Craig Chan

TF to ONNX export fails with CLI using example from docs

2

### System Info - `transformers` version: 4.21.1 - Platform: Linux-4.15.0-187-generic-x86_64-with-debian-buster-sid - Python version: 3.7.5 - Huggingface_hub version: 0.8.1 - PyTorch version (GPU?): not installed (NA) - Tensorflow version (GPU?): 2.7.0...

TF to ONNX export fails with large models

4

### System Info - `transformers` version: 4.21.1 - Platform: Linux-4.15.0-187-generic-x86_64-with-debian-buster-sid - Python version: 3.7.5 - Huggingface_hub version: 0.8.1 - PyTorch version (GPU?): not installed (NA) - Tensorflow version (GPU?): 2.7.0...

tf.unstack / tf.split on input loses dynamic batch size

3

**Describe the bug** When `tf.unstack` or `tf.split` is used on an input, dynamic batch dimension is lost. This was found while trying to export [NodLabs's DLRM](https://github.com/NodLabs/tensorflow-dlrm). An example script is...

THC/THC.h: No such file or directory

7

It appears since PyTorch 1.11, TH/THC include files have been removed. `warp_rnnt` wasn't the only one affected. Other repos relying on these include files were as well (see https://github.com/open-mmlab/mmdetection3d/issues/1332). Python...

### 🐛 Describe the bug Not quite sure if this is expected behavior, but we are using `create_feature_extractor` on Swin-Tiny as the backbone for a model. I believe starting in...