")

ctmodel = ct.convert(model,source="tensorflow")

I also tried downloading the .h5 file locally and feeding it to coremltools convert but that also fails with the same error. How do I set the upper_bound? I'm new to this. I have tested a .tflite version of the Tensorflow model with the TensorflowLite framework but I'd rather convert the model to CoreML and use the native APIs. If possible I'd like to avoid having to add such a large dependency (TFLite) to my project.

Thanks in advance.

0