As a result of abusive and unreasonable anonymous traffic patterns, we're now requiring authentication to access some GitLab pages and content. We will investigate further to determine if these changes are permanent or temporary.
We apologize for the inconvenience caused.
Eclipse GitLab will be undergoing disruptive maintenance on June 15 at 9h00 CEST. Please see (and follow) this announcement: https://www.eclipsestatus.io/maintenance/911919
Since 1.19 and in particular https://github.com/onnx/onnx/issues/4554 onnx.mapping is removed. It was deprecated since 1.14.
In ort_inference:
np_dtype = onnx.mapping.TENSOR_TYPE_TO_NP_TYPE.get(dtype, np.float32)The code could be changed to:
onnx._mapping but it is private.helper.tensor_dtype_to_np_dtype() but it doesn't support default values. Is it useful?Copyright © Eclipse Foundation AISBL. All rights reserved. Privacy Policy | Terms of Use | Copyright Agent