Text-to-Image
Diffusers
Safetensors
English
StableDiffusionXLPipeline
Inference Endpoints
File size: 953 Bytes
a3a596a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
enable_conv = true

# An example for use different algo/settings in "full" preset
unet_target_module = [
  "Transformer2DModel", 
  "ResnetBlock2D", 
  "Downsample2D", 
  "Upsample2D",
]
unet_target_name = [
  ".*time_embed\\..+",
  ".*label_emb\\..+",
  ".*input_blocks\\.0.+",
  "^out\\..+"
]
text_encoder_target_module = [
  "CLIPAttention",
  "CLIPMLP",
]
text_encoder_target_name = [
  # "token_embedding", # not supported, Embedding module in CLIP
]


[module_algo_map]
  [module_algo_map.CrossAttention]  
    algo = "lokr"
    dim = 10000
    alpha = 10000
    factor = 4
  
  [module_algo_map.FeedForward]     # MLP Layer in UNet
    algo = "lokr"
    dim = 10000
    alpha = 10000
    factor = 2
  
  [module_algo_map.CLIPAttention]   # Attention Layer in TE
    algo = "lokr"
    dim = 10000
    alpha = 10000
    factor = 8
  
  [module_algo_map.CLIPMLP]   # MLP Layer in TE
    algo = "lokr"
    dim = 10000
    alpha = 10000
    factor = 8