FlagOpen / FlagGems

FlagGems is an operator library for large language models implemented in Triton Language.
Apache License 2.0
347 stars 48 forks source link

[Operator] Add upsample_nearest2d op [MooreThreads] #193

Closed ZaccurLi closed 1 month ago

ZaccurLi commented 3 months ago

PR Category

Operator upsample_nearest2d

Type of Change

New Feature

Description

https://github.com/pytorch/pytorch/blob/main/aten/src/ATen/native/native_functions.yaml#L12777C6-L12777C7 func: upsample_nearest2d(Tensor self, SymInt[2] output_size, float? scales_h=None, float? scales_w=None) -> Tensor

Issue

Progress

Performance

image

ZaccurLi commented 2 months ago

This PR has some common issue with PR #203

upsample_nearest2d fixed all issues mentioned in PR #203

ZaccurLi commented 1 month ago

PTAL @Bowen12992