ducha-aiki

I1223 12:53:51.375224 23647 caffe.cpp:134] Use GPU with device ID 0
I1223 12:53:51.488831 23647 net.cpp:42] Initializing net from parameters: 
name: "GoogleNet"
layers {
  top: "data"
  top: "label"
  name: "data"
  type: DATA
  data_param {
    source: "/home/share/storage/datasets/imagenet/dbs/ilsvrc12_val_lmdb"

ducha-aiki

I1229 14:22:43.806607 23733 caffe.cpp:134] Use GPU with device ID 0
I1229 14:22:44.181466 23733 net.cpp:42] Initializing net from parameters: 
name: "AlexNet"
layers {
  top: "data"
  top: "label"
  name: "data"
  type: DATA
  data_param {
    source: "/home/share/storage/datasets/imagenet/dbs/ilsvrc12_shai_val_lmdb"

ducha-aiki

I1021 21:37:50.680150  3320 caffe.cpp:184] Using GPUs 0
I1021 21:37:50.799264  3320 solver.cpp:47] Initializing solver from parameters: 
test_iter: 10
test_interval: 1000
base_lr: 0.001
display: 100
max_iter: 5000
lr_policy: "poly"
power: 0.5
momentum: 0.9

ducha-aiki

I1021 21:39:31.658416  3427 caffe.cpp:184] Using GPUs 0
I1021 21:39:31.932163  3427 solver.cpp:47] Initializing solver from parameters: 
test_iter: 10
test_interval: 1000
base_lr: 0.001
display: 100
max_iter: 5000
lr_policy: "poly"
power: 0.5
momentum: 0.9

ducha-aiki

layer {
  name: "data"
  type: "Input"
  top: "data"
  input_param { shape: { dim: 2 dim: 1 dim: 65 dim: 65 } }
}

 layer {
  name: "Gx"
  type: "Convolution"

ducha-aiki

name: "CIFAR10_full"
layer {
  name: "cifar"
  type: "Data"
  top: "data"
  top: "label"
  include {
    phase: TRAIN
  }
  transform_param {

ducha-aiki

import torch
import torch.nn.functional as F
from  torch.autograd import Variable

def batched_grid_apply(img, grid, batch_size):
    n_patches = len(grid)
    if n_patches > batch_size:
        bs = batch_size
        n_batches = n_patches / bs + 1
        for batch_idx in range(n_batches):

ducha-aiki

hb_setup();
%%
res = rproc.read('scoresroot', ...
  fullfile(hb_path, 'matlab', 'scores', 'default'));

norm_splits = {};
norms_path = fullfile(hb_path, 'matlab', 'data', 'best_normalizations.csv');
norms = readtable(norms_path, 'delimiter', ',');
norms.Properties.RowNames = norms.descriptor;

ducha-aiki

import torch

class CompactBilinearPooling(torch.nn.Module):
    def __init__(self, input_dim1, input_dim2, output_dim, sum_pool = True):
        super(CompactBilinearPooling, self).__init__()
        self.output_dim = output_dim
        self.sum_pool = sum_pool
        generate_sketch_matrix = lambda rand_h, rand_s, input_dim, output_dim: torch.sparse.FloatTensor(torch.stack([torch.arange(input_dim, out = torch.LongTensor()), rand_h.long()]), rand_s.float(), [input_dim, output_dim]).to_dense()
        self.sketch_matrix1 = torch.nn.Parameter(generate_sketch_matrix(torch.randint(output_dim, size = (input_dim1,)), 2 * torch.randint(2, size = (input_dim1,)) - 1, input_dim1, output_dim))
        self.sketch_matrix2 = torch.nn.Parameter(generate_sketch_matrix(torch.randint(output_dim, size = (input_dim2,)), 2 * torch.randint(2, size = (input_dim2,)) - 1, input_dim2, output_dim))

ducha-aiki

# The train/test net protocol buffer definition
net: "examples/cifar10/cifar10_full_sigmoid_train_test.prototxt"
# test_iter specifies how many forward passes the test should carry out.
# In the case of CIFAR10, we have test batch size 100 and 100 test iterations,
# covering the full 10,000 testing images.
test_iter: 100
# Carry out testing every 1000 training iterations.
test_interval: 1000
# The base learning rate, momentum and the weight decay of the network.
base_lr: 0.001