fxmarty

@kapilsingh93 Can you share a reproduction in fp16?

@mht-sharma I'm not sure about this - actually we may be better off sticking with `GraphOptimizationLevel::ORT_ENABLE_ALL` during validation as this is the default of ORT, better to catch issues ahead?...

@Kaya-P probably not short term but happy to review contributions.

@BrightXiaoHan Could you add ernie as well to https://github.com/huggingface/optimum/blob/7e08a820b65a359a61444abe51df4eb96b26b2e3/tests/exporters/exporters_utils.py#L82 ? Also can you use https://huggingface.co/hf-internal-testing/tiny-random-ErnieModel instead of https://huggingface.co/hf-internal-testing/tiny-random-ErnieMModel? They appear to be two different model types in transformers and the...

Thank you @YooSungHyun and apologizes for the delay, I'll give a review on Monday!

Can you run `make style`? The check_code_quality workflow does not pass currently.

Hi @kazssym, Optimum ONNX export uses `model.onnx_data` for the external data file, contrary to other tools that use `model.onnx.data`. Hence the check on the Hub. Has this caused problems to...

Thank you @kazssym, I am not sure to understand. Could you open an issue about that with a reproduction of the issue? For example, ```bash optimum-cli export onnx --model saibo/llama-1B...

@kazssym Thanks, let me try.

Hi @JunyiYe thank you, could you share which behavior you expect, and which behavior you are getting?