From cb425024101d882b3672c8dca075eb1fdfd22f57 Mon Sep 17 00:00:00 2001 From: Billy Cao Date: Fri, 1 Jul 2022 21:19:13 +0800 Subject: [PATCH] Fix typo in perf_train_gpu_one.mdx (#17983) --- docs/source/en/perf_train_gpu_one.mdx | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/docs/source/en/perf_train_gpu_one.mdx b/docs/source/en/perf_train_gpu_one.mdx index 5b120391e..682ba663c 100644 --- a/docs/source/en/perf_train_gpu_one.mdx +++ b/docs/source/en/perf_train_gpu_one.mdx @@ -400,7 +400,7 @@ TrainingArguments(tf32=True) ``` By default the PyTorch default is used. -Note: tf32 mode is internal to CUDA and can't be accessed directly via `tensor.to(dtype=torch.tf32)` as `torch.tf32` doesn't exit. +Note: tf32 mode is internal to CUDA and can't be accessed directly via `tensor.to(dtype=torch.tf32)` as `torch.tf32` doesn't exist. Note: you need `torch>=1.7` to enjoy this feature.