resnet docs

examples/DoReFa-Net/alexnet-dorefa.py

@@ -65,7 +65,9 @@ To Run Pretrained Model:
 BITW = 1
 BITA = 2
 BITG = 6
-BATCH_SIZE = 32
+TOTAL_BATCH_SIZE = 128
+NUM_GPU = 4
+BATCH_SIZE = TOTAL_BATCH_SIZE // NUM_GPU
 
 class Model(ModelDesc):
     def _get_input_vars(self):

examples/GAN/README.md

@@ -2,13 +2,13 @@
 
 Reproduce DCGAN following the setup in [dcgan.torch](https://github.com/soumith/dcgan.torch).
 
-Play with the [pretrained model](https://drive.google.com/drive/folders/0B9IPQTvr2BBkLUF2M0RXU1NYSkE?usp=sharing) on CelebA face dataset.
+Play with the [pretrained model](https://drive.google.com/drive/folders/0B9IPQTvr2BBkLUF2M0RXU1NYSkE?usp=sharing) on CelebA face dataset:
 
-Generated samples:
+1. Generated samples
 
 ![sample](demo/CelebA-samples.jpg)
 
-Vector arithmetic: smiling woman - neutral woman + neutral man = smiling man
+2. Vector arithmetic: smiling woman - neutral woman + neutral man = smiling man
 
 ![vec](demo/CelebA-vec.jpg)
 

examples/ResNet/README.md

@@ -3,7 +3,7 @@
 
 Training code of pre-activation ResNet on ImageNet. It follows the setup in
 [fb.resnet.torch](https://github.com/facebook/fb.resnet.torch) and gets similar performance (with much fewer lines of code).
-More results to come.
+Models can be [downloaded here](https://drive.google.com/open?id=0B9IPQTvr2BBkTXBlZmh1cmlnQ0k).
 
 | Model              | Top 5 Error | Top 1 Error |
 |:-------------------|-------------|------------:|
@@ -16,7 +16,7 @@ More results to come.
 
 ## load-resnet.py
 
-A script to convert and run ResNet{50,101,152} caffe models trained on ImageNet [released by Kaiming](https://github.com/KaimingHe/deep-residual-networks).
+A script to convert and run ImageNet-ResNet{50,101,152} Caffe models [released by Kaiming](https://github.com/KaimingHe/deep-residual-networks).
 
 Example usage:
 ```bash

tensorpack/tfutils/sessinit.py

@@ -134,7 +134,8 @@ class SaverRestore(SessionInit):
         if len(chkpt_vars_used) < len(vars_available):
             unused = vars_available - chkpt_vars_used
             for name in unused:
-                logger.warn("Variable {} in checkpoint not found in the graph!".format(name))
+                if not is_training_name(name):
+                    logger.warn("Variable {} in checkpoint not found in the graph!".format(name))
         return var_dict
 
 class ParamRestore(SessionInit):