VSehwag

## Make sure to first download the model_best_dense.pth.tar from https://www.dropbox.com/sh/56yyfy16elwbnr8/AADmr7bXgFkrNdoHjKWwIFKqa?dl=0

import os
import argparse
import torch
import torch.nn as nn
import torchvision.datasets as datasets
import torch.utils.data as data
import torchvision.transforms as transforms
import torch.nn.functional as F

VSehwag

'''
A standalone PyTorch implementation for fast and efficient bicubic resampling.
Well suited for Pytorch-XLA, which doesn't support pytorch native bicubic resampling,
i.e., only nearest/bilinear are lowered for xla. As of now, bicubic resampling is very 
slow on TPUs with PyTorch XLA. 

This implementation dramatically reduces this overhead and (almost) makes it as fast as 
on a gpu.

## Hacked by:      Vikash Sehwag

VSehwag

""" ConvNeXt

Paper: `A ConvNet for the 2020s` - https://arxiv.org/pdf/2201.03545.pdf

Original code and weights from https://github.com/facebookresearch/ConvNeXt, original copyright below

Model defs atto, femto, pico, nano and _ols / _hnf variants are timm specific.

Modifications and additions for timm hacked together by / Copyright 2022, Ross Wightman