Added caffenet 4x wider

Complexity.md

@@ -14,10 +14,11 @@ Default augmentation: random crop 128x128 from 144xN image, 50% random horizonta
 
 | Name    | Accuracy      | LogLoss | Comments  |
 | -------|---------:| -------:|:-----------|
-| 2 wider| **0.533** | **2.04** | |
-| sqrt(2) wider | 0.506 | 2.17 | |
+| 4x wider | **0.563** | **1.92** | Still no overfitting %) | 
+| 2x wider| 0.533 | 2.04 | |
+| sqrt(2)x wider | 0.506 | 2.17 | |
 | Default | 0.471 | 2.36 | |
-| sqrt(2) narrower | 0.460 | 2.41 | |
+| sqrt(2)x narrower | 0.460 | 2.41 | |
 | 2x narrower |  0.416 | 2.68 | |
 
 

README.md

@@ -255,11 +255,11 @@ Others
 | [DarkNetBN](http://pjreddie.com/darknet/imagenet/#reference) |0.502| 2.25 | 16C3->MP2->32C3->MP2->64C3->MP2->128C3->MP2->256C3->MP2->512C3->MP2->1024C3->1000CLF.BN|
 | [HeNet2x2](http://arxiv.org/abs/1412.1710) |0.561 | 1.88|No SPP, Pool5 = 3x3, VLReLU, J' from paper |
 | [HeNet3x1](http://arxiv.org/abs/1412.1710) |0.560 | 1.88|No SPP, Pool5 = 3x3, VLReLU, J' from paper, 2x2->3x1 |
-| [GoogLeNet128](http://arxiv.org/abs/1409.4842) | **0.619** | **1.61** |linear lr_policy, batch_size=256. obviously slower than caffenet |
+| [GoogLeNet128](http://arxiv.org/abs/1409.4842) | 0.619 | 1.61 |linear lr_policy, batch_size=256. obviously slower than caffenet |
 | googlenet_loss2_clf| 0.571 | 1.80 | from net above, aux classifier after inception_4d |
 | googlenet_loss1_clf| 0.520 | 2.06 | from net above, aux classifier after inception_4a |
 | [fitnet1_elu](http://arxiv.org/abs/1412.6550) | 0.333 | 3.21 | |
-| [VGGNet16_128](https://arxiv.org/abs/1409.1556.pdf) |  0.651 | 1.46 |Surprisingly much better that GoogLeNet128, even later is with step-based solver. |
+| [VGGNet16_128](https://arxiv.org/abs/1409.1556.pdf) |  **0.651** | **1.46** |Surprisingly much better that GoogLeNet128, even later is with step-based solver. |
 
 
 ResNet attempts are moved to [ResNets.md](ResNets.md)
@@ -420,7 +420,8 @@ There difference in filters (main, 5x5 -> 3x3 + 3x3 or 1x5+5x1) and solver.
 
 | Name    | Accuracy      | LogLoss | Comments  |
 | -------|---------:| -------:|:-----------|
-| 2 wider| **0.533** | **2.04** | |
+| 4x wider | **0.563** | **1.92** | Still no overfitting %) |
+| 2 wider| 0.533 | 2.04 | |
 | sqrt(2) wider | 0.506 | 2.17 | |
 | Default | 0.471 | 2.36 | |
 | sqrt(2) narrower | 0.460 | 2.41 | |