Time to Say Goodbye, torch 1.7 and 1.8 #22291
Conversation
There was a problem hiding this comment.
There is a bit more work to do ;-) The goal of pinning a minimum version is to remove all the code specific to older PyTorch versions.
In this case I haven't caught some PyTorch 1.7 specific code but there are a lot of PyTorch 1.8-specific code. In particular we can replace all instance of torch_int_div which was there to execute a different line depending on the version.
|
Thank you @sgugger . I will try to make a clean breakup |
|
The documentation is not available anymore as the PR was closed or merged. |
ydshieh
changed the title
ydshieh
changed the title
|
Hope I don't miss anything |
There was a problem hiding this comment.
Thanks a lot! Diving into the codebase a bit more, I see a test for PyTorch 1.8.0 in onnx.convert (and anything using TORCH_ONNX_DICT_INPUTS_MINIMUM_VERSION which is 1.8.0).
In modeling_utils there are also some require_version_core("torch>=1.9") which can now be removed.
|
ok, the deepspeed CI is running pt-1.8 - how do we solve that then? I have passed this change to the Deepspeed team let's see what they say. edit: they followed suit microsoft/DeepSpeed#3082 |
* time to say goodbye, torch 1.7 and 1.8 * clean up torch_int_div * clean up is_torch_less_than_1_8-9 * update --------- Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
* time to say goodbye, torch 1.7 and 1.8 * clean up torch_int_div * clean up is_torch_less_than_1_8-9 * update --------- Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
What does this PR do?
We have been together for more than 2 years ❤️
(see this discussion)