Architecture reference
This page lists all available parameters for each architecture in traiNNer-redux. While the default configs use the official defaults and shouldn’t need to be modified by most users, advanced users may wish to inspect or modify architectures to train to their liking. Please keep in mind that changing parameters for generator architectures can affect compatibility with using pretrain models.
Generator architectures (network_g)
ATD
atd
type: atd
in_chans: 3
img_size: 96
embed_dim: 210
depths: [6, 6, 6, 6, 6, 6]
num_heads: [6, 6, 6, 6, 6, 6]
window_size: 16
category_size: 256
num_tokens: 128
reducted_dim: 20
convffn_kernel_size: 5
img_range: 1.0
mlp_ratio: 2
upsampler: pixelshuffle
resi_connection: 1conv
atd_light
type: atd_light
in_chans: 3
img_size: 64
embed_dim: 48
depths: [6, 6, 6, 6]
num_heads: [6, 6, 6, 6]
window_size: 16
category_size: 128
num_tokens: 64
reducted_dim: 8
convffn_kernel_size: 7
img_range: 1.0
mlp_ratio: 1
upsampler: pixelshuffledirect
resi_connection: 1conv
ArtCNN
artcnn_r16f96
type: artcnn_r16f96
in_ch: 3
filters: 96
n_block: 16
kernel_size: 3
artcnn_r3f24
type: artcnn_r3f24
in_ch: 3
filters: 24
n_block: 3
kernel_size: 3
artcnn_r5f48
type: artcnn_r5f48
in_ch: 3
filters: 48
n_block: 5
kernel_size: 3
artcnn_r8f24
type: artcnn_r8f24
in_ch: 3
filters: 24
n_block: 8
kernel_size: 3
artcnn_r8f48
type: artcnn_r8f48
in_ch: 3
filters: 48
n_block: 8
kernel_size: 3
artcnn_r8f64
type: artcnn_r8f64
in_ch: 3
filters: 64
n_block: 8
kernel_size: 3
AutoEncoder
autoencoder
type: autoencoder
nf: 64
CRAFT
craft
type: craft
window_size: 16
embed_dim: 48
depths: [2, 2, 2, 2]
num_heads: [6, 6, 6, 6]
split_size_0: 4
split_size_1: 16
mlp_ratio: 2.0
qkv_bias: true
qk_scale: ~
img_range: 1.0
resi_connection: 1conv
CascadedGaze
cascadedgaze
type: cascadedgaze
img_channel: 3
width: 60
middle_blk_num: 10
enc_blk_nums: ~
dec_blk_nums: ~
GCE_CONVS_nums: ~
DAT
dat
type: dat
in_chans: 3
img_size: 64
img_range: 1.0
split_size: [8, 32]
depth: [6, 6, 6, 6, 6, 6]
embed_dim: 180
num_heads: [6, 6, 6, 6, 6, 6]
expansion_factor: 4
resi_connection: 1conv
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
use_chk: false
upsampler: pixelshuffle
unshuffle_mod: false
dat_2
type: dat_2
in_chans: 3
img_size: 64
img_range: 1.0
split_size: [8, 32]
depth: [6, 6, 6, 6, 6, 6]
embed_dim: 180
num_heads: [6, 6, 6, 6, 6, 6]
expansion_factor: 2
resi_connection: 1conv
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
use_chk: false
upsampler: pixelshuffle
unshuffle_mod: false
dat_2_l
type: dat_2_l
in_chans: 3
img_size: 64
img_range: 1.0
split_size: [8, 32]
depth: [6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6]
embed_dim: 180
num_heads: [6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6]
expansion_factor: 2
resi_connection: 1conv
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
use_chk: false
upsampler: pixelshuffle
unshuffle_mod: false
dat_light
type: dat_light
in_chans: 3
img_size: 64
img_range: 1.0
split_size: [8, 32]
depth: [18]
embed_dim: 60
num_heads: [6]
expansion_factor: 2
resi_connection: 3conv
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
use_chk: false
upsampler: pixelshuffle
unshuffle_mod: false
dat_s
type: dat_s
in_chans: 3
img_size: 64
img_range: 1.0
split_size: [8, 16]
depth: [6, 6, 6, 6, 6, 6]
embed_dim: 180
num_heads: [6, 6, 6, 6, 6, 6]
expansion_factor: 2
resi_connection: 1conv
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
use_chk: false
upsampler: pixelshuffle
unshuffle_mod: false
DAT_iLN
dat_iln
type: dat_iln
in_chans: 3
img_size: 64
split_size: [8, 32]
depth: [6, 6, 6, 6, 6, 6]
embed_dim: 180
num_heads: [6, 6, 6, 6, 6, 6]
expansion_factor: 4
resi_connection: 1conv
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
use_chk: false
upsampler: pixelshuffle
unshuffle_mod: false
dat_iln_2
type: dat_iln_2
in_chans: 3
img_size: 64
split_size: [8, 32]
depth: [6, 6, 6, 6, 6, 6]
embed_dim: 180
num_heads: [6, 6, 6, 6, 6, 6]
expansion_factor: 2
resi_connection: 1conv
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
use_chk: false
upsampler: pixelshuffle
unshuffle_mod: false
dat_iln_light
type: dat_iln_light
in_chans: 3
img_size: 64
split_size: [8, 32]
depth: [18]
embed_dim: 60
num_heads: [6]
expansion_factor: 2
resi_connection: 3conv
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
use_chk: false
upsampler: pixelshuffle
unshuffle_mod: false
dat_iln_s
type: dat_iln_s
in_chans: 3
img_size: 64
split_size: [8, 16]
depth: [6, 6, 6, 6, 6, 6]
embed_dim: 180
num_heads: [6, 6, 6, 6, 6, 6]
expansion_factor: 2
resi_connection: 1conv
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
use_chk: false
upsampler: pixelshuffle
unshuffle_mod: false
DCTLSA
dctlsa
type: dctlsa
in_nc: 3
nf: 55
num_modules: 6
out_nc: 3
num_head: 5
DIS
dis_balanced
type: dis_balanced
in_channels: 3
out_channels: 3
num_features: 32
num_blocks: 12
use_depthwise: false
dis_fast
type: dis_fast
in_channels: 3
out_channels: 3
num_features: 32
num_blocks: 8
use_depthwise: false
DITN_Real
ditn_real
type: ditn_real
inp_channels: 3
dim: 60
ITL_blocks: 4
SAL_blocks: 4
UFONE_blocks: 1
ffn_expansion_factor: 2
bias: false
LayerNorm_type: WithBias
patch_size: 8
DRCT
drct
type: drct
in_chans: 3
img_size: 64
window_size: 16
img_range: 1.0
depths: [6, 6, 6, 6, 6, 6]
embed_dim: 180
num_heads: [6, 6, 6, 6, 6, 6]
mlp_ratio: 2.0
upsampler: pixelshuffle
resi_connection: 1conv
use_checkpoint: false
drct_l
type: drct_l
in_chans: 3
img_size: 64
window_size: 16
img_range: 1.0
depths: [6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6]
embed_dim: 180
num_heads: [6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6]
mlp_ratio: 2.0
upsampler: pixelshuffle
resi_connection: 1conv
use_checkpoint: false
drct_xl
type: drct_xl
in_chans: 3
img_size: 64
window_size: 16
img_range: 1.0
depths: [6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6]
embed_dim: 180
num_heads: [6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6]
mlp_ratio: 2.0
upsampler: pixelshuffle
resi_connection: 1conv
use_checkpoint: false
DWT
dwt
type: dwt
img_size: 64
patch_size: 1
in_chans: 3
embed_dim: 180
depths: [6, 6, 6, 6, 6, 6]
num_heads: [6, 6, 6, 6, 6, 6]
window_size: 16
mlp_ratio: 2.0
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
ape: false
patch_norm: true
use_checkpoint: false
img_range: 1.0
upsampler: pixelshuffle
resi_connection: 1conv
layer_kinds: [[0, -1, 0, -1, 0, -1], [0, -1, 0, -1, 0, -1], [0, 2, 0, 2, 0, 2], [0, 2, 0, 2, 0, 2], [0, 1, 0, 1, 0, 1], [0, 1, 0, 1, 0, 1]]
dwt_s
type: dwt_s
img_size: 64
patch_size: 1
in_chans: 3
embed_dim: 180
depths: [6, 6, 6, 6, 6, 6]
num_heads: [6, 6, 6, 6, 6, 6]
window_size: 8
mlp_ratio: 2.0
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
ape: false
patch_norm: true
use_checkpoint: false
img_range: 1.0
upsampler: pixelshuffle
resi_connection: 1conv
layer_kinds: [[0, -1, 0, -1, 0, -1], [0, -1, 0, -1, 0, -1], [0, 2, 0, 2, 0, 2], [0, 2, 0, 2, 0, 2], [0, 1, 0, 1, 0, 1], [0, 1, 0, 1, 0, 1]]
DocRes
docres
type: docres
inp_channels: 3
out_channels: 3
dim: 48
num_blocks: [2, 3, 3, 4]
num_refinement_blocks: 4
heads: [1, 2, 4, 8]
ffn_expansion_factor: 2.66
bias: false
LayerNorm_type: WithBias
dual_pixel_task: true
docres_base
type: docres_base
inp_channels: 3
out_channels: 3
dim: 48
num_blocks: [2, 3, 3, 4]
num_refinement_blocks: 4
heads: [1, 2, 4, 8]
ffn_expansion_factor: 2.66
bias: false
LayerNorm_type: WithBias
dual_pixel_task: true
docres_large
type: docres_large
inp_channels: 3
out_channels: 3
dim: 64
num_blocks: [3, 4, 4, 6]
num_refinement_blocks: 6
heads: [2, 4, 8, 16]
ffn_expansion_factor: 2.66
bias: false
LayerNorm_type: WithBias
dual_pixel_task: true
EIMN
eimn_a
type: eimn_a
embed_dims: 64
depths: 1
mlp_ratios: 2.66
drop_rate: 0.0
drop_path_rate: 0.0
num_stages: 14
freeze_param: false
eimn_l
type: eimn_l
embed_dims: 64
depths: 1
mlp_ratios: 2.66
drop_rate: 0.0
drop_path_rate: 0.0
num_stages: 16
freeze_param: false
ELAN
elan
type: elan
colors: 3
rgb_range: 255
norm: false
window_sizes: ~
m_elan: 36
c_elan: 180
n_share: 0
r_expand: 2
elan_light
type: elan_light
colors: 3
rgb_range: 255
norm: false
window_sizes: ~
m_elan: 24
c_elan: 60
n_share: 1
r_expand: 2
EMT
emt
type: emt
num_in_ch: 3
num_out_ch: 3
upsampler: pixelshuffle
dim: 60
n_blocks: 6
n_layers: 6
num_heads: 3
mlp_ratio: 2
n_GTLs: 2
window_list: [[32, 8], [8, 32]]
shift_list: [[16, 4], [4, 16]]
ESCRealM
escrealm
type: escrealm
dim: 64
pdim: 16
kernel_size: 13
n_blocks: 10
conv_blocks: 5
window_size: 32
num_heads: 4
exp_ratio: 2.0
attn_type: Flex
mid_dim: 64
upsampler: transpose+conv
unshuffle_mod: false
escrealm_xl
type: escrealm_xl
dim: 128
pdim: 32
kernel_size: 13
n_blocks: 16
conv_blocks: 5
window_size: 32
num_heads: 8
exp_ratio: 2.0
attn_type: Flex
mid_dim: 64
upsampler: pixelshuffle
unshuffle_mod: false
FDAT
fdat
type: fdat
num_in_ch: 3
num_out_ch: 3
embed_dim: 120
num_groups: 4
depth_per_group: 3
num_heads: 4
window_size: 8
ffn_expansion_ratio: 2.0
aim_reduction_ratio: 8
group_block_pattern: ~
drop_path_rate: 0.1
mid_dim: 64
upsampler_type: pixelshuffle
img_range: 1.0
unshuffle_mod: false
fdat_large
type: fdat_large
num_in_ch: 3
num_out_ch: 3
embed_dim: 180
num_groups: 4
depth_per_group: 4
num_heads: 6
window_size: 8
ffn_expansion_ratio: 2.0
aim_reduction_ratio: 8
group_block_pattern: ~
drop_path_rate: 0.1
upsampler_type: transpose+conv
img_range: 1.0
fdat_light
type: fdat_light
num_in_ch: 3
num_out_ch: 3
embed_dim: 108
num_groups: 3
depth_per_group: 2
num_heads: 4
window_size: 8
ffn_expansion_ratio: 2.0
aim_reduction_ratio: 8
group_block_pattern: ~
drop_path_rate: 0.08
upsampler_type: transpose+conv
img_range: 1.0
unshuffle_mod: false
fdat_medium
type: fdat_medium
num_in_ch: 3
num_out_ch: 3
embed_dim: 120
num_groups: 4
depth_per_group: 3
num_heads: 4
window_size: 8
ffn_expansion_ratio: 2.0
aim_reduction_ratio: 8
group_block_pattern: ~
drop_path_rate: 0.1
upsampler_type: transpose+conv
img_range: 1.0
unshuffle_mod: false
fdat_medium_aligned
type: fdat_medium_aligned
num_in_ch: 3
num_out_ch: 3
embed_dim: 128
num_groups: 4
depth_per_group: 3
num_heads: 4
window_size: 8
ffn_expansion_ratio: 2.0
aim_reduction_ratio: 8
group_block_pattern: ~
drop_path_rate: 0.1
upsampler_type: transpose+conv
img_range: 1.0
unshuffle_mod: false
fdat_tiny
type: fdat_tiny
num_in_ch: 3
num_out_ch: 3
embed_dim: 96
num_groups: 2
depth_per_group: 2
num_heads: 3
window_size: 8
ffn_expansion_ratio: 1.5
aim_reduction_ratio: 8
group_block_pattern: ~
drop_path_rate: 0.05
upsampler_type: transpose+conv
img_range: 1.0
unshuffle_mod: false
fdat_xl
type: fdat_xl
num_in_ch: 3
num_out_ch: 3
embed_dim: 180
num_groups: 6
depth_per_group: 6
num_heads: 6
window_size: 8
ffn_expansion_ratio: 2.0
aim_reduction_ratio: 8
group_block_pattern: ~
drop_path_rate: 0.1
upsampler_type: transpose+conv
img_range: 1.0
FlexNet
flexnet
type: flexnet
inp_channels: 3
out_channels: 3
dim: 64
num_blocks: [6, 6, 6, 6, 6, 6]
window_size: 8
hidden_rate: 4
channel_norm: false
attn_drop: 0.0
proj_drop: 0.0
pipeline_type: linear
upsampler: pixelshuffle
metaflexnet
type: metaflexnet
inp_channels: 3
out_channels: 3
dim: 64
num_blocks: [4, 6, 6, 8]
window_size: 8
hidden_rate: 4
channel_norm: false
attn_drop: 0.0
proj_drop: 0.0
pipeline_type: meta
upsampler: pixelshuffle
GRL
grl_b
type: grl_b
img_size: 64
in_channels: 3
out_channels: 3
embed_dim: 180
img_range: 1.0
upsampler: pixelshuffle
depths: [4, 4, 8, 8, 8, 4, 4]
num_heads_window: [3, 3, 3, 3, 3, 3, 3]
num_heads_stripe: [3, 3, 3, 3, 3, 3, 3]
window_size: 32
stripe_size: [64, 64]
stripe_groups: [None, None]
stripe_shift: true
mlp_ratio: 2.0
qkv_bias: true
qkv_proj_type: linear
anchor_proj_type: avgpool
anchor_one_stage: true
anchor_window_down_factor: 4
out_proj_type: linear
local_connection: true
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
pretrained_window_size: [0, 0]
pretrained_stripe_size: [0, 0]
conv_type: 1conv
init_method: n
fairscale_checkpoint: false
offload_to_cpu: false
euclidean_dist: false
grl_s
type: grl_s
img_size: 64
in_channels: 3
out_channels: 3
embed_dim: 128
img_range: 1.0
upsampler: pixelshuffle
depths: [4, 4, 4, 4]
num_heads_window: [2, 2, 2, 2]
num_heads_stripe: [2, 2, 2, 2]
window_size: 32
stripe_size: [64, 64]
stripe_groups: [None, None]
stripe_shift: true
mlp_ratio: 2.0
qkv_bias: true
qkv_proj_type: linear
anchor_proj_type: avgpool
anchor_one_stage: true
anchor_window_down_factor: 4
out_proj_type: linear
local_connection: false
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
pretrained_window_size: [0, 0]
pretrained_stripe_size: [0, 0]
conv_type: 1conv
init_method: n
fairscale_checkpoint: false
offload_to_cpu: false
euclidean_dist: false
grl_t
type: grl_t
img_size: 64
in_channels: 3
out_channels: 3
embed_dim: 64
img_range: 1.0
upsampler: pixelshuffledirect
depths: [4, 4, 4, 4]
num_heads_window: [2, 2, 2, 2]
num_heads_stripe: [2, 2, 2, 2]
window_size: 32
stripe_size: [64, 64]
stripe_groups: [None, None]
stripe_shift: true
mlp_ratio: 2.0
qkv_bias: true
qkv_proj_type: linear
anchor_proj_type: avgpool
anchor_one_stage: true
anchor_window_down_factor: 4
out_proj_type: linear
local_connection: false
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
pretrained_window_size: [0, 0]
pretrained_stripe_size: [0, 0]
conv_type: 1conv
init_method: n
fairscale_checkpoint: false
offload_to_cpu: false
euclidean_dist: false
GateRV3
gaterv3
type: gaterv3
in_ch: 3
dim: 32
enc_blocks: [2, 2, 4, 6]
dec_blocks: [2, 2, 2, 2]
num_latent: 8
upsample: pixelshuffle
upsample_mid_dim: 48
end_gamma_init: 1
attention: false
sisr_blocks: 4
flash: true
gaterv3_r
type: gaterv3_r
gaterv3_s
type: gaterv3_s
HAT
hat_l
type: hat_l
img_size: 64
patch_size: 1
in_chans: 3
embed_dim: 180
depths: [6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6]
num_heads: [6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6]
window_size: 16
compress_ratio: 3
squeeze_factor: 30
conv_scale: 0.01
overlap_ratio: 0.5
mlp_ratio: 2.0
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
ape: false
patch_norm: true
use_checkpoint: false
img_range: 1.0
upsampler: pixelshuffle
resi_connection: 1conv
num_feat: 64
hat_l_aligned
type: hat_l_aligned
img_size: 64
patch_size: 1
in_chans: 3
embed_dim: 192
depths: [6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6]
num_heads: [6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6]
window_size: 16
compress_ratio: 3
squeeze_factor: 30
conv_scale: 0.01
overlap_ratio: 0.5
mlp_ratio: 2.0
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
ape: false
patch_norm: true
use_checkpoint: false
img_range: 1.0
upsampler: pixelshuffle
resi_connection: 1conv
num_feat: 64
hat_l_aligned_down
type: hat_l_aligned_down
img_size: 64
patch_size: 1
in_chans: 3
embed_dim: 160
depths: [6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6]
num_heads: [5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5]
window_size: 16
compress_ratio: 3
squeeze_factor: 30
conv_scale: 0.01
overlap_ratio: 0.5
mlp_ratio: 2.0
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
ape: false
patch_norm: true
use_checkpoint: false
img_range: 1.0
upsampler: pixelshuffle
resi_connection: 1conv
num_feat: 64
hat_m
type: hat_m
img_size: 64
patch_size: 1
in_chans: 3
embed_dim: 180
depths: [6, 6, 6, 6, 6, 6]
num_heads: [6, 6, 6, 6, 6, 6]
window_size: 16
compress_ratio: 3
squeeze_factor: 30
conv_scale: 0.01
overlap_ratio: 0.5
mlp_ratio: 2.0
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
ape: false
patch_norm: true
use_checkpoint: false
img_range: 1.0
upsampler: pixelshuffle
resi_connection: 1conv
num_feat: 64
hat_s
type: hat_s
img_size: 64
patch_size: 1
in_chans: 3
embed_dim: 144
depths: [6, 6, 6, 6, 6, 6]
num_heads: [6, 6, 6, 6, 6, 6]
window_size: 16
compress_ratio: 24
squeeze_factor: 24
conv_scale: 0.01
overlap_ratio: 0.5
mlp_ratio: 2.0
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
ape: false
patch_norm: true
use_checkpoint: false
img_range: 1.0
upsampler: pixelshuffle
resi_connection: 1conv
num_feat: 64
HAT_iLN
hat_iln_l
type: hat_iln_l
img_size: 64
patch_size: 1
in_chans: 3
embed_dim: 180
depths: [6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6]
num_heads: [6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6]
window_size: 16
compress_ratio: 3
squeeze_factor: 30
conv_scale: 0.01
overlap_ratio: 0.5
mlp_ratio: 2.0
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
ape: false
patch_norm: true
use_checkpoint: false
upsampler: pixelshuffle
resi_connection: 1conv
num_feat: 64
hat_iln_l_aligned
type: hat_iln_l_aligned
img_size: 64
patch_size: 1
in_chans: 3
embed_dim: 192
depths: [6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6]
num_heads: [6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6]
window_size: 16
compress_ratio: 3
squeeze_factor: 30
conv_scale: 0.01
overlap_ratio: 0.5
mlp_ratio: 2.0
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
ape: false
patch_norm: true
use_checkpoint: false
upsampler: pixelshuffle
resi_connection: 1conv
num_feat: 64
hat_iln_m
type: hat_iln_m
img_size: 64
patch_size: 1
in_chans: 3
embed_dim: 180
depths: [6, 6, 6, 6, 6, 6]
num_heads: [6, 6, 6, 6, 6, 6]
window_size: 16
compress_ratio: 3
squeeze_factor: 30
conv_scale: 0.01
overlap_ratio: 0.5
mlp_ratio: 2.0
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
ape: false
patch_norm: true
use_checkpoint: false
upsampler: pixelshuffle
resi_connection: 1conv
num_feat: 64
hat_iln_m_aligned
type: hat_iln_m_aligned
img_size: 64
patch_size: 1
in_chans: 3
embed_dim: 192
depths: [6, 6, 6, 6, 6, 6]
num_heads: [6, 6, 6, 6, 6, 6]
window_size: 16
compress_ratio: 3
squeeze_factor: 30
conv_scale: 0.01
overlap_ratio: 0.5
mlp_ratio: 2.0
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
ape: false
patch_norm: true
use_checkpoint: false
upsampler: pixelshuffle
resi_connection: 1conv
num_feat: 64
hat_iln_s
type: hat_iln_s
img_size: 64
patch_size: 1
in_chans: 3
embed_dim: 144
depths: [6, 6, 6, 6, 6, 6]
num_heads: [6, 6, 6, 6, 6, 6]
window_size: 16
compress_ratio: 24
squeeze_factor: 24
conv_scale: 0.01
overlap_ratio: 0.5
mlp_ratio: 2.0
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
ape: false
patch_norm: true
use_checkpoint: false
upsampler: pixelshuffle
resi_connection: 1conv
num_feat: 64
HiT_SIR
hit_sir
type: hit_sir
in_chans: 3
img_size: 64
base_win_size: [8, 8]
img_range: 1.0
depths: [6, 6, 6, 6]
embed_dim: 60
num_heads: [6, 6, 6, 6]
expansion_factor: 2
resi_connection: 1conv
hier_win_ratios: [0.5, 1, 2, 4, 6, 8]
upsampler: pixelshuffledirect
HiT_SNG
hit_sng
type: hit_sng
in_chans: 3
img_size: 64
base_win_size: [8, 8]
img_range: 1.0
depths: [6, 6, 6, 6]
embed_dim: 60
num_heads: [6, 6, 6, 6]
expansion_factor: 2
resi_connection: 1conv
hier_win_ratios: [0.5, 1, 2, 4, 6, 8]
upsampler: pixelshuffledirect
HiT_SRF
hit_srf
type: hit_srf
in_chans: 3
img_size: 64
base_win_size: [8, 8]
img_range: 1.0
depths: [6, 6, 6, 6]
embed_dim: 60
num_heads: [6, 6, 6, 6]
expansion_factor: 2
resi_connection: 1conv
hier_win_ratios: [0.5, 1, 2, 4, 6, 8]
upsampler: pixelshuffledirect
LKFMixer
lkfmixer_b
type: lkfmixer_b
in_channels: 3
channels: 48
out_channels: 3
num_block: 8
large_kernel: 31
split_factor: 0.25
lkfmixer_l
type: lkfmixer_l
in_channels: 3
channels: 64
out_channels: 3
num_block: 12
large_kernel: 31
split_factor: 0.25
lkfmixer_t
type: lkfmixer_t
in_channels: 3
channels: 40
out_channels: 3
num_block: 6
large_kernel: 31
split_factor: 0.25
LMLT
lmlt_base
type: lmlt_base
dim: 60
n_blocks: 8
ffn_scale: 2.0
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.0
lmlt_large
type: lmlt_large
dim: 84
n_blocks: 8
ffn_scale: 2.0
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.0
lmlt_tiny
type: lmlt_tiny
dim: 36
n_blocks: 8
ffn_scale: 2.0
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.0
MAN
man
type: man
n_resblocks: 36
n_resgroups: 1
n_colors: 3
n_feats: 180
man_light
type: man_light
n_resblocks: 24
n_resgroups: 1
n_colors: 3
n_feats: 60
man_tiny
type: man_tiny
n_resblocks: 5
n_resgroups: 1
n_colors: 3
n_feats: 48
MetaGAN3
metagan3
type: metagan3
in_ch: 3
dims: [64, 128, 192, 256]
blocks: [2, 3, 5, 2]
MoESR2
moesr2
type: moesr2
in_ch: 3
out_ch: 3
dim: 64
n_blocks: 9
n_block: 4
expansion_factor: 2.6666666666666665
expansion_msg: 1.5
upsampler: pixelshuffledirect
upsample_dim: 64
MoSR
mosr
type: mosr
in_ch: 3
out_ch: 3
n_block: 24
dim: 64
upsampler: pixelshuffle
drop_path: 0.0
kernel_size: 7
expansion_ratio: 1.5
conv_ratio: 1.0
mosr_t
type: mosr_t
in_ch: 3
out_ch: 3
n_block: 5
dim: 48
upsampler: pixelshuffle
drop_path: 0.0
kernel_size: 7
expansion_ratio: 1.5
conv_ratio: 1.0
MoSRv2
mosrv2
type: mosrv2
in_ch: 3
n_block: 24
dim: 64
upsampler: pixelshuffledirect
expansion_ratio: 1.5
mid_dim: 32
unshuffle_mod: true
rms_norm: false
OmniSR
omnisr
type: omnisr
num_in_ch: 3
num_out_ch: 3
num_feat: 64
block_num: 1
pe: true
window_size: 8
res_num: 5
bias: true
PLKSR
plksr
type: plksr
dim: 64
n_blocks: 28
ccm_type: DCCM
kernel_size: 17
split_ratio: 0.25
lk_type: PLK
use_max_kernel: false
sparse_kernels: [5, 5, 5, 5]
sparse_dilations: [1, 2, 3, 4]
with_idt: false
use_ea: true
plksr_tiny
type: plksr_tiny
dim: 64
n_blocks: 12
ccm_type: DCCM
kernel_size: 13
split_ratio: 0.25
lk_type: PLK
use_max_kernel: false
sparse_kernels: [5, 5, 5, 5]
sparse_dilations: [1, 2, 3, 4]
with_idt: false
use_ea: false
RCAN
rcan
type: rcan
n_resgroups: 10
n_resblocks: 20
n_feats: 64
n_colors: 3
rgb_range: 255
norm: false
kernel_size: 3
reduction: 16
res_scale: 1
act_mode: relu
unshuffle_mod: false
rcan_l
type: rcan_l
n_resgroups: 10
n_resblocks: 20
n_feats: 96
n_colors: 3
rgb_range: 255
norm: false
kernel_size: 3
reduction: 16
res_scale: 1
act_mode: relu
unshuffle_mod: false
rcan_unshuffle
type: rcan_unshuffle
n_resgroups: 10
n_resblocks: 20
n_feats: 64
n_colors: 3
rgb_range: 255
norm: false
kernel_size: 3
reduction: 16
res_scale: 1
act_mode: relu
unshuffle_mod: true
RGT
rgt
type: rgt
img_size: 64
in_chans: 3
embed_dim: 180
depth: [6, 6, 6, 6, 6, 6, 6, 6]
num_heads: [6, 6, 6, 6, 6, 6, 6, 6]
mlp_ratio: 2.0
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
use_chk: false
img_range: 1.0
resi_connection: 1conv
split_size: [8, 32]
c_ratio: 0.5
rgt_s
type: rgt_s
img_size: 64
in_chans: 3
embed_dim: 180
depth: [6, 6, 6, 6, 6, 6]
num_heads: [6, 6, 6, 6, 6, 6]
mlp_ratio: 2.0
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
use_chk: false
img_range: 1.0
resi_connection: 1conv
split_size: [8, 32]
c_ratio: 0.5
RRDBNet
esrgan
type: esrgan
use_pixel_unshuffle: true
in_nc: 3
out_nc: 3
num_filters: 64
num_blocks: 23
esrgan_lite
type: esrgan_lite
use_pixel_unshuffle: true
in_nc: 3
out_nc: 3
num_filters: 32
num_blocks: 12
RTMoSR
rtmosr
type: rtmosr
dim: 32
ffn_expansion: 2
n_blocks: 2
unshuffle_mod: false
dccm: true
se: true
rtmosr_l
type: rtmosr_l
dim: 32
ffn_expansion: 2
n_blocks: 2
unshuffle_mod: true
dccm: true
se: true
rtmosr_ul
type: rtmosr_ul
dim: 32
ffn_expansion: 1.5
n_blocks: 2
unshuffle_mod: true
dccm: false
se: true
RealPLKSR
realplksr
type: realplksr
in_ch: 3
out_ch: 3
dim: 64
n_blocks: 28
kernel_size: 17
split_ratio: 0.25
use_ea: true
norm_groups: 4
dropout: 0
upsampler: pixelshuffle
layer_norm: true
realplksr_large
type: realplksr_large
in_ch: 3
out_ch: 3
dim: 96
n_blocks: 28
kernel_size: 17
split_ratio: 0.25
use_ea: true
norm_groups: 4
dropout: 0
upsampler: pixelshuffle
layer_norm: true
realplksr_tiny
type: realplksr_tiny
in_ch: 3
out_ch: 3
dim: 64
n_blocks: 12
kernel_size: 13
split_ratio: 0.25
use_ea: false
norm_groups: 4
dropout: 0
upsampler: pixelshuffle
layer_norm: true
SAFMN
safmn
type: safmn
dim: 36
n_blocks: 8
ffn_scale: 2.0
safmn_l
type: safmn_l
dim: 128
n_blocks: 16
ffn_scale: 2.0
SCUNet_aaf6aa
scunet_aaf6aa
type: scunet_aaf6aa
in_nc: 3
out_nc: 3
config: ~
dim: 64
drop_path_rate: 0.0
input_resolution: 256
residual: true
state: ~
SPAN
span
type: span
num_in_ch: 3
num_out_ch: 3
feature_channels: 52
bias: true
norm: false
img_range: 255.0
rgb_mean: [0.4488, 0.4371, 0.404]
span_f32
type: span_f32
num_in_ch: 3
num_out_ch: 3
feature_channels: 32
bias: true
norm: false
img_range: 255.0
rgb_mean: [0.4488, 0.4371, 0.404]
span_f64
type: span_f64
num_in_ch: 3
num_out_ch: 3
feature_channels: 64
bias: true
norm: false
img_range: 255.0
rgb_mean: [0.4488, 0.4371, 0.404]
span_f96
type: span_f96
num_in_ch: 3
num_out_ch: 3
feature_channels: 96
bias: true
norm: false
img_range: 255.0
rgb_mean: [0.4488, 0.4371, 0.404]
span_s
type: span_s
num_in_ch: 3
num_out_ch: 3
feature_channels: 48
bias: true
norm: false
img_range: 255.0
rgb_mean: [0.4488, 0.4371, 0.404]
SPANF
spanf
type: spanf
num_in_ch: 3
num_out_ch: 3
feature_channels: 32
bias: true
SRFormer
srformer
type: srformer
img_size: 48
patch_size: 1
in_chans: 3
embed_dim: 180
depths: [6, 6, 6, 6, 6, 6]
num_heads: [6, 6, 6, 6, 6, 6]
window_size: 24
mlp_ratio: 2.0
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
ape: false
patch_norm: true
use_checkpoint: false
img_range: 1.0
upsampler: pixelshuffle
resi_connection: 1conv
srformer_light
type: srformer_light
img_size: 64
patch_size: 1
in_chans: 3
embed_dim: 180
depths: [6, 6, 6, 6]
num_heads: [6, 6, 6, 6]
window_size: 16
mlp_ratio: 2.0
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
ape: false
patch_norm: true
use_checkpoint: false
img_range: 1.0
upsampler: pixelshuffledirect
resi_connection: 1conv
SRFormerv2
srformerv2
type: srformerv2
in_chans: 3
img_size: 64
window_size: 36
squeeze_dim: 60
img_range: 1
depths: [4, 4, 4, 4, 4, 4]
embed_dim: 240
num_heads: [8, 8, 8, 8, 8, 8]
mlp_ratio: 2
upsampler: pixelshuffle
resi_connection: 1conv
use_checkpoint: false
SRVGGNetCompact
compact
type: compact
num_in_ch: 3
num_out_ch: 3
num_feat: 64
num_conv: 16
act_type: prelu
learn_residual: true
superultracompact
type: superultracompact
num_in_ch: 3
num_out_ch: 3
num_feat: 24
num_conv: 8
act_type: prelu
learn_residual: true
ultracompact
type: ultracompact
num_in_ch: 3
num_out_ch: 3
num_feat: 64
num_conv: 8
act_type: prelu
learn_residual: true
Sebica
sebica
type: sebica
N: 16
sebica_mini
type: sebica_mini
N: 8
SeemoRe
seemore_t
type: seemore_t
in_chans: 3
num_experts: 3
num_layers: 6
embedding_dim: 36
img_range: 1.0
use_shuffle: true
global_kernel_size: 11
recursive: 2
lr_space: exp
topk: 1
SpanPlus
spanplus
type: spanplus
num_in_ch: 3
num_out_ch: 3
blocks: ~
feature_channels: 48
drop_rate: 0.0
upsampler: dys
spanplus_s
type: spanplus_s
num_in_ch: 3
num_out_ch: 3
blocks: [2]
feature_channels: 32
drop_rate: 0.0
upsampler: dys
spanplus_st
type: spanplus_st
num_in_ch: 3
num_out_ch: 3
blocks: ~
feature_channels: 48
drop_rate: 0.0
upsampler: ps
spanplus_sts
type: spanplus_sts
num_in_ch: 3
num_out_ch: 3
blocks: [2]
feature_channels: 32
drop_rate: 0.0
upsampler: ps
Swin2SR
swin2sr_l
type: swin2sr_l
img_size: 64
patch_size: 1
in_chans: 3
embed_dim: 240
depths: [6, 6, 6, 6, 6, 6, 6, 6, 6]
num_heads: [8, 8, 8, 8, 8, 8, 8, 8, 8]
window_size: 8
mlp_ratio: 2.0
qkv_bias: true
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
ape: false
patch_norm: true
use_checkpoint: false
img_range: 1.0
upsampler: nearest+conv
resi_connection: 3conv
swin2sr_m
type: swin2sr_m
img_size: 64
patch_size: 1
in_chans: 3
embed_dim: 180
depths: [6, 6, 6, 6, 6, 6]
num_heads: [6, 6, 6, 6, 6, 6]
window_size: 8
mlp_ratio: 2.0
qkv_bias: true
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
ape: false
patch_norm: true
use_checkpoint: false
img_range: 1.0
upsampler: pixelshuffle
resi_connection: 1conv
swin2sr_s
type: swin2sr_s
img_size: 64
patch_size: 1
in_chans: 3
embed_dim: 60
depths: [6, 6, 6, 6]
num_heads: [6, 6, 6, 6]
window_size: 8
mlp_ratio: 2.0
qkv_bias: true
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
ape: false
patch_norm: true
use_checkpoint: false
img_range: 1.0
upsampler: pixelshuffledirect
resi_connection: 1conv
SwinIR
swinir_l
type: swinir_l
img_size: 64
patch_size: 1
in_chans: 3
embed_dim: 240
depths: [6, 6, 6, 6, 6, 6, 6, 6, 6]
num_heads: [8, 8, 8, 8, 8, 8, 8, 8, 8]
window_size: 8
mlp_ratio: 2.0
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
ape: false
patch_norm: true
use_checkpoint: false
img_range: 1.0
upsampler: nearest+conv
resi_connection: 3conv
start_unshuffle: 1
swinir_m
type: swinir_m
img_size: 48
patch_size: 1
in_chans: 3
embed_dim: 180
depths: [6, 6, 6, 6, 6, 6]
num_heads: [6, 6, 6, 6, 6, 6]
window_size: 8
mlp_ratio: 2.0
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
ape: false
patch_norm: true
use_checkpoint: false
img_range: 1.0
upsampler: pixelshuffle
resi_connection: 1conv
start_unshuffle: 1
swinir_s
type: swinir_s
img_size: 64
patch_size: 1
in_chans: 3
embed_dim: 60
depths: [6, 6, 6, 6]
num_heads: [6, 6, 6, 6]
window_size: 8
mlp_ratio: 2.0
qkv_bias: true
qk_scale: ~
drop_rate: 0.0
attn_drop_rate: 0.0
drop_path_rate: 0.1
ape: false
patch_norm: true
use_checkpoint: false
img_range: 1.0
upsampler: pixelshuffledirect
resi_connection: 1conv
start_unshuffle: 1
TFDAT
tfdat
type: tfdat
num_in_ch: 3
num_out_ch: 3
embed_dim: 120
num_groups: 4
depth_per_group: 3
num_heads: 4
window_size: 8
ffn_expansion_ratio: 2.0
aim_reduction_ratio: 8
group_block_pattern: ~
drop_path_rate: 0.1
upsampler_type: transpose+conv
flow_base_ch: 32
TSCUNet
tscunet
type: tscunet
in_nc: 3
out_nc: 3
clip_size: 5
nb: 2
dim: 64
drop_path_rate: 0.0
input_resolution: 256
residual: true
sigma: false
state: ~
TSPANv2
temporalspanv2
type: temporalspanv2
num_in_ch: 3
num_out_ch: 3
num_frames: 5
feature_channels: 48
bias: false
num_blocks: 6
residual: true
img_range: 255.0
use_checkpoint: false
TemporalSPAN
temporalspan
type: temporalspan
num_in_ch: 3
num_out_ch: 3
num_frames: 5
feature_channels: 48
bias: true
history_channels: 12
Unet
unetsegmentation
type: unetsegmentation
encoder_name: resnet50
encoder_weights: imagenet
in_ch: 3
classes: 1
activation: ~
UpCunet4x
realcugan
type: realcugan
pro: false
fast: false
in_channels: 3
out_channels: 3
Discriminator architectures (network_d)
DUnet
dunet
type: dunet
num_in_ch: 3
num_feat: 64
MetaGan2
metagan2
type: metagan2
in_ch: 3
n_class: 1
dims: [32, 64, 128, 192]
blocks: [3, 3, 15, 3]
downs: [4, 2, 2, 2]
MultiscalePatchGANDiscriminatorSN
multiscalepatchgandiscriminatorsn
type: multiscalepatchgandiscriminatorsn
input_nc: 3
ndf: 64
n_layers: 3
use_sigmoid: false
num_d: 3
PatchGANDiscriminatorSN
patchgandiscriminatorsn
type: patchgandiscriminatorsn
input_nc: 3
ndf: 64
n_layers: 3
use_sigmoid: false
UNetDiscriminatorSN
unetdiscriminatorsn
type: unetdiscriminatorsn
num_in_ch: 3
num_feat: 64
skip_connection: true
VGGStyleDiscriminator
vggstylediscriminator
type: vggstylediscriminator
num_in_ch: 3
num_feat: 64
input_size: 128