Skip to content

Please add support for torch.tensor_split #2226

Description

@mallman
  • Name of layer type: torch.tensor_split
  • Is this a PyTorch or a TensorFlow layer type: PyTorch
  • Your version of coremltools: 7.2
  • Your version of PyTorch/TensorFlow: 2.3.0
  • Impact of supporting this layer type. Why is adding support for this layer type important? Is it necessary to support a popular model or use case?

This layer/op is used by EVA-02, a model for image classification, segmentation and object detection. Personally, I'm interested in using it for image classification in a Mac app.

As of this writing (May 21st, 2024), various sizes of pre-trained EVA and EVA-02 models dominate the leaderboard for image classification on ImageNet 1k among the models curated by the Pytorch Image Models Hugging Face org. See https://huggingface.co/collections/timm/timm-top-20-imagenet-1k-models-655d78909af37bae32381f61

FYI, it looks like this is (essentially) the same op as tf.split from TensorFlow.

Metadata

Metadata

Assignees

No one assigned

    Labels

    PyTorch (traced)missing layer typeUnable to convert a layer type from the relevant frameworktriagedReviewed and examined, release as been assigned if applicable (status)

    Type

    No type

    Fields

    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions