HyperGAN Foundation commits | Ecosyste.ms: OpenCollective

[fix] revert class name

github.com/HyperGAN/HyperGAN - 40cfa19742342296e0b873681028451443c447bd authored over 5 years ago

[feature] skip_restore for input_fitness_train_hook

github.com/HyperGAN/HyperGAN - e63507a737c5f7f51efba782772b65ec9cef2d88 authored over 5 years ago

[experiment] alignment

github.com/HyperGAN/HyperGAN - 16ac109595df1775f6448e86e9268952e423497f authored over 5 years ago

[experiment] alignment

github.com/HyperGAN/HyperGAN - 412a75e6618f5ee73b8b182e552833d1f1e6172f authored over 5 years ago

[config] match_support example

github.com/HyperGAN/HyperGAN - 42f52eb0a27efdc53657694a023cd5755a75a7d2 authored over 5 years ago

[experiment] parameters for match_support train hook

github.com/HyperGAN/HyperGAN - a153f985b8d71486049fb0509d109012955f06fb authored over 5 years ago

[feature] support search_steps = 0 for input fitness

github.com/HyperGAN/HyperGAN - 33f273dc89130a455bf70d7966fbdd13977ef94b authored over 5 years ago

[feature] negative momentum optimizer

github.com/HyperGAN/HyperGAN - 2d06455879b175af06c23df274a65089d36e4adf authored over 5 years ago

[feature] search heuristic for learn rate on match_support optimizer

github.com/HyperGAN/HyperGAN - dde64019ba3c48c68def58c77cf935669dc92b77 authored over 5 years ago

[feature] match_support parameterization and convergence end condition

github.com/HyperGAN/HyperGAN - 77e2e8f781b41a02866ca9a32371e50dbbbc4a21 authored over 5 years ago

[experiment] match_support trains X and G before each step

github.com/HyperGAN/HyperGAN - 594b14c6a5bd7d78a38ea1110adddc02d75feb7d authored over 5 years ago

[feature] zero_penalty_train_hook applied l2 loss to zero, used in previously committed config

github.com/HyperGAN/HyperGAN - c4a41d561a211633d2e8290d4271cc52f92a243d authored over 5 years ago

[feature] add dfgan noise layer to debug sampler

github.com/HyperGAN/HyperGAN - 99483e54b3076930498fcc211470f56dd84c44d5 authored over 5 years ago

[feature] distribution filtering gan https://arxiv.org/pdf/1906.04612v1.pdf

github.com/HyperGAN/HyperGAN - 96d1c04a9a1d00659cc14c0aa3100b945cd1fb0e authored over 5 years ago

[feature] l2nn configuration and l2nn changes for stability

github.com/HyperGAN/HyperGAN - a1d44e13820582072bcfa9f0654714c96b226bd3 authored over 5 years ago

[fix] depth trainer call train hooks

github.com/HyperGAN/HyperGAN - 8bf401abd4a1aea4d2f4400cfd5483d79030853f authored over 5 years ago

[feature] force equilibrium train hook

github.com/HyperGAN/HyperGAN - 9b5636f3738237e7310ac7f8e0c41be6fc7851c7 authored over 5 years ago

[feature] turn on optimization for building (size = 1/4)

github.com/HyperGAN/HyperGAN - e1b3e93e95a7e559d3db25ae8bc366499f83664b authored over 5 years ago

release 0.0.1

github.com/HyperGAN/HyperGAN-android - 6224ff03eca77900ca6daab77ada88fc9a552de2 authored over 5 years ago

Merge pull request #155 from HyperGAN/fix-share-image

Fix share image

github.com/HyperGAN/HyperGAN-android - 8189809fb0ed09e84cd840b85310c71f31fcfdd6 authored over 5 years ago

[feature] support faster fitness search with heuristic = zero

github.com/HyperGAN/HyperGAN - b49b79476ba14a2cc400bc69240a5820984688f1 authored over 5 years ago

[feature] heuristic integer lets the fitness search until no new best fit entries are found

github.com/HyperGAN/HyperGAN - 21608c70f94278f4799cc3bd01c8908e58c24615 authored over 5 years ago

[fix] sorting by d_real

github.com/HyperGAN/HyperGAN - 641ddf7d2c3ea2e1ff953d0d0e38d5f917853c04 authored over 5 years ago

update assets

github.com/HyperGAN/HyperGAN-android - 996289bc7d0b43878a5d8ae37e57140820dde2c9 authored over 5 years ago

style feature

github.com/HyperGAN/HyperGAN-android - daa42b405b0d7fffa0d0037306b0e8df1bfae741 authored over 5 years ago

bought tag

github.com/HyperGAN/HyperGAN-android - 792ab7add21b1da44d10efedb4c918da70d1203d authored over 5 years ago

remove unused arg

github.com/HyperGAN/HyperGAN-android - 1401119ba004bd5839cbaf3637e12db5bc0e2724 authored over 5 years ago

upgrade tflite

github.com/HyperGAN/HyperGAN-android - 2569bfc7d59476b5a13784bc7bc238a9409238e9 authored over 5 years ago

[feature] slice layer for slicing z

github.com/HyperGAN/HyperGAN - 6a02779b9d8e0827f7a46576565066e4b668520f authored over 5 years ago

[fix] sorting method for input fitness train hook

github.com/HyperGAN/HyperGAN - be8f13e0e5ec11fbd74b992d0ab4514ccc753106 authored over 5 years ago

[feature] configurable component layer `split`

github.com/HyperGAN/HyperGAN - 403256d6ea70a22079d5fb78665e89508908ed90 authored over 5 years ago

[fix] adabound

github.com/HyperGAN/HyperGAN - c17d78fee79285a86f588d0310324a667fd5a756 authored over 5 years ago

[fix] Sample->refresh does not save files

github.com/HyperGAN/HyperGAN - d50faa05fc61ad80b472544fb20de747f9c4c26e authored over 5 years ago

[refactor] sort layers

github.com/HyperGAN/HyperGAN - f737b4b867db50e9cf74c4475ad665447bddc136 authored over 5 years ago

[fix] ali/ali8 experiments

github.com/HyperGAN/HyperGAN - a9b36562c13158380e62244b82d04b8524fe08a6 authored over 5 years ago

[feature] some configurable component layers

github.com/HyperGAN/HyperGAN - 7910d16e58e29e0cffc46a50088a6db4bc018f2d authored over 5 years ago

[feature] repeat config keyword. example: decay(steps=100 range=0:1 repeat)

github.com/HyperGAN/HyperGAN - 8d4800c16232282c9074dde0b3925745d287d417 authored over 5 years ago

[feature] use gan.configurable_params in adabound

github.com/HyperGAN/HyperGAN - 1d7e980c0cb59dffca3fca5c1fd0b51a21092fbd authored over 5 years ago

[fix] share image

github.com/HyperGAN/HyperGAN-android - b8f423ebcdec08d14cc959603c267a7b0755a70e authored over 5 years ago

[fix] memory leak

github.com/HyperGAN/HyperGAN-android - 9454dbe4f64efdc00d9b670ffcd2eac963cd32a6 authored over 5 years ago

Merge pull request #154 from HyperGAN/upgrade-tflite

Upgrade tflite

github.com/HyperGAN/HyperGAN-android - f570588ab0bcef4530514259845e99415d427526 authored over 5 years ago

[feature] drag on image for slide effect and disabled choose image

github.com/HyperGAN/HyperGAN-android - ee0128564d518e15ed30de2a48cf69acce1f3c54 authored over 5 years ago

[feature] gan loss_type for adversarial_robust_train_hook

github.com/HyperGAN/HyperGAN - 8cbf5ab1e789886e23214572ab314aae3b704faa authored over 5 years ago

[feature] lambdas on adversarial examples

github.com/HyperGAN/HyperGAN - dcc9a315696f7753a1fa3bc3ed92142c8185de99 authored over 5 years ago

[experiment] adversarial robust train hook https://openreview.net/pdf\?id\=HJE6X305Fm

github.com/HyperGAN/HyperGAN - e25589a2604c8233d5d0dc2278c77de7b5156921 authored over 5 years ago

[fix] upgrade sdk

[feature] slider continuously renders

github.com/HyperGAN/HyperGAN-android - 2f5f458a1f6c5966ee94f47eb472d85d609d19b3 authored over 5 years ago

[fix] add_n on ewc

github.com/HyperGAN/HyperGAN - ed260786857a0826fc4c28e1a256ccf9d3c04594 authored over 5 years ago

[feature] dont save if nan in d_fake

github.com/HyperGAN/HyperGAN - 291430b618269f41cd037b2eeaf66a5dcdb5df1c authored over 5 years ago

[feature] AdaBound from https://github.com/taki0112/AdaBound-Tensorflow

github.com/HyperGAN/HyperGAN - 3c7e26ce18b2dc06428f4d2c1fdc5d43c31cc9b4 authored over 5 years ago

[feature] sgd example. no gradient penalty

github.com/HyperGAN/HyperGAN - 17473f528bbd4120e7d6d37e7edcc5c6ef0a7be0 authored over 5 years ago

[feature] batch_fitness_trainer and fixes for input_fitness_trainer

github.com/HyperGAN/HyperGAN - c4721816152afa736d1273b9ffcfca13ed0c054c authored over 5 years ago

[refactor] config.fixed_input

github.com/HyperGAN/HyperGAN - 10eb92902c5f2fbe716b11c5d543b7c3afdfe1ef authored over 5 years ago

[feature] fixed_input standardgan option

github.com/HyperGAN/HyperGAN - 548258bbe31ab59b4228dcabc9dabd3e1cc82220 authored over 5 years ago

[feature] select input by fitness

github.com/HyperGAN/HyperGAN - cfdf068dbc80bf9412b3a9ce456e146dbca11cc5 authored over 5 years ago

[experiment] u_to_z instead of reencode_u_to_z on ali manifold_guided

github.com/HyperGAN/HyperGAN - a0b3fce8e786c658aeee13c322b0cdf7fd7f9c9f authored almost 6 years ago

[feature] different types of minibatch gans

github.com/HyperGAN/HyperGAN - e6b42dbc60c416f578da6aa444e6ae0421db828f authored almost 6 years ago

[feature] minibatch train hook - gan type

github.com/HyperGAN/HyperGAN - b4a65e51b6a71bebdb8e1cac7dd54519281b4277 authored almost 6 years ago

[fix] minibatch train hook

github.com/HyperGAN/HyperGAN - 645d090278a0b2637696f03707bc1a62f4fcda1c authored almost 6 years ago

[refactor] gp_sn_train_hook becomes max_gp_train_hook and paper link added.

github.com/HyperGAN/HyperGAN - 61618e0acd35312eef2e940529fed7c9827391af authored almost 6 years ago

[feature] stylegan-custom1 config

github.com/HyperGAN/HyperGAN - 6a6caa6f40d3448a3fc2ce9f3974a39243d39d1e authored almost 6 years ago

[fix] fitness_trainer working with new structure

github.com/HyperGAN/HyperGAN - 6679cb361edebded637f75ae0da89c6f0a808380 authored almost 6 years ago

[fix] create f_decay and gradient_scale once in elastic_weight_consolidation

github.com/HyperGAN/HyperGAN - bd326360ad971bd059b1ecc3b7e6a406daee45c1 authored almost 6 years ago

[feature] weight penalty train hook with l2nn/ortho support

github.com/HyperGAN/HyperGAN - a9f5fc91ce6f85c85bbd5bf054efe7d9ca8fc91e authored almost 6 years ago

[feature] curl / ewc configurable params

github.com/HyperGAN/HyperGAN - 39ce7cdc5390a3f3801ede56507fc2a3ef803109 authored almost 6 years ago

[fix] configurable_component - reference looks at layers if they exist

github.com/HyperGAN/HyperGAN - 98e86ae46eb6626f2f5a9aa7fe25bcbe9db1efce authored almost 6 years ago

[fix] classification config

github.com/HyperGAN/HyperGAN - ad8ed2f0af29bbacaa6775c8f71a00b74f49e5a5 authored almost 6 years ago

[fix] stride as a list in configs

github.com/HyperGAN/HyperGAN - 73a026442816275db5f331af49dad22013f1c0c3 authored almost 6 years ago

Merge pull request #217 from HyperGAN/develop

0.10 alpha 22

github.com/HyperGAN/HyperGAN - aa847e0c12f854ca95ac2d86e3dfa7cb7309410e authored almost 6 years ago

Merge branch 'master' into develop

github.com/HyperGAN/HyperGAN - e278e86d17a07b284fcd1299752e6da6e4e0267a authored almost 6 years ago

[experiment] option.mask_with for combining layers

github.com/HyperGAN/HyperGAN - a5a99afa5b4e53587989f5b4964da05956ee9408 authored almost 6 years ago

[fix] avg_pool=N on layer

github.com/HyperGAN/HyperGAN - c5da4e1bcffe882c38a2e43a8a3c718776d9e468 authored almost 6 years ago

[fix] config looks in default as well as the base namespace

github.com/HyperGAN/HyperGAN - fc933f0c41091651f8339331446fc8e06b000632 authored almost 6 years ago

[feature] gradient_penalty, config.flex can be an array.

github.com/HyperGAN/HyperGAN - 025a050890969ab9bdc1ce622802aefefa21770d authored almost 6 years ago

[fix] add missing abs call for flex: square(relu(abs(gradient)- flex))

github.com/HyperGAN/HyperGAN - 29636f8507845c8a944a2cc89874c104f4a9d4a0 authored almost 6 years ago

[feature] gradient_penalty_train_hook config.flex lets gp vary from zero

github.com/HyperGAN/HyperGAN - 346c8fd14fe4f5855f8c7e7651473d8775d5c0cd authored almost 6 years ago

[feature] gradient_penalty.lambda is a configurable_param

github.com/HyperGAN/HyperGAN - 1c34374957e6ccfb05e5c33becf4fcd88c5bab86 authored almost 6 years ago

[feature] gradient penalty defaults to being applied to all gan variables, not just discriminator variables

github.com/HyperGAN/HyperGAN - c6b5cf7e8b7ed1e96465b354d42451012c1876ad authored almost 6 years ago

[fix] default decay() configurable param to range[0]

github.com/HyperGAN/HyperGAN - e419248f4f7ea307dd6ea255eb9eda661360b661 authored almost 6 years ago

[fix] skip discriminator concatting x/g if different sizes (for classifier)

github.com/HyperGAN/HyperGAN - 4ebb637738376b544940b5ddfd40db9395277547 authored almost 6 years ago

[feature] examples/classification. Train a generator than classifies

github.com/HyperGAN/HyperGAN - 3b680bcaaa013e78f0f2135ca75d7cd2d481a9ce authored almost 6 years ago

[fix] specify components with defaulting to discriminator target in gradient_penalty_train_hook

github.com/HyperGAN/HyperGAN - b3d8d2b709f93f9ce808b44e0946b4e00d019c55 authored almost 6 years ago

[feature] configurable_param allows parameters to change over training steps

github.com/HyperGAN/HyperGAN - 6f326abbf9b943152a25fd12ec53f89514e442be authored almost 6 years ago

[fix] update stylegan config to stabler version

github.com/HyperGAN/HyperGAN - 4021b3b5651b0440a3ae838ddf4417ac53dac41a authored almost 6 years ago

[wip] alignment

github.com/HyperGAN/HyperGAN - 1193d7e26ae202ff26e5fee1296552fb076dd983 authored almost 6 years ago

[fix] allow layer name lookup across components

github.com/HyperGAN/HyperGAN - e3d701eaca495347ee4484cdc52ba9736388a122 authored almost 6 years ago

[fix] alpha channel in tk_viewer would crash sampler

github.com/HyperGAN/HyperGAN - 000bfadfb896d3432d08b6f1ba58a3f601219159 authored almost 6 years ago

[fix] batch size 1 on 2d-distribution maps contours/points

github.com/HyperGAN/HyperGAN - 54d7a5d515f1df71bbe9fe0e49fae3a09fc0db94 authored almost 6 years ago

[feature] better visualization of 2d example

github.com/HyperGAN/HyperGAN - 521579b4cc2ba7c1ec865e847ab3d4cff56ac79f authored almost 6 years ago

[fix] 2d-distribution example

github.com/HyperGAN/HyperGAN - 8c8a0015a5b43acd992846436c3ad494c549a8cb authored almost 6 years ago

[experiment] alignment

github.com/HyperGAN/HyperGAN - c46bce2ee4a2afad38abaacdb4c104c0b50eb447 authored almost 6 years ago

Merge pull request #216 from HyperGAN/develop

0.10 alpha 21

github.com/HyperGAN/HyperGAN - 2c4df74bb308b931a3cbb63073a83679847a3b3a authored almost 6 years ago

Merge branch 'master' into develop

github.com/HyperGAN/HyperGAN - 8912506c8aedc47f56c6b589b7f07b8e65804191 authored almost 6 years ago

[feature] stylegan config

github.com/HyperGAN/HyperGAN - f3dac197b8f2548b1b668ab88ff80f644805f3a9 authored almost 6 years ago

[fix] add initializer= and trainable= options to 'const'/'noise learned' layers.\n[fix] Move stylegan weight multiplier to get_weights

github.com/HyperGAN/HyperGAN - 63d5c4449077d9f0124e93bd4fff2d5ba5e2c2e8 authored almost 6 years ago

[fix] initializers were not being parsed from config.default

github.com/HyperGAN/HyperGAN - 5b01095c5a117a5577bae6459fe936e16f99f5f8 authored almost 6 years ago

[feature] output number of parameters per layer during network construction

github.com/HyperGAN/HyperGAN - 1e0015801eff833b5d00838a0e1c1413c5f0b6b5 authored almost 6 years ago

[feature] interesting and small colorizer config for 1024x1024

github.com/HyperGAN/HyperGAN - d4814c5908f9b6f65e54ff677e8003731c8946a5 authored almost 6 years ago

[feature] interesting colorizer-resizeconv config

github.com/HyperGAN/HyperGAN - 548b8a04d5494f6451b5550d8395d48e82a198bf authored almost 6 years ago

[feature] output training command with "hypergan new -l ."

github.com/HyperGAN/HyperGAN - 81a05e803fc468ccaccaa1bdd3537a9198c94e6f authored almost 6 years ago

[refactor] remove resolution from config filenames.

github.com/HyperGAN/HyperGAN - 02064c4aed4ac1014c49e2be424adfd55cc5c7a2 authored almost 6 years ago