github.com/HyperGAN/HyperGAN commits | Ecosyste.ms: OpenCollective

[feature] optimize & fitness distributions

e631df61841cce67e6e8d65a2d5c8843f68e3a46 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[feature] config.input

[refactor] latent.next()
[fix] sample --steps

414f89f41439870b5f0c8a672647ee87fa204cc6 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[feature] fitnessinput

70d268858f44e1b147651798c27bdaad7e401a90 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[refactor] input loader

a76a814d113a4b1aa4cb7151748a8141d4db5c0c authored almost 5 years ago by Mikkel Garcia <[email protected]>

[fix] extragradient formulation

53debd93663ab5314905b3a858c0961d0aaa2dcb authored almost 5 years ago by Mikkel Garcia <[email protected]>

[feature] extragradient formulation

d7fdad78e6f3083e9a03af7c80e115cf9d584c6e authored almost 5 years ago by Mikkel Garcia <[email protected]>

[fix] disable beta1 for now

289430968418213661f852426a4d12276a8d6f65 authored almost 5 years ago by Martyn Garcia <[email protected]>

[feature] layer_dropout

042d4756e4efe71ef0376a186ed15e40a5e770ca authored almost 5 years ago by Mikkel Garcia <[email protected]>

[fix] missing calculate_gradients method

1ea11fbf4c2c2050ad2503a82816d181c9b1f4e1 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[feature] negative momentum train hook

894b2c70f9bcbe229aba5576ffc3fb55d873e024 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[feature] extragradient train hook

5f72bb69d05fee8535ccda0ffedab6744262e18e authored almost 5 years ago by Mikkel Garcia <[email protected]>

Merge branch 'pytorch' of github.com:255BITS/HyperGAN into pytorch

* 'pytorch' of github.com:255BITS/HyperGAN:
[fix] add deps
[feature] chargan

0b272ad51191b1636c120cd4374eb910eff439b2 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[fix] batch walk sampler

314f925d0a089d5a44331c95eb2bfefcae1b0bc7 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[feature] chargan

fb659be27c144f9ec913a81c1cb8af2777f68cab authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] add deps

902c37505debcf2f00a9c6857b95503dd2633cbe authored almost 5 years ago by Martyn Garcia <[email protected]>

[feature] batch walk sampler with pytorch

345fce3d67cafe3a1728b177d376193f48efd4bd authored almost 5 years ago by Mikkel Garcia <[email protected]>

[fix] input range

85d0efd0774bf9e63d08791d6f640817b3abb26b authored almost 5 years ago by Mikkel Garcia <[email protected]>

[cleanup] remove unused files and remove --classloss (use -a)

15fa82c468603843aff7f18cd75221e87ab89964 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[fix] disable removed layers

db726ff0020598f16c3c661f457637a28a05f850 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[fix] disable removed layers

8a6c19b983cbe2060025f2c8d36cd93a00a8f1ae authored almost 5 years ago by Mikkel Garcia <[email protected]>

[fix] remove old tf code and mark what layers need to be converted still

23ae68b22e477997c875e04f7646dd9c67d28720 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[fix] next frame bug

07a6d24ef9d772e15e1712c64f78f4dd7671437a authored almost 5 years ago by Mikkel Garcia <[email protected]>

[fix] next frame bug

54428ffb51914d582fa90b97dac57e72b991917e authored almost 5 years ago by Mikkel Garcia <[email protected]>

[feature] variational layer

7cd98dec8e834c4818108d1bb4c0c102f2047de6 authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] optimistic adam

1726e8e6778b07279362a06a1a92b117cabf7213 authored almost 5 years ago by Martyn Garcia <[email protected]>

[wip] multi marginal

d61e75d0a2c2472a5d78d9fb1874680476d8931e authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] report g_loss/d_loss correctly and remove old code

9e12d2527d718697c13742ec4b1fa29e183737d2 authored almost 5 years ago by Martyn Garcia <[email protected]>

[feature] ali + vae loss

77d6ef906b95a4ebbcaaa4067bac4b3b2b9db15c authored almost 5 years ago by Martyn Garcia <[email protected]>

[feature] new layers: vae, conv3d, adaptive_avg_pool3d

2b4a7b0ceba071e026d618cf5898fcaa3d0e862f authored almost 5 years ago by Martyn Garcia <[email protected]>

[wip] next frame

fb5a8a3e0a73eea528ac398c04ba20a96ef575e1 authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] remove unused flag

85a49fda28720b6ec06008f0c064d257036b4f9c authored almost 5 years ago by Martyn Garcia <[email protected]>

[feature] hardtanh activation

7728013d3b31814efd3ab4a7fd5f3346610d7b01 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[fix] step counter

0fb710444b274e1b3aa9d63e63aebcfc04b69397 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[feature] hook.gradients

6e7bc01007c8d762096a61ec39cd5334feabe0e5 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[wip] next frame example

cf72dcaef367bf5625aa3a83a8bce8bbe823d892 authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] conjecture train hook

851e233c41cc071046d2386991235e739862c04e authored almost 5 years ago by Mikkel Garcia <[email protected]>

[feature] learn rate dropout train hook

bf1b18ea0691e010be3d09e349f2becc16ade150 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[fix] ali

4d3848ea2a2b22ad9d9a8364061e8ffc83b91dae authored almost 5 years ago by Mikkel Garcia <[email protected]>

[fix] use ordered samples

f7dc6fc130bfbcbfbb1a68eacbc2948be4b40859 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[refactor] gradient_norm_train_hook

69c205fc0c0c0ffa8429afbb487c6b055810a5c9 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[wip] improve 2 class ali, needs refactoring

eb444522eb3077e517d37b906393f5e92eaabe32 authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] default to adam

27480b383a7b4c4457b077cd90a9bb139e693766 authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] optimizers

35ef007642f4307e9bc2b64cd6979b0935db2ed4 authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] add "concat layer z"

04dda1b7f13e5bbcdf8bb5d2f3836571ea2acf7f authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] save/load

8451a2d1b3264a92b85c0eef49e86d76f704f9f0 authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] ali gan

a700d1c7ebe4f9a56c45ecfe4a8a99cc51deedd5 authored almost 5 years ago by Martyn Garcia <[email protected]>

[feature] loss option

d8759a5edcf4f6f42fccf6bf1e12e384f75c1f3d authored almost 5 years ago by Mikkel Garcia <[email protected]>

[feature] optimization for zero gradient

dffff0ff849e6a57b510b8e1f13c9a5f408b9377 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[feature] new regularizer

e0924eeb790c02a024bb1f8f3a34386f3e4507a1 authored almost 5 years ago by Mikkel Garcia <[email protected]>

Merge branch 'pytorch' of github.com:255BITS/HyperGAN into pytorch

* 'pytorch' of github.com:255BITS/HyperGAN:
[fix] adamirror
[fix] ali_loss
[feature] layer...

afd4bed459794358626feba3070613d17a9c8449 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[feature] gradient_penalty options

b7e3219192084087500fc74238f5db6c44daee43 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[fix] adamirror

299476117b11feb23e878a365c73e3359a3cab48 authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] ali_loss

53035228038b6c0ebbd62347f0566af0c68a4d2e authored almost 5 years ago by Martyn Garcia <[email protected]>

[feature] layer_initializer

9590ac0626eba16eb85156562c6b30dfc5e1961b authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] --random_crop

8c3a52b25dd9c5a814de483abbc77eb6445fbd63 authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] sizing

3525764324d15cd1654209cfdc3fba303deb07bf authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] reshape

6919039a463150f50f0cbf60bf14bfc273172e1d authored almost 5 years ago by Mikkel Garcia <[email protected]>

[fix] reshape

f10257705758df75422292e0afe3a42704d80c52 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[fix] add missing files

ceb1f1a6f32cf00279423ab4d6ed2c0925d40bd5 authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] layer 'layer', 'identity', 'latent', and dont use isinstance for speed

5d191cb062b75a48885acdbe29e4f7e3ff184d1f authored almost 5 years ago by Martyn Garcia <[email protected]>

[feature] layer adaptive_instance_norm

e2cf1e4f51b9e10056fb79bdefd3de769131a8cc authored almost 5 years ago by Martyn Garcia <[email protected]>

[doc] start updating docs

701f2b05f7cf36407ca4c5620b4cc50c81f8325a authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] instance_norm affine=False option

444e8cd5a1069b26466a86a11c35b13195e9e67c authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] linear bias option

8927f56419e922ff86ba6387742c4fa25aaa40ae authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] activations on resizablegenerator

9f526d266d23a6c887966bbd550555bd188926eb authored almost 5 years ago by Martyn Garcia <[email protected]>

[wip] conv/linear now output logits - now must specify activation such as lrelu/relu/etc as independent layers. also allow nn.Module as layers

f9a7e7e73dbbd22201faf9a68164e712c1dc99a0 authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] dcgan

8ca7f4a22620ae4327f64f78211b47ddcea59ad1 authored almost 5 years ago by Martyn Garcia <[email protected]>

[feature] layers residual, instance_norm

693d9ed19342ce88bd70ebc594e7bbe3bcf3cd6f authored almost 5 years ago by Martyn Garcia <[email protected]>

[feature] layer "concat noise", no options for now

04852ee5fee60bfd669c5a500b16857edec07f6a authored almost 5 years ago by Martyn Garcia <[email protected]>

Merge branch 'pytorch' of github.com:255BITS/HyperGAN into pytorch

* 'pytorch' of github.com:255BITS/HyperGAN:
[fix] save/load - kinda hacked for now
[fix] mul...

b5aa2dc4cb71c9754eb4d651d368eb89476bd885 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[fix] wgan-gp

840e02d43c5724847f767d6bc82981b3aed5eaa6 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[fix] save/load - kinda hacked for now

7fe6218de3595f4c0253fa89fe231de12297cd55 authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] multiple outputs on sampler

a664eeed47dc15717dd93ea0c2f39569f79f99c5 authored almost 5 years ago by Martyn Garcia <[email protected]>

[feature] test config for multi marginal

a3e907b8fb5c6b998bcc76f99f81704dc878d991 authored almost 5 years ago by Martyn Garcia <[email protected]>

[wip] convert multi_marginal_gan

c5af1a87fe4ebe936ca1323c66dbf4557920e2bd authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] requirements.txt

69b5cb7d4933ddba1954542fca4a213f11e66d8d authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] remove tensorflow requirement

ca4bd3d5600832230e3430e2148e47be7ec379b4 authored almost 5 years ago by Martyn Garcia <[email protected]>

[feature] Reshape module

256667692e65ad98232e1c9ca8065aa3388cb8c9 authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] activation selection

b499c2fd5b0907290d0778ec8c76144134f3d776 authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] ConfigurableComponent linear 4*4*512

1d50d1fbba5a48af58c1c5f5e732c207d58a7a87 authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] static learn_rate

d421451c71f9c0553a2d87725eb4c9097c2c88a7 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[fix] stabilizing example

a1b072971c3878be5e3fa35057cae578a30f9d61 authored almost 5 years ago by Mikkel Garcia <[email protected]>

Merge branch 'pytorch' of github.com:255BITS/HyperGAN into pytorch

* 'pytorch' of github.com:255BITS/HyperGAN:
[fix] resize_conv and subpixel on resizable
[ref...

685035b7c8f3fcaacd6a584db54cefa86c97cd70 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[feature] stabilizing, broken

e0404b5ade2cde48a00aed1ad6cee7b36f69aa73 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[feature] stabilizing, broken

d0f63a4b302cbfac711d69ae6c15deabc5f1a001 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[fix] resize_conv and subpixel on resizable

941163d8847b1974c3c66e3bf7e0fa5387e263c3 authored almost 5 years ago by Martyn Garcia <[email protected]>

[refactor] configurable component

3e95bf89db1958951eb1583727c599bb9abc75ef authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] initial resizable working

c2b21cba9612f83bc2b9d7ef2a9c43da1fd4cbdb authored almost 5 years ago by Martyn Garcia <[email protected]>

[wip] test config

ffe7fef66aeab078f465c4a2cf7a8ab38dc32b80 authored almost 5 years ago by Martyn Garcia <[email protected]>

[wip] resizable

4991e9fedc1615297077091b192feed630b02f89 authored almost 5 years ago by Martyn Garcia <[email protected]>

[feature] dcgan-wgangp

289dcf404046a2009ce583f06cb604a4b60a6187 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[fix] handle stopiteration on standardgan

ea34be9141bf10ed59737dab9c5f64f42cbcb23e authored almost 5 years ago by Mikkel Garcia <[email protected]>

Merge branch 'pytorch' of github.com:255BITS/HyperGAN into pytorch

* 'pytorch' of github.com:255BITS/HyperGAN:
[feature] start porting configurable component

0b6b4344eaf79de1564f643b75957d817ea4ff1d authored almost 5 years ago by Mikkel Garcia <[email protected]>

[fix] drop_last=True

94a8bba09cc6b9b2b5b18c421040ff75bf187845 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[feature] start porting configurable component

894a2b921ea0ba9969cec62b484d03ab706ad25a authored almost 5 years ago by Martyn Garcia <[email protected]>

Merge branch 'pytorch' of github.com:255BITS/HyperGAN into pytorch

* 'pytorch' of github.com:255BITS/HyperGAN:
[feature] port alternating trainer with example

5e6e50c518a9f0853736bcacdfbb0035005fb1cc authored almost 5 years ago by Mikkel Garcia <[email protected]>

[fix] simultaneous

160bb6395e3690b96290a81aa4b5fb04020eb5a5 authored almost 5 years ago by Mikkel Garcia <[email protected]>

[feature] port alternating trainer with example

d78a6f10ce5360c49d25166c4acb1e4ccb32734d authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] gan.parameters

361dfd46b54b20fc8da4159aaebb4f63cf90fb38 authored almost 5 years ago by Martyn Garcia <[email protected]>

[fix] loss is GANComponent

93ab27b7afbc1b04379b9c78ef8969b8e286338c authored almost 5 years ago by Martyn Garcia <[email protected]>