Jean-Baptiste Cordonnier

Export to ONNX fails

1

Docker image `nvcr.io/nvidia/pytorch:23.09-py3` with installation of `transformer-engine==1.0.0+66d91d5`. Exporting a `TransformerLayer` to ONNX format fails. Minimal reproducible example below. Am I doing something wrong? ```python import transformer_engine import torch # #...

Loading dataset from large GCS bucket very slow since 2.14

1

### Describe the bug Since updating to >2.14 we have very slow access to our parquet files on GCS when loading a dataset (>30 min vs 3s). Our GCS bucket...

Very large lockfile and clean up

9

We noticed that our `uv.lock` file grows in size, up to 2.5MB and 35k lines with a lot of `resolution-markers` and the time to `uv lock` explodes to more than...