Rolling Memory

Search…

Rolling Memory

(no paper)

Rolling memory is a type of experience replay. Each training step, a memory is replaced with the top scoring batch item.
Each types pairing becomes a discriminator that is added to the loss.
examples
1
{
2
    "class": "function:hypergan.train_hooks.experimental.rolling_memory_2_train_hook.RollingMemoryTrainHook",
3
    "types": ["mx-/g(mz-)"]
4
}
Copied!
mx- is a memory of x that gets updated each training step. g(mz-) is a memory of z that gets run through a generator and updated each trainng step.
A discriminator d(mx-, g(mz-)) is created and added to the gan loss.
options
attribute
description
type
types
What memories and how they are paired.  See memory types below
array of strings
top_k
How many memory items to replace per frame.  Defaults to 1
integer
only
Overrides all other losses when this is set. Defaults to false
boolean
memory types
memory
description
mx-
x reverse sorted by d_real
mx+
x sorted by d_real
mg-
memory of g reverse sorted by d_fake
mg+
memory of g sorted by d_fake
g(mz-)
generator of memory of z reverse sorted by d_fake
g(mz+)
generator of memory of z sorted by d_fake
x
gan.inputs.x
g
gan.generator.sample

Gradient Penalty

Next - Components

Other GAN implementations

Last modified 1yr ago

Copy link

Contents

examples

options

memory types