enas

enas copied to clipboard

Hello Mr Does this source support to generate RNN, LSTM or CNN-LSTM(LRCN)?

ganav

Explanation of `cifar10.micro_child._factorized_reduction`

1

The docstring for `cifar10.micro_child._factorized_reduction` says ``` """Reduces the shape of x without information loss due to striding.""" ``` Could you explain what that means? When `stride=2`, ``` path1 = tf.nn.avg_pool(x,...

bkj

In [this](https://github.com/melodyguan/enas/blob/master/src/cifar10/micro_controller.py#L178-L185) part of the code: ```python logits = tf.matmul(next_h[-1], self.w_soft) + self.b_soft if self.temperature is not None: logits /= self.temperature if self.tanh_constant is not None: op_tanh = self.tanh_constant /...

karandwivedi42

Child Hyper Parameters

1

Hi, could you briefly explain the meaning of: "child_out_filters", "child_num_branches", "child_num_cell_layers", and "child_keep_prob" I am fairly new to ML and am trying to figure out how to properly train the...

ghost

[In this code](https://github.com/melodyguan/enas/blob/master/src/cifar10/micro_child.py#L275-L283): ```python if self.fixed_arc is None: x = self._factorized_reduction(x, out_filters, 2, is_training) layers = [layers[-1], x] #

karandwivedi42

quesion about selecting architectures

3

Aftering running cifar10_micro_search.sh, I got a bunch of architecures and their accuracy, then i selected one architeture with a relatively higher accuracy and retrained it from scratch to get a...

hsl0529

Difference between `_fixed_layer` and `_enas_layer` in `cifar10/micro_child.py`

4

There are a number of differences between `_fixed_layer` and `_enas_layer` in `cifar10/micro_child.py`. 1) [layer_base variable scope](https://github.com/melodyguan/enas/blob/156b81272731f7a8f34f1791131d275597900221/src/cifar10/micro_child.py#L416) 2) [strided pooling layers and convolutions](https://github.com/melodyguan/enas/blob/156b81272731f7a8f34f1791131d275597900221/src/cifar10/micro_child.py#L442) 3) [possible _factorized_reduction for output](https://github.com/melodyguan/enas/blob/156b81272731f7a8f34f1791131d275597900221/src/cifar10/micro_child.py#L389) Are you able...

bkj

In the final phase where you choose the best architecture based on their reward, the reward of ptb and cifar10 is set to be c/ppl + (entropy term) and accuracy,...

AranKomat

now i know that the model result is just a sequence number,if i want to use the model to classify or to visualize the network architecture,do you have the tool?

weidong8405347

I am attempting to run the cifar10 macro search on a set of images that I have converted into the same format as for cifar10. During the child training phase,...

ghost