Fix benchmark_low_bit_adam.py reference (#2287)

Signed-off-by: Emmanuel Ferdman <emmanuelferdman@gmail.com>

Author: emmanuel-ferdman

torchao/optim/README.md

@@ -32,7 +32,7 @@ NOTE:
 
 ## Benchmarks
 
-Fine-tune [timm](https://github.com/huggingface/pytorch-image-models)'s [ViT-H](https://huggingface.co/timm/vit_huge_patch14_224.orig_in21k) (630M params) on [resisc45](https://huggingface.co/datasets/timm/resisc45) dataset. PyTorch 2.4, BF16 AMP, compiled model, 1 epoch, batch size 8, cosine LR scheduler, 4070Ti SUPER, fixed random seed. Benchmark script is available at [benchmarks/benchmark_low_bit_adam.py](../../../benchmarks/benchmark_low_bit_adam.py).
+Fine-tune [timm](https://github.com/huggingface/pytorch-image-models)'s [ViT-H](https://huggingface.co/timm/vit_huge_patch14_224.orig_in21k) (630M params) on [resisc45](https://huggingface.co/datasets/timm/resisc45) dataset. PyTorch 2.4, BF16 AMP, compiled model, 1 epoch, batch size 8, cosine LR scheduler, 4070Ti SUPER, fixed random seed. Benchmark script is available at [benchmarks/benchmark_low_bit_adam.py](../../benchmarks/benchmark_low_bit_adam.py).
 
 AdamW impl      | Peak memory allocated (GB) | imgs/s | accuracy
 ----------------|----------------------------|--------|----------