add aa result and download link for policy

README.md

@@ -1,14 +1,17 @@
 # Fast-AutoAug-Torch
 
-Search for [Fast AutoAugment](https://arxiv.org/abs/1905.00397) using [AutoTorch](http://autotorch.org/). [AutoAugment](https://arxiv.org/abs/1805.09501) and [RandAugment](https://arxiv.org/abs/1909.13719) are also implemented for comparison.
+Search for [Fast AutoAugment](https://arxiv.org/abs/1905.00397) using [AutoTorch](http://autotorch.org/). [AutoAugment](https://arxiv.org/abs/1805.09501) and [RandAugment](https://arxiv.org/abs/1909.13719) are also implemented for comparison. 
 
-| model | augment| epoch | Acc | Weights |
+This example will be used in the tutorial on [From HPO to NAS: Automated Deep Learning](https://hangzhang.org/CVPR2020/) at CVPR 2020.
+
+| model | augment| epoch | Acc | weights |
 |-------|--------|-------|-----|---------|
 |ResNet-50| baseline | 120 | 76.48 |
 |ResNet-50| AA | 120 | 76.66 |
 |ResNet-50| Fast AA | 120| 76.88 |
 |ResNet-50| Rand AA | 120 | 76.79 |
 |ResNet-50| baseline | 270| 77.17 |
+|ResNet-50| AA | 270 | 77.78 | [link](https://hangzh.s3-us-west-1.amazonaws.com/others/resnet50_aa-0cb27f8e.pth) |
 |ResNet-50| Fast AA | 270| 77.73 | [link](https://hangzh.s3-us-west-1.amazonaws.com/others/resnet50_fast_aa-3342410e.pth) |
 |ResNet-50| Rand AA | 270| **77.97** | [link](https://hangzh.s3-us-west-1.amazonaws.com/others/resnet50_rand_aug-e38097c7.pth) |
 
@@ -38,7 +41,7 @@ n=2, m=12,
 ``
 
 
-## Quick Start
+## Setup
 ### Install dependencies
 - PyTorch, please follow the [instructions](https://pytorch.org/get-started/locally/).
 
@@ -63,7 +66,10 @@ pip install -v --no-cache-dir --global-option="--cpp_ext" --global-option="--cud
 python prepare_imagenet.py --download-dir ./
 ```
 
+## Fast AutoAugment Training
 ### Search policy
+If you want to skip this step, you can download the searched policy [here](https://hangzh.s3-us-west-1.amazonaws.com/others/imagenet_policy.at).
+
 ```bash
 python search_policy.py --reduced-size 60000 --epochs 120 --nfolds 8 --num-trials 200  --save-policy imagenet_policy.at
 ```
@@ -76,7 +82,18 @@ python search_policy.py --reduced-size 60000 --epochs 120 --nfolds 8 --num-trial
 python train.py --dataset imagenet --model resnet50 --lr-scheduler cos --epochs 270 --checkname resnet50_fast_aa --lr 0.025 --batch-size 64 --auto-policy imagenet_policy.at
 ```
 
-### Train with random AA
+## Auto Augment and RandomAugment
+
+### Training with AutoAugment policy
+This repo only enables training with searched policy from the original paper. 
+
+```bash
+# download the policy
+wget https://hangzh.s3-us-west-1.amazonaws.com/others/aa_policy.at
+python train.py --dataset imagenet --model resnet50 --lr-scheduler cos --epochs 270 --checkname resnet50_fast_aa --lr 0.025 --batch-size 64 --auto-policy  aa_policy.at
+```
+
+### Training with RandomAugment
 ```bash
 python train.py --dataset imagenet --model resnet50 --lr-scheduler cos --epochs 270 --checkname resnet50_rand_aug --lr 0.025 --batch-size 64 --rand-aug
 ```