initial commit
This commit is contained in:
Miller Liang
@@ -0,0 +1,9 @@
|
||||
# IC-Light-ComfyUI-Node
|
||||
|
||||
Original repo: https://github.com/lllyasviel/IC-Light/
|
||||
|
||||
Models: https://huggingface.co/lllyasviel/ic-light/tree/main
|
||||
|
||||
models go into `ComfyUI/models/unet`
|
||||
|
||||
|
||||
@@ -0,0 +1,3 @@
|
||||
from .nodes import NODE_CLASS_MAPPINGS, NODE_DISPLAY_NAME_MAPPINGS
|
||||
|
||||
__all__ = ["NODE_CLASS_MAPPINGS", "NODE_DISPLAY_NAME_MAPPINGS"]
|
||||
Executable
+25
@@ -0,0 +1,25 @@
|
||||
{
|
||||
"_name_or_path": "openai/clip-vit-large-patch14",
|
||||
"architectures": [
|
||||
"CLIPTextModel"
|
||||
],
|
||||
"attention_dropout": 0.0,
|
||||
"bos_token_id": 0,
|
||||
"dropout": 0.0,
|
||||
"eos_token_id": 2,
|
||||
"hidden_act": "quick_gelu",
|
||||
"hidden_size": 768,
|
||||
"initializer_factor": 1.0,
|
||||
"initializer_range": 0.02,
|
||||
"intermediate_size": 3072,
|
||||
"layer_norm_eps": 1e-05,
|
||||
"max_position_embeddings": 77,
|
||||
"model_type": "clip_text_model",
|
||||
"num_attention_heads": 12,
|
||||
"num_hidden_layers": 12,
|
||||
"pad_token_id": 1,
|
||||
"projection_dim": 768,
|
||||
"torch_dtype": "float32",
|
||||
"transformers_version": "4.22.0.dev0",
|
||||
"vocab_size": 49408
|
||||
}
|
||||
@@ -0,0 +1,171 @@
|
||||
{
|
||||
"_name_or_path": "clip-vit-large-patch14/",
|
||||
"architectures": [
|
||||
"CLIPModel"
|
||||
],
|
||||
"initializer_factor": 1.0,
|
||||
"logit_scale_init_value": 2.6592,
|
||||
"model_type": "clip",
|
||||
"projection_dim": 768,
|
||||
"text_config": {
|
||||
"_name_or_path": "",
|
||||
"add_cross_attention": false,
|
||||
"architectures": null,
|
||||
"attention_dropout": 0.0,
|
||||
"bad_words_ids": null,
|
||||
"bos_token_id": 0,
|
||||
"chunk_size_feed_forward": 0,
|
||||
"cross_attention_hidden_size": null,
|
||||
"decoder_start_token_id": null,
|
||||
"diversity_penalty": 0.0,
|
||||
"do_sample": false,
|
||||
"dropout": 0.0,
|
||||
"early_stopping": false,
|
||||
"encoder_no_repeat_ngram_size": 0,
|
||||
"eos_token_id": 2,
|
||||
"finetuning_task": null,
|
||||
"forced_bos_token_id": null,
|
||||
"forced_eos_token_id": null,
|
||||
"hidden_act": "quick_gelu",
|
||||
"hidden_size": 768,
|
||||
"id2label": {
|
||||
"0": "LABEL_0",
|
||||
"1": "LABEL_1"
|
||||
},
|
||||
"initializer_factor": 1.0,
|
||||
"initializer_range": 0.02,
|
||||
"intermediate_size": 3072,
|
||||
"is_decoder": false,
|
||||
"is_encoder_decoder": false,
|
||||
"label2id": {
|
||||
"LABEL_0": 0,
|
||||
"LABEL_1": 1
|
||||
},
|
||||
"layer_norm_eps": 1e-05,
|
||||
"length_penalty": 1.0,
|
||||
"max_length": 20,
|
||||
"max_position_embeddings": 77,
|
||||
"min_length": 0,
|
||||
"model_type": "clip_text_model",
|
||||
"no_repeat_ngram_size": 0,
|
||||
"num_attention_heads": 12,
|
||||
"num_beam_groups": 1,
|
||||
"num_beams": 1,
|
||||
"num_hidden_layers": 12,
|
||||
"num_return_sequences": 1,
|
||||
"output_attentions": false,
|
||||
"output_hidden_states": false,
|
||||
"output_scores": false,
|
||||
"pad_token_id": 1,
|
||||
"prefix": null,
|
||||
"problem_type": null,
|
||||
"projection_dim" : 768,
|
||||
"pruned_heads": {},
|
||||
"remove_invalid_values": false,
|
||||
"repetition_penalty": 1.0,
|
||||
"return_dict": true,
|
||||
"return_dict_in_generate": false,
|
||||
"sep_token_id": null,
|
||||
"task_specific_params": null,
|
||||
"temperature": 1.0,
|
||||
"tie_encoder_decoder": false,
|
||||
"tie_word_embeddings": true,
|
||||
"tokenizer_class": null,
|
||||
"top_k": 50,
|
||||
"top_p": 1.0,
|
||||
"torch_dtype": null,
|
||||
"torchscript": false,
|
||||
"transformers_version": "4.16.0.dev0",
|
||||
"use_bfloat16": false,
|
||||
"vocab_size": 49408
|
||||
},
|
||||
"text_config_dict": {
|
||||
"hidden_size": 768,
|
||||
"intermediate_size": 3072,
|
||||
"num_attention_heads": 12,
|
||||
"num_hidden_layers": 12,
|
||||
"projection_dim": 768
|
||||
},
|
||||
"torch_dtype": "float32",
|
||||
"transformers_version": null,
|
||||
"vision_config": {
|
||||
"_name_or_path": "",
|
||||
"add_cross_attention": false,
|
||||
"architectures": null,
|
||||
"attention_dropout": 0.0,
|
||||
"bad_words_ids": null,
|
||||
"bos_token_id": null,
|
||||
"chunk_size_feed_forward": 0,
|
||||
"cross_attention_hidden_size": null,
|
||||
"decoder_start_token_id": null,
|
||||
"diversity_penalty": 0.0,
|
||||
"do_sample": false,
|
||||
"dropout": 0.0,
|
||||
"early_stopping": false,
|
||||
"encoder_no_repeat_ngram_size": 0,
|
||||
"eos_token_id": null,
|
||||
"finetuning_task": null,
|
||||
"forced_bos_token_id": null,
|
||||
"forced_eos_token_id": null,
|
||||
"hidden_act": "quick_gelu",
|
||||
"hidden_size": 1024,
|
||||
"id2label": {
|
||||
"0": "LABEL_0",
|
||||
"1": "LABEL_1"
|
||||
},
|
||||
"image_size": 224,
|
||||
"initializer_factor": 1.0,
|
||||
"initializer_range": 0.02,
|
||||
"intermediate_size": 4096,
|
||||
"is_decoder": false,
|
||||
"is_encoder_decoder": false,
|
||||
"label2id": {
|
||||
"LABEL_0": 0,
|
||||
"LABEL_1": 1
|
||||
},
|
||||
"layer_norm_eps": 1e-05,
|
||||
"length_penalty": 1.0,
|
||||
"max_length": 20,
|
||||
"min_length": 0,
|
||||
"model_type": "clip_vision_model",
|
||||
"no_repeat_ngram_size": 0,
|
||||
"num_attention_heads": 16,
|
||||
"num_beam_groups": 1,
|
||||
"num_beams": 1,
|
||||
"num_hidden_layers": 24,
|
||||
"num_return_sequences": 1,
|
||||
"output_attentions": false,
|
||||
"output_hidden_states": false,
|
||||
"output_scores": false,
|
||||
"pad_token_id": null,
|
||||
"patch_size": 14,
|
||||
"prefix": null,
|
||||
"problem_type": null,
|
||||
"projection_dim" : 768,
|
||||
"pruned_heads": {},
|
||||
"remove_invalid_values": false,
|
||||
"repetition_penalty": 1.0,
|
||||
"return_dict": true,
|
||||
"return_dict_in_generate": false,
|
||||
"sep_token_id": null,
|
||||
"task_specific_params": null,
|
||||
"temperature": 1.0,
|
||||
"tie_encoder_decoder": false,
|
||||
"tie_word_embeddings": true,
|
||||
"tokenizer_class": null,
|
||||
"top_k": 50,
|
||||
"top_p": 1.0,
|
||||
"torch_dtype": null,
|
||||
"torchscript": false,
|
||||
"transformers_version": "4.16.0.dev0",
|
||||
"use_bfloat16": false
|
||||
},
|
||||
"vision_config_dict": {
|
||||
"hidden_size": 1024,
|
||||
"intermediate_size": 4096,
|
||||
"num_attention_heads": 16,
|
||||
"num_hidden_layers": 24,
|
||||
"patch_size": 14,
|
||||
"projection_dim": 768
|
||||
}
|
||||
}
|
||||
File diff suppressed because it is too large
Load Diff
@@ -0,0 +1,19 @@
|
||||
{
|
||||
"crop_size": 224,
|
||||
"do_center_crop": true,
|
||||
"do_normalize": true,
|
||||
"do_resize": true,
|
||||
"feature_extractor_type": "CLIPFeatureExtractor",
|
||||
"image_mean": [
|
||||
0.48145466,
|
||||
0.4578275,
|
||||
0.40821073
|
||||
],
|
||||
"image_std": [
|
||||
0.26862954,
|
||||
0.26130258,
|
||||
0.27577711
|
||||
],
|
||||
"resample": 3,
|
||||
"size": 224
|
||||
}
|
||||
@@ -0,0 +1 @@
|
||||
{"bos_token": {"content": "<|startoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, "eos_token": {"content": "<|endoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, "unk_token": {"content": "<|endoftext|>", "single_word": false, "lstrip": false, "rstrip": false, "normalized": true}, "pad_token": "<|endoftext|>"}
|
||||
File diff suppressed because it is too large
Load Diff
@@ -0,0 +1,34 @@
|
||||
{
|
||||
"unk_token": {
|
||||
"content": "<|endoftext|>",
|
||||
"single_word": false,
|
||||
"lstrip": false,
|
||||
"rstrip": false,
|
||||
"normalized": true,
|
||||
"__type": "AddedToken"
|
||||
},
|
||||
"bos_token": {
|
||||
"content": "<|startoftext|>",
|
||||
"single_word": false,
|
||||
"lstrip": false,
|
||||
"rstrip": false,
|
||||
"normalized": true,
|
||||
"__type": "AddedToken"
|
||||
},
|
||||
"eos_token": {
|
||||
"content": "<|endoftext|>",
|
||||
"single_word": false,
|
||||
"lstrip": false,
|
||||
"rstrip": false,
|
||||
"normalized": true,
|
||||
"__type": "AddedToken"
|
||||
},
|
||||
"pad_token": "<|endoftext|>",
|
||||
"add_prefix_space": false,
|
||||
"errors": "replace",
|
||||
"do_lower_case": true,
|
||||
"name_or_path": "openai/clip-vit-base-patch32",
|
||||
"model_max_length": 77,
|
||||
"special_tokens_map_file": "./special_tokens_map.json",
|
||||
"tokenizer_class": "CLIPTokenizer"
|
||||
}
|
||||
File diff suppressed because one or more lines are too long
Executable
+34
@@ -0,0 +1,34 @@
|
||||
{
|
||||
"add_prefix_space": false,
|
||||
"bos_token": {
|
||||
"__type": "AddedToken",
|
||||
"content": "<|startoftext|>",
|
||||
"lstrip": false,
|
||||
"normalized": true,
|
||||
"rstrip": false,
|
||||
"single_word": false
|
||||
},
|
||||
"do_lower_case": true,
|
||||
"eos_token": {
|
||||
"__type": "AddedToken",
|
||||
"content": "<|endoftext|>",
|
||||
"lstrip": false,
|
||||
"normalized": true,
|
||||
"rstrip": false,
|
||||
"single_word": false
|
||||
},
|
||||
"errors": "replace",
|
||||
"model_max_length": 77,
|
||||
"name_or_path": "openai/clip-vit-large-patch14",
|
||||
"pad_token": "<|endoftext|>",
|
||||
"special_tokens_map_file": "./special_tokens_map.json",
|
||||
"tokenizer_class": "CLIPTokenizer",
|
||||
"unk_token": {
|
||||
"__type": "AddedToken",
|
||||
"content": "<|endoftext|>",
|
||||
"lstrip": false,
|
||||
"normalized": true,
|
||||
"rstrip": false,
|
||||
"single_word": false
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,70 @@
|
||||
model:
|
||||
base_learning_rate: 1.0e-04
|
||||
target: ldm.models.diffusion.ddpm.LatentDiffusion
|
||||
params:
|
||||
linear_start: 0.00085
|
||||
linear_end: 0.0120
|
||||
num_timesteps_cond: 1
|
||||
log_every_t: 200
|
||||
timesteps: 1000
|
||||
first_stage_key: "jpg"
|
||||
cond_stage_key: "txt"
|
||||
image_size: 64
|
||||
channels: 4
|
||||
cond_stage_trainable: false # Note: different from the one we trained before
|
||||
conditioning_key: crossattn
|
||||
monitor: val/loss_simple_ema
|
||||
scale_factor: 0.18215
|
||||
use_ema: False
|
||||
|
||||
scheduler_config: # 10000 warmup steps
|
||||
target: ldm.lr_scheduler.LambdaLinearScheduler
|
||||
params:
|
||||
warm_up_steps: [ 10000 ]
|
||||
cycle_lengths: [ 10000000000000 ] # incredibly large number to prevent corner cases
|
||||
f_start: [ 1.e-6 ]
|
||||
f_max: [ 1. ]
|
||||
f_min: [ 1. ]
|
||||
|
||||
unet_config:
|
||||
target: ldm.modules.diffusionmodules.openaimodel.UNetModel
|
||||
params:
|
||||
image_size: 32 # unused
|
||||
in_channels: 4
|
||||
out_channels: 4
|
||||
model_channels: 320
|
||||
attention_resolutions: [ 4, 2, 1 ]
|
||||
num_res_blocks: 2
|
||||
channel_mult: [ 1, 2, 4, 4 ]
|
||||
num_heads: 8
|
||||
use_spatial_transformer: True
|
||||
transformer_depth: 1
|
||||
context_dim: 768
|
||||
use_checkpoint: True
|
||||
legacy: False
|
||||
|
||||
first_stage_config:
|
||||
target: ldm.models.autoencoder.AutoencoderKL
|
||||
params:
|
||||
embed_dim: 4
|
||||
monitor: val/rec_loss
|
||||
ddconfig:
|
||||
double_z: true
|
||||
z_channels: 4
|
||||
resolution: 256
|
||||
in_channels: 3
|
||||
out_ch: 3
|
||||
ch: 128
|
||||
ch_mult:
|
||||
- 1
|
||||
- 2
|
||||
- 4
|
||||
- 4
|
||||
num_res_blocks: 2
|
||||
attn_resolutions: []
|
||||
dropout: 0.0
|
||||
lossconfig:
|
||||
target: torch.nn.Identity
|
||||
|
||||
cond_stage_config:
|
||||
target: ldm.modules.encoders.modules.FrozenCLIPEmbedder
|
||||
@@ -0,0 +1,850 @@
|
||||
{
|
||||
"last_node_id": 36,
|
||||
"last_link_id": 63,
|
||||
"nodes": [
|
||||
{
|
||||
"id": 5,
|
||||
"type": "diffusers_model_loader",
|
||||
"pos": [
|
||||
996,
|
||||
292
|
||||
],
|
||||
"size": {
|
||||
"0": 267,
|
||||
"1": 66
|
||||
},
|
||||
"flags": {},
|
||||
"order": 3,
|
||||
"mode": 0,
|
||||
"inputs": [
|
||||
{
|
||||
"name": "model",
|
||||
"type": "MODEL",
|
||||
"link": 47,
|
||||
"slot_index": 0
|
||||
},
|
||||
{
|
||||
"name": "clip",
|
||||
"type": "CLIP",
|
||||
"link": 2
|
||||
},
|
||||
{
|
||||
"name": "vae",
|
||||
"type": "VAE",
|
||||
"link": 3
|
||||
}
|
||||
],
|
||||
"outputs": [
|
||||
{
|
||||
"name": "diffusers_model",
|
||||
"type": "DIFFUSERSMODEL",
|
||||
"links": [
|
||||
26
|
||||
],
|
||||
"shape": 3,
|
||||
"slot_index": 0
|
||||
}
|
||||
],
|
||||
"properties": {
|
||||
"Node name for S&R": "diffusers_model_loader"
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": 22,
|
||||
"type": "LoadICLightUnetDiffusers",
|
||||
"pos": [
|
||||
1343,
|
||||
301
|
||||
],
|
||||
"size": {
|
||||
"0": 447,
|
||||
"1": 58
|
||||
},
|
||||
"flags": {},
|
||||
"order": 6,
|
||||
"mode": 0,
|
||||
"inputs": [
|
||||
{
|
||||
"name": "diffusersmodel",
|
||||
"type": "DIFFUSERSMODEL",
|
||||
"link": 26
|
||||
}
|
||||
],
|
||||
"outputs": [
|
||||
{
|
||||
"name": "DIFFUSERSMODEL",
|
||||
"type": "DIFFUSERSMODEL",
|
||||
"links": [
|
||||
58
|
||||
],
|
||||
"shape": 3,
|
||||
"slot_index": 0
|
||||
}
|
||||
],
|
||||
"properties": {
|
||||
"Node name for S&R": "LoadICLightUnetDiffusers"
|
||||
},
|
||||
"widgets_values": [
|
||||
"iclight_sd15_fc.safetensors"
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": 8,
|
||||
"type": "VAEEncode",
|
||||
"pos": [
|
||||
1022,
|
||||
469
|
||||
],
|
||||
"size": {
|
||||
"0": 210,
|
||||
"1": 46
|
||||
},
|
||||
"flags": {},
|
||||
"order": 4,
|
||||
"mode": 0,
|
||||
"inputs": [
|
||||
{
|
||||
"name": "pixels",
|
||||
"type": "IMAGE",
|
||||
"link": 41,
|
||||
"slot_index": 0
|
||||
},
|
||||
{
|
||||
"name": "vae",
|
||||
"type": "VAE",
|
||||
"link": 8
|
||||
}
|
||||
],
|
||||
"outputs": [
|
||||
{
|
||||
"name": "LATENT",
|
||||
"type": "LATENT",
|
||||
"links": [
|
||||
59
|
||||
],
|
||||
"shape": 3,
|
||||
"slot_index": 0
|
||||
}
|
||||
],
|
||||
"properties": {
|
||||
"Node name for S&R": "VAEEncode"
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": 25,
|
||||
"type": "VAEEncode",
|
||||
"pos": [
|
||||
1055,
|
||||
649
|
||||
],
|
||||
"size": {
|
||||
"0": 210,
|
||||
"1": 46
|
||||
},
|
||||
"flags": {},
|
||||
"order": 10,
|
||||
"mode": 0,
|
||||
"inputs": [
|
||||
{
|
||||
"name": "pixels",
|
||||
"type": "IMAGE",
|
||||
"link": 56,
|
||||
"slot_index": 0
|
||||
},
|
||||
{
|
||||
"name": "vae",
|
||||
"type": "VAE",
|
||||
"link": 32,
|
||||
"slot_index": 1
|
||||
}
|
||||
],
|
||||
"outputs": [
|
||||
{
|
||||
"name": "LATENT",
|
||||
"type": "LATENT",
|
||||
"links": [
|
||||
60
|
||||
],
|
||||
"shape": 3,
|
||||
"slot_index": 0
|
||||
}
|
||||
],
|
||||
"properties": {
|
||||
"Node name for S&R": "VAEEncode"
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": 34,
|
||||
"type": "GrowMaskWithBlur",
|
||||
"pos": [
|
||||
620,
|
||||
1030
|
||||
],
|
||||
"size": {
|
||||
"0": 315,
|
||||
"1": 246
|
||||
},
|
||||
"flags": {},
|
||||
"order": 7,
|
||||
"mode": 0,
|
||||
"inputs": [
|
||||
{
|
||||
"name": "mask",
|
||||
"type": "MASK",
|
||||
"link": 52
|
||||
}
|
||||
],
|
||||
"outputs": [
|
||||
{
|
||||
"name": "mask",
|
||||
"type": "MASK",
|
||||
"links": [
|
||||
53
|
||||
],
|
||||
"shape": 3,
|
||||
"slot_index": 0
|
||||
},
|
||||
{
|
||||
"name": "mask_inverted",
|
||||
"type": "MASK",
|
||||
"links": null,
|
||||
"shape": 3
|
||||
}
|
||||
],
|
||||
"properties": {
|
||||
"Node name for S&R": "GrowMaskWithBlur"
|
||||
},
|
||||
"widgets_values": [
|
||||
0,
|
||||
0,
|
||||
true,
|
||||
false,
|
||||
59.400000000000006,
|
||||
1,
|
||||
1,
|
||||
false
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": 9,
|
||||
"type": "LoadImage",
|
||||
"pos": [
|
||||
104,
|
||||
474
|
||||
],
|
||||
"size": {
|
||||
"0": 315,
|
||||
"1": 314
|
||||
},
|
||||
"flags": {},
|
||||
"order": 0,
|
||||
"mode": 0,
|
||||
"outputs": [
|
||||
{
|
||||
"name": "IMAGE",
|
||||
"type": "IMAGE",
|
||||
"links": [
|
||||
40
|
||||
],
|
||||
"shape": 3,
|
||||
"slot_index": 0
|
||||
},
|
||||
{
|
||||
"name": "MASK",
|
||||
"type": "MASK",
|
||||
"links": null,
|
||||
"shape": 3
|
||||
}
|
||||
],
|
||||
"properties": {
|
||||
"Node name for S&R": "LoadImage"
|
||||
},
|
||||
"widgets_values": [
|
||||
"oldman.jpg",
|
||||
"image"
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": 28,
|
||||
"type": "ImageResize+",
|
||||
"pos": [
|
||||
482,
|
||||
479
|
||||
],
|
||||
"size": {
|
||||
"0": 315,
|
||||
"1": 218
|
||||
},
|
||||
"flags": {},
|
||||
"order": 2,
|
||||
"mode": 0,
|
||||
"inputs": [
|
||||
{
|
||||
"name": "image",
|
||||
"type": "IMAGE",
|
||||
"link": 40
|
||||
}
|
||||
],
|
||||
"outputs": [
|
||||
{
|
||||
"name": "IMAGE",
|
||||
"type": "IMAGE",
|
||||
"links": [
|
||||
41
|
||||
],
|
||||
"shape": 3,
|
||||
"slot_index": 0
|
||||
},
|
||||
{
|
||||
"name": "width",
|
||||
"type": "INT",
|
||||
"links": [
|
||||
54
|
||||
],
|
||||
"shape": 3
|
||||
},
|
||||
{
|
||||
"name": "height",
|
||||
"type": "INT",
|
||||
"links": [
|
||||
55
|
||||
],
|
||||
"shape": 3
|
||||
}
|
||||
],
|
||||
"properties": {
|
||||
"Node name for S&R": "ImageResize+"
|
||||
},
|
||||
"widgets_values": [
|
||||
512,
|
||||
512,
|
||||
"nearest",
|
||||
false,
|
||||
"always",
|
||||
8
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": 30,
|
||||
"type": "MaskToImage",
|
||||
"pos": [
|
||||
663,
|
||||
806
|
||||
],
|
||||
"size": {
|
||||
"0": 210,
|
||||
"1": 26
|
||||
},
|
||||
"flags": {},
|
||||
"order": 9,
|
||||
"mode": 0,
|
||||
"inputs": [
|
||||
{
|
||||
"name": "mask",
|
||||
"type": "MASK",
|
||||
"link": 50
|
||||
}
|
||||
],
|
||||
"outputs": [
|
||||
{
|
||||
"name": "IMAGE",
|
||||
"type": "IMAGE",
|
||||
"links": [
|
||||
56,
|
||||
57
|
||||
],
|
||||
"shape": 3,
|
||||
"slot_index": 0
|
||||
}
|
||||
],
|
||||
"properties": {
|
||||
"Node name for S&R": "MaskToImage"
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": 29,
|
||||
"type": "CreateShapeMask",
|
||||
"pos": [
|
||||
262,
|
||||
912
|
||||
],
|
||||
"size": {
|
||||
"0": 315,
|
||||
"1": 270
|
||||
},
|
||||
"flags": {},
|
||||
"order": 5,
|
||||
"mode": 0,
|
||||
"inputs": [
|
||||
{
|
||||
"name": "frame_width",
|
||||
"type": "INT",
|
||||
"link": 54,
|
||||
"widget": {
|
||||
"name": "frame_width"
|
||||
},
|
||||
"slot_index": 0
|
||||
},
|
||||
{
|
||||
"name": "frame_height",
|
||||
"type": "INT",
|
||||
"link": 55,
|
||||
"widget": {
|
||||
"name": "frame_height"
|
||||
},
|
||||
"slot_index": 1
|
||||
}
|
||||
],
|
||||
"outputs": [
|
||||
{
|
||||
"name": "mask",
|
||||
"type": "MASK",
|
||||
"links": [
|
||||
52
|
||||
],
|
||||
"shape": 3,
|
||||
"slot_index": 0
|
||||
},
|
||||
{
|
||||
"name": "mask_inverted",
|
||||
"type": "MASK",
|
||||
"links": null,
|
||||
"shape": 3
|
||||
}
|
||||
],
|
||||
"properties": {
|
||||
"Node name for S&R": "CreateShapeMask"
|
||||
},
|
||||
"widgets_values": [
|
||||
"circle",
|
||||
1,
|
||||
256,
|
||||
256,
|
||||
0,
|
||||
512,
|
||||
512,
|
||||
256,
|
||||
256
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": 32,
|
||||
"type": "RemapMaskRange",
|
||||
"pos": [
|
||||
610,
|
||||
890
|
||||
],
|
||||
"size": {
|
||||
"0": 315,
|
||||
"1": 82
|
||||
},
|
||||
"flags": {},
|
||||
"order": 8,
|
||||
"mode": 0,
|
||||
"inputs": [
|
||||
{
|
||||
"name": "mask",
|
||||
"type": "MASK",
|
||||
"link": 53
|
||||
}
|
||||
],
|
||||
"outputs": [
|
||||
{
|
||||
"name": "mask",
|
||||
"type": "MASK",
|
||||
"links": [
|
||||
50
|
||||
],
|
||||
"shape": 3,
|
||||
"slot_index": 0
|
||||
}
|
||||
],
|
||||
"properties": {
|
||||
"Node name for S&R": "RemapMaskRange"
|
||||
},
|
||||
"widgets_values": [
|
||||
0,
|
||||
0.9
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": 33,
|
||||
"type": "PreviewImage",
|
||||
"pos": [
|
||||
989,
|
||||
893
|
||||
],
|
||||
"size": {
|
||||
"0": 342,
|
||||
"1": 347
|
||||
},
|
||||
"flags": {},
|
||||
"order": 11,
|
||||
"mode": 0,
|
||||
"inputs": [
|
||||
{
|
||||
"name": "images",
|
||||
"type": "IMAGE",
|
||||
"link": 57
|
||||
}
|
||||
],
|
||||
"properties": {
|
||||
"Node name for S&R": "PreviewImage"
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": 10,
|
||||
"type": "PreviewImage",
|
||||
"pos": [
|
||||
2076,
|
||||
347
|
||||
],
|
||||
"size": {
|
||||
"0": 526.656982421875,
|
||||
"1": 580.8809814453125
|
||||
},
|
||||
"flags": {},
|
||||
"order": 14,
|
||||
"mode": 0,
|
||||
"inputs": [
|
||||
{
|
||||
"name": "images",
|
||||
"type": "IMAGE",
|
||||
"link": 63
|
||||
}
|
||||
],
|
||||
"properties": {
|
||||
"Node name for S&R": "PreviewImage"
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": 35,
|
||||
"type": "iclight_diffusers_sampler",
|
||||
"pos": [
|
||||
1372,
|
||||
434
|
||||
],
|
||||
"size": {
|
||||
"0": 400,
|
||||
"1": 406
|
||||
},
|
||||
"flags": {},
|
||||
"order": 12,
|
||||
"mode": 0,
|
||||
"inputs": [
|
||||
{
|
||||
"name": "diffusers_model",
|
||||
"type": "DIFFUSERSMODEL",
|
||||
"link": 58
|
||||
},
|
||||
{
|
||||
"name": "latent",
|
||||
"type": "LATENT",
|
||||
"link": 59
|
||||
},
|
||||
{
|
||||
"name": "bg_latent",
|
||||
"type": "LATENT",
|
||||
"link": 60
|
||||
}
|
||||
],
|
||||
"outputs": [
|
||||
{
|
||||
"name": "samples",
|
||||
"type": "LATENT",
|
||||
"links": [
|
||||
61
|
||||
],
|
||||
"shape": 3,
|
||||
"slot_index": 0
|
||||
}
|
||||
],
|
||||
"properties": {
|
||||
"Node name for S&R": "iclight_diffusers_sampler"
|
||||
},
|
||||
"widgets_values": [
|
||||
512,
|
||||
512,
|
||||
25,
|
||||
2,
|
||||
0.9,
|
||||
819223299872531,
|
||||
"randomize",
|
||||
"DPMSolverMultistepScheduler",
|
||||
"positive",
|
||||
"negative",
|
||||
false,
|
||||
true
|
||||
]
|
||||
},
|
||||
{
|
||||
"id": 36,
|
||||
"type": "VAEDecode",
|
||||
"pos": [
|
||||
1817,
|
||||
427
|
||||
],
|
||||
"size": {
|
||||
"0": 210,
|
||||
"1": 46
|
||||
},
|
||||
"flags": {},
|
||||
"order": 13,
|
||||
"mode": 0,
|
||||
"inputs": [
|
||||
{
|
||||
"name": "samples",
|
||||
"type": "LATENT",
|
||||
"link": 61
|
||||
},
|
||||
{
|
||||
"name": "vae",
|
||||
"type": "VAE",
|
||||
"link": 62,
|
||||
"slot_index": 1
|
||||
}
|
||||
],
|
||||
"outputs": [
|
||||
{
|
||||
"name": "IMAGE",
|
||||
"type": "IMAGE",
|
||||
"links": [
|
||||
63
|
||||
],
|
||||
"shape": 3,
|
||||
"slot_index": 0
|
||||
}
|
||||
],
|
||||
"properties": {
|
||||
"Node name for S&R": "VAEDecode"
|
||||
}
|
||||
},
|
||||
{
|
||||
"id": 6,
|
||||
"type": "CheckpointLoaderSimple",
|
||||
"pos": [
|
||||
563,
|
||||
285
|
||||
],
|
||||
"size": {
|
||||
"0": 315,
|
||||
"1": 98
|
||||
},
|
||||
"flags": {},
|
||||
"order": 1,
|
||||
"mode": 0,
|
||||
"outputs": [
|
||||
{
|
||||
"name": "MODEL",
|
||||
"type": "MODEL",
|
||||
"links": [
|
||||
47
|
||||
],
|
||||
"shape": 3,
|
||||
"slot_index": 0
|
||||
},
|
||||
{
|
||||
"name": "CLIP",
|
||||
"type": "CLIP",
|
||||
"links": [
|
||||
2
|
||||
],
|
||||
"shape": 3,
|
||||
"slot_index": 1
|
||||
},
|
||||
{
|
||||
"name": "VAE",
|
||||
"type": "VAE",
|
||||
"links": [
|
||||
3,
|
||||
8,
|
||||
32,
|
||||
62
|
||||
],
|
||||
"shape": 3,
|
||||
"slot_index": 2
|
||||
}
|
||||
],
|
||||
"properties": {
|
||||
"Node name for S&R": "CheckpointLoaderSimple"
|
||||
},
|
||||
"widgets_values": [
|
||||
"1_5\\photon_v1.safetensors"
|
||||
]
|
||||
}
|
||||
],
|
||||
"links": [
|
||||
[
|
||||
2,
|
||||
6,
|
||||
1,
|
||||
5,
|
||||
1,
|
||||
"CLIP"
|
||||
],
|
||||
[
|
||||
3,
|
||||
6,
|
||||
2,
|
||||
5,
|
||||
2,
|
||||
"VAE"
|
||||
],
|
||||
[
|
||||
8,
|
||||
6,
|
||||
2,
|
||||
8,
|
||||
1,
|
||||
"VAE"
|
||||
],
|
||||
[
|
||||
26,
|
||||
5,
|
||||
0,
|
||||
22,
|
||||
0,
|
||||
"DIFFUSERSMODEL"
|
||||
],
|
||||
[
|
||||
32,
|
||||
6,
|
||||
2,
|
||||
25,
|
||||
1,
|
||||
"VAE"
|
||||
],
|
||||
[
|
||||
40,
|
||||
9,
|
||||
0,
|
||||
28,
|
||||
0,
|
||||
"IMAGE"
|
||||
],
|
||||
[
|
||||
41,
|
||||
28,
|
||||
0,
|
||||
8,
|
||||
0,
|
||||
"IMAGE"
|
||||
],
|
||||
[
|
||||
47,
|
||||
6,
|
||||
0,
|
||||
5,
|
||||
0,
|
||||
"MODEL"
|
||||
],
|
||||
[
|
||||
50,
|
||||
32,
|
||||
0,
|
||||
30,
|
||||
0,
|
||||
"MASK"
|
||||
],
|
||||
[
|
||||
52,
|
||||
29,
|
||||
0,
|
||||
34,
|
||||
0,
|
||||
"MASK"
|
||||
],
|
||||
[
|
||||
53,
|
||||
34,
|
||||
0,
|
||||
32,
|
||||
0,
|
||||
"MASK"
|
||||
],
|
||||
[
|
||||
54,
|
||||
28,
|
||||
1,
|
||||
29,
|
||||
0,
|
||||
"INT"
|
||||
],
|
||||
[
|
||||
55,
|
||||
28,
|
||||
2,
|
||||
29,
|
||||
1,
|
||||
"INT"
|
||||
],
|
||||
[
|
||||
56,
|
||||
30,
|
||||
0,
|
||||
25,
|
||||
0,
|
||||
"IMAGE"
|
||||
],
|
||||
[
|
||||
57,
|
||||
30,
|
||||
0,
|
||||
33,
|
||||
0,
|
||||
"IMAGE"
|
||||
],
|
||||
[
|
||||
58,
|
||||
22,
|
||||
0,
|
||||
35,
|
||||
0,
|
||||
"DIFFUSERSMODEL"
|
||||
],
|
||||
[
|
||||
59,
|
||||
8,
|
||||
0,
|
||||
35,
|
||||
1,
|
||||
"LATENT"
|
||||
],
|
||||
[
|
||||
60,
|
||||
25,
|
||||
0,
|
||||
35,
|
||||
2,
|
||||
"LATENT"
|
||||
],
|
||||
[
|
||||
61,
|
||||
35,
|
||||
0,
|
||||
36,
|
||||
0,
|
||||
"LATENT"
|
||||
],
|
||||
[
|
||||
62,
|
||||
6,
|
||||
2,
|
||||
36,
|
||||
1,
|
||||
"VAE"
|
||||
],
|
||||
[
|
||||
63,
|
||||
36,
|
||||
0,
|
||||
10,
|
||||
0,
|
||||
"IMAGE"
|
||||
]
|
||||
],
|
||||
"groups": [],
|
||||
"config": {},
|
||||
"extra": {},
|
||||
"version": 0.4
|
||||
}
|
||||
File diff suppressed because it is too large
Load Diff
@@ -0,0 +1,3 @@
|
||||
from .hidiffusion import apply_hidiffusion, remove_hidiffusion
|
||||
|
||||
__all__ = ["apply_hidiffusion", "remove_hidiffusion"]
|
||||
File diff suppressed because it is too large
Load Diff
@@ -0,0 +1,685 @@
|
||||
conv_in
|
||||
time_proj
|
||||
time_embedding
|
||||
time_embedding.linear_1
|
||||
time_embedding.act
|
||||
time_embedding.linear_2
|
||||
down_blocks
|
||||
down_blocks.0
|
||||
down_blocks.0.attentions
|
||||
down_blocks.0.attentions.0
|
||||
down_blocks.0.attentions.0.norm
|
||||
down_blocks.0.attentions.0.proj_in
|
||||
down_blocks.0.attentions.0.transformer_blocks
|
||||
down_blocks.0.attentions.0.transformer_blocks.0
|
||||
down_blocks.0.attentions.0.transformer_blocks.0.norm1
|
||||
down_blocks.0.attentions.0.transformer_blocks.0.attn1
|
||||
down_blocks.0.attentions.0.transformer_blocks.0.attn1.to_q
|
||||
down_blocks.0.attentions.0.transformer_blocks.0.attn1.to_k
|
||||
down_blocks.0.attentions.0.transformer_blocks.0.attn1.to_v
|
||||
down_blocks.0.attentions.0.transformer_blocks.0.attn1.to_out
|
||||
down_blocks.0.attentions.0.transformer_blocks.0.attn1.to_out.0
|
||||
down_blocks.0.attentions.0.transformer_blocks.0.attn1.to_out.1
|
||||
down_blocks.0.attentions.0.transformer_blocks.0.norm2
|
||||
down_blocks.0.attentions.0.transformer_blocks.0.attn2
|
||||
down_blocks.0.attentions.0.transformer_blocks.0.attn2.to_q
|
||||
down_blocks.0.attentions.0.transformer_blocks.0.attn2.to_k
|
||||
down_blocks.0.attentions.0.transformer_blocks.0.attn2.to_v
|
||||
down_blocks.0.attentions.0.transformer_blocks.0.attn2.to_out
|
||||
down_blocks.0.attentions.0.transformer_blocks.0.attn2.to_out.0
|
||||
down_blocks.0.attentions.0.transformer_blocks.0.attn2.to_out.1
|
||||
down_blocks.0.attentions.0.transformer_blocks.0.norm3
|
||||
down_blocks.0.attentions.0.transformer_blocks.0.ff
|
||||
down_blocks.0.attentions.0.transformer_blocks.0.ff.net
|
||||
down_blocks.0.attentions.0.transformer_blocks.0.ff.net.0
|
||||
down_blocks.0.attentions.0.transformer_blocks.0.ff.net.0.proj
|
||||
down_blocks.0.attentions.0.transformer_blocks.0.ff.net.1
|
||||
down_blocks.0.attentions.0.transformer_blocks.0.ff.net.2
|
||||
down_blocks.0.attentions.0.proj_out
|
||||
down_blocks.0.attentions.1
|
||||
down_blocks.0.attentions.1.norm
|
||||
down_blocks.0.attentions.1.proj_in
|
||||
down_blocks.0.attentions.1.transformer_blocks
|
||||
down_blocks.0.attentions.1.transformer_blocks.0
|
||||
down_blocks.0.attentions.1.transformer_blocks.0.norm1
|
||||
down_blocks.0.attentions.1.transformer_blocks.0.attn1
|
||||
down_blocks.0.attentions.1.transformer_blocks.0.attn1.to_q
|
||||
down_blocks.0.attentions.1.transformer_blocks.0.attn1.to_k
|
||||
down_blocks.0.attentions.1.transformer_blocks.0.attn1.to_v
|
||||
down_blocks.0.attentions.1.transformer_blocks.0.attn1.to_out
|
||||
down_blocks.0.attentions.1.transformer_blocks.0.attn1.to_out.0
|
||||
down_blocks.0.attentions.1.transformer_blocks.0.attn1.to_out.1
|
||||
down_blocks.0.attentions.1.transformer_blocks.0.norm2
|
||||
down_blocks.0.attentions.1.transformer_blocks.0.attn2
|
||||
down_blocks.0.attentions.1.transformer_blocks.0.attn2.to_q
|
||||
down_blocks.0.attentions.1.transformer_blocks.0.attn2.to_k
|
||||
down_blocks.0.attentions.1.transformer_blocks.0.attn2.to_v
|
||||
down_blocks.0.attentions.1.transformer_blocks.0.attn2.to_out
|
||||
down_blocks.0.attentions.1.transformer_blocks.0.attn2.to_out.0
|
||||
down_blocks.0.attentions.1.transformer_blocks.0.attn2.to_out.1
|
||||
down_blocks.0.attentions.1.transformer_blocks.0.norm3
|
||||
down_blocks.0.attentions.1.transformer_blocks.0.ff
|
||||
down_blocks.0.attentions.1.transformer_blocks.0.ff.net
|
||||
down_blocks.0.attentions.1.transformer_blocks.0.ff.net.0
|
||||
down_blocks.0.attentions.1.transformer_blocks.0.ff.net.0.proj
|
||||
down_blocks.0.attentions.1.transformer_blocks.0.ff.net.1
|
||||
down_blocks.0.attentions.1.transformer_blocks.0.ff.net.2
|
||||
down_blocks.0.attentions.1.proj_out
|
||||
down_blocks.0.resnets
|
||||
down_blocks.0.resnets.0
|
||||
down_blocks.0.resnets.0.norm1
|
||||
down_blocks.0.resnets.0.conv1
|
||||
down_blocks.0.resnets.0.time_emb_proj
|
||||
down_blocks.0.resnets.0.norm2
|
||||
down_blocks.0.resnets.0.dropout
|
||||
down_blocks.0.resnets.0.conv2
|
||||
down_blocks.0.resnets.1
|
||||
down_blocks.0.resnets.1.norm1
|
||||
down_blocks.0.resnets.1.conv1
|
||||
down_blocks.0.resnets.1.time_emb_proj
|
||||
down_blocks.0.resnets.1.norm2
|
||||
down_blocks.0.resnets.1.dropout
|
||||
down_blocks.0.resnets.1.conv2
|
||||
down_blocks.0.downsamplers
|
||||
down_blocks.0.downsamplers.0
|
||||
down_blocks.0.downsamplers.0.conv
|
||||
down_blocks.1
|
||||
down_blocks.1.attentions
|
||||
down_blocks.1.attentions.0
|
||||
down_blocks.1.attentions.0.norm
|
||||
down_blocks.1.attentions.0.proj_in
|
||||
down_blocks.1.attentions.0.transformer_blocks
|
||||
down_blocks.1.attentions.0.transformer_blocks.0
|
||||
down_blocks.1.attentions.0.transformer_blocks.0.norm1
|
||||
down_blocks.1.attentions.0.transformer_blocks.0.attn1
|
||||
down_blocks.1.attentions.0.transformer_blocks.0.attn1.to_q
|
||||
down_blocks.1.attentions.0.transformer_blocks.0.attn1.to_k
|
||||
down_blocks.1.attentions.0.transformer_blocks.0.attn1.to_v
|
||||
down_blocks.1.attentions.0.transformer_blocks.0.attn1.to_out
|
||||
down_blocks.1.attentions.0.transformer_blocks.0.attn1.to_out.0
|
||||
down_blocks.1.attentions.0.transformer_blocks.0.attn1.to_out.1
|
||||
down_blocks.1.attentions.0.transformer_blocks.0.norm2
|
||||
down_blocks.1.attentions.0.transformer_blocks.0.attn2
|
||||
down_blocks.1.attentions.0.transformer_blocks.0.attn2.to_q
|
||||
down_blocks.1.attentions.0.transformer_blocks.0.attn2.to_k
|
||||
down_blocks.1.attentions.0.transformer_blocks.0.attn2.to_v
|
||||
down_blocks.1.attentions.0.transformer_blocks.0.attn2.to_out
|
||||
down_blocks.1.attentions.0.transformer_blocks.0.attn2.to_out.0
|
||||
down_blocks.1.attentions.0.transformer_blocks.0.attn2.to_out.1
|
||||
down_blocks.1.attentions.0.transformer_blocks.0.norm3
|
||||
down_blocks.1.attentions.0.transformer_blocks.0.ff
|
||||
down_blocks.1.attentions.0.transformer_blocks.0.ff.net
|
||||
down_blocks.1.attentions.0.transformer_blocks.0.ff.net.0
|
||||
down_blocks.1.attentions.0.transformer_blocks.0.ff.net.0.proj
|
||||
down_blocks.1.attentions.0.transformer_blocks.0.ff.net.1
|
||||
down_blocks.1.attentions.0.transformer_blocks.0.ff.net.2
|
||||
down_blocks.1.attentions.0.proj_out
|
||||
down_blocks.1.attentions.1
|
||||
down_blocks.1.attentions.1.norm
|
||||
down_blocks.1.attentions.1.proj_in
|
||||
down_blocks.1.attentions.1.transformer_blocks
|
||||
down_blocks.1.attentions.1.transformer_blocks.0
|
||||
down_blocks.1.attentions.1.transformer_blocks.0.norm1
|
||||
down_blocks.1.attentions.1.transformer_blocks.0.attn1
|
||||
down_blocks.1.attentions.1.transformer_blocks.0.attn1.to_q
|
||||
down_blocks.1.attentions.1.transformer_blocks.0.attn1.to_k
|
||||
down_blocks.1.attentions.1.transformer_blocks.0.attn1.to_v
|
||||
down_blocks.1.attentions.1.transformer_blocks.0.attn1.to_out
|
||||
down_blocks.1.attentions.1.transformer_blocks.0.attn1.to_out.0
|
||||
down_blocks.1.attentions.1.transformer_blocks.0.attn1.to_out.1
|
||||
down_blocks.1.attentions.1.transformer_blocks.0.norm2
|
||||
down_blocks.1.attentions.1.transformer_blocks.0.attn2
|
||||
down_blocks.1.attentions.1.transformer_blocks.0.attn2.to_q
|
||||
down_blocks.1.attentions.1.transformer_blocks.0.attn2.to_k
|
||||
down_blocks.1.attentions.1.transformer_blocks.0.attn2.to_v
|
||||
down_blocks.1.attentions.1.transformer_blocks.0.attn2.to_out
|
||||
down_blocks.1.attentions.1.transformer_blocks.0.attn2.to_out.0
|
||||
down_blocks.1.attentions.1.transformer_blocks.0.attn2.to_out.1
|
||||
down_blocks.1.attentions.1.transformer_blocks.0.norm3
|
||||
down_blocks.1.attentions.1.transformer_blocks.0.ff
|
||||
down_blocks.1.attentions.1.transformer_blocks.0.ff.net
|
||||
down_blocks.1.attentions.1.transformer_blocks.0.ff.net.0
|
||||
down_blocks.1.attentions.1.transformer_blocks.0.ff.net.0.proj
|
||||
down_blocks.1.attentions.1.transformer_blocks.0.ff.net.1
|
||||
down_blocks.1.attentions.1.transformer_blocks.0.ff.net.2
|
||||
down_blocks.1.attentions.1.proj_out
|
||||
down_blocks.1.resnets
|
||||
down_blocks.1.resnets.0
|
||||
down_blocks.1.resnets.0.norm1
|
||||
down_blocks.1.resnets.0.conv1
|
||||
down_blocks.1.resnets.0.time_emb_proj
|
||||
down_blocks.1.resnets.0.norm2
|
||||
down_blocks.1.resnets.0.dropout
|
||||
down_blocks.1.resnets.0.conv2
|
||||
down_blocks.1.resnets.0.conv_shortcut
|
||||
down_blocks.1.resnets.1
|
||||
down_blocks.1.resnets.1.norm1
|
||||
down_blocks.1.resnets.1.conv1
|
||||
down_blocks.1.resnets.1.time_emb_proj
|
||||
down_blocks.1.resnets.1.norm2
|
||||
down_blocks.1.resnets.1.dropout
|
||||
down_blocks.1.resnets.1.conv2
|
||||
down_blocks.1.downsamplers
|
||||
down_blocks.1.downsamplers.0
|
||||
down_blocks.1.downsamplers.0.conv
|
||||
down_blocks.2
|
||||
down_blocks.2.attentions
|
||||
down_blocks.2.attentions.0
|
||||
down_blocks.2.attentions.0.norm
|
||||
down_blocks.2.attentions.0.proj_in
|
||||
down_blocks.2.attentions.0.transformer_blocks
|
||||
down_blocks.2.attentions.0.transformer_blocks.0
|
||||
down_blocks.2.attentions.0.transformer_blocks.0.norm1
|
||||
down_blocks.2.attentions.0.transformer_blocks.0.attn1
|
||||
down_blocks.2.attentions.0.transformer_blocks.0.attn1.to_q
|
||||
down_blocks.2.attentions.0.transformer_blocks.0.attn1.to_k
|
||||
down_blocks.2.attentions.0.transformer_blocks.0.attn1.to_v
|
||||
down_blocks.2.attentions.0.transformer_blocks.0.attn1.to_out
|
||||
down_blocks.2.attentions.0.transformer_blocks.0.attn1.to_out.0
|
||||
down_blocks.2.attentions.0.transformer_blocks.0.attn1.to_out.1
|
||||
down_blocks.2.attentions.0.transformer_blocks.0.norm2
|
||||
down_blocks.2.attentions.0.transformer_blocks.0.attn2
|
||||
down_blocks.2.attentions.0.transformer_blocks.0.attn2.to_q
|
||||
down_blocks.2.attentions.0.transformer_blocks.0.attn2.to_k
|
||||
down_blocks.2.attentions.0.transformer_blocks.0.attn2.to_v
|
||||
down_blocks.2.attentions.0.transformer_blocks.0.attn2.to_out
|
||||
down_blocks.2.attentions.0.transformer_blocks.0.attn2.to_out.0
|
||||
down_blocks.2.attentions.0.transformer_blocks.0.attn2.to_out.1
|
||||
down_blocks.2.attentions.0.transformer_blocks.0.norm3
|
||||
down_blocks.2.attentions.0.transformer_blocks.0.ff
|
||||
down_blocks.2.attentions.0.transformer_blocks.0.ff.net
|
||||
down_blocks.2.attentions.0.transformer_blocks.0.ff.net.0
|
||||
down_blocks.2.attentions.0.transformer_blocks.0.ff.net.0.proj
|
||||
down_blocks.2.attentions.0.transformer_blocks.0.ff.net.1
|
||||
down_blocks.2.attentions.0.transformer_blocks.0.ff.net.2
|
||||
down_blocks.2.attentions.0.proj_out
|
||||
down_blocks.2.attentions.1
|
||||
down_blocks.2.attentions.1.norm
|
||||
down_blocks.2.attentions.1.proj_in
|
||||
down_blocks.2.attentions.1.transformer_blocks
|
||||
down_blocks.2.attentions.1.transformer_blocks.0
|
||||
down_blocks.2.attentions.1.transformer_blocks.0.norm1
|
||||
down_blocks.2.attentions.1.transformer_blocks.0.attn1
|
||||
down_blocks.2.attentions.1.transformer_blocks.0.attn1.to_q
|
||||
down_blocks.2.attentions.1.transformer_blocks.0.attn1.to_k
|
||||
down_blocks.2.attentions.1.transformer_blocks.0.attn1.to_v
|
||||
down_blocks.2.attentions.1.transformer_blocks.0.attn1.to_out
|
||||
down_blocks.2.attentions.1.transformer_blocks.0.attn1.to_out.0
|
||||
down_blocks.2.attentions.1.transformer_blocks.0.attn1.to_out.1
|
||||
down_blocks.2.attentions.1.transformer_blocks.0.norm2
|
||||
down_blocks.2.attentions.1.transformer_blocks.0.attn2
|
||||
down_blocks.2.attentions.1.transformer_blocks.0.attn2.to_q
|
||||
down_blocks.2.attentions.1.transformer_blocks.0.attn2.to_k
|
||||
down_blocks.2.attentions.1.transformer_blocks.0.attn2.to_v
|
||||
down_blocks.2.attentions.1.transformer_blocks.0.attn2.to_out
|
||||
down_blocks.2.attentions.1.transformer_blocks.0.attn2.to_out.0
|
||||
down_blocks.2.attentions.1.transformer_blocks.0.attn2.to_out.1
|
||||
down_blocks.2.attentions.1.transformer_blocks.0.norm3
|
||||
down_blocks.2.attentions.1.transformer_blocks.0.ff
|
||||
down_blocks.2.attentions.1.transformer_blocks.0.ff.net
|
||||
down_blocks.2.attentions.1.transformer_blocks.0.ff.net.0
|
||||
down_blocks.2.attentions.1.transformer_blocks.0.ff.net.0.proj
|
||||
down_blocks.2.attentions.1.transformer_blocks.0.ff.net.1
|
||||
down_blocks.2.attentions.1.transformer_blocks.0.ff.net.2
|
||||
down_blocks.2.attentions.1.proj_out
|
||||
down_blocks.2.resnets
|
||||
down_blocks.2.resnets.0
|
||||
down_blocks.2.resnets.0.norm1
|
||||
down_blocks.2.resnets.0.conv1
|
||||
down_blocks.2.resnets.0.time_emb_proj
|
||||
down_blocks.2.resnets.0.norm2
|
||||
down_blocks.2.resnets.0.dropout
|
||||
down_blocks.2.resnets.0.conv2
|
||||
down_blocks.2.resnets.0.conv_shortcut
|
||||
down_blocks.2.resnets.1
|
||||
down_blocks.2.resnets.1.norm1
|
||||
down_blocks.2.resnets.1.conv1
|
||||
down_blocks.2.resnets.1.time_emb_proj
|
||||
down_blocks.2.resnets.1.norm2
|
||||
down_blocks.2.resnets.1.dropout
|
||||
down_blocks.2.resnets.1.conv2
|
||||
down_blocks.2.downsamplers
|
||||
down_blocks.2.downsamplers.0
|
||||
down_blocks.2.downsamplers.0.conv
|
||||
down_blocks.3
|
||||
down_blocks.3.resnets
|
||||
down_blocks.3.resnets.0
|
||||
down_blocks.3.resnets.0.norm1
|
||||
down_blocks.3.resnets.0.conv1
|
||||
down_blocks.3.resnets.0.time_emb_proj
|
||||
down_blocks.3.resnets.0.norm2
|
||||
down_blocks.3.resnets.0.dropout
|
||||
down_blocks.3.resnets.0.conv2
|
||||
down_blocks.3.resnets.1
|
||||
down_blocks.3.resnets.1.norm1
|
||||
down_blocks.3.resnets.1.conv1
|
||||
down_blocks.3.resnets.1.time_emb_proj
|
||||
down_blocks.3.resnets.1.norm2
|
||||
down_blocks.3.resnets.1.dropout
|
||||
down_blocks.3.resnets.1.conv2
|
||||
up_blocks
|
||||
up_blocks.0
|
||||
up_blocks.0.resnets
|
||||
up_blocks.0.resnets.0
|
||||
up_blocks.0.resnets.0.norm1
|
||||
up_blocks.0.resnets.0.conv1
|
||||
up_blocks.0.resnets.0.time_emb_proj
|
||||
up_blocks.0.resnets.0.norm2
|
||||
up_blocks.0.resnets.0.dropout
|
||||
up_blocks.0.resnets.0.conv2
|
||||
up_blocks.0.resnets.0.conv_shortcut
|
||||
up_blocks.0.resnets.1
|
||||
up_blocks.0.resnets.1.norm1
|
||||
up_blocks.0.resnets.1.conv1
|
||||
up_blocks.0.resnets.1.time_emb_proj
|
||||
up_blocks.0.resnets.1.norm2
|
||||
up_blocks.0.resnets.1.dropout
|
||||
up_blocks.0.resnets.1.conv2
|
||||
up_blocks.0.resnets.1.conv_shortcut
|
||||
up_blocks.0.resnets.2
|
||||
up_blocks.0.resnets.2.norm1
|
||||
up_blocks.0.resnets.2.conv1
|
||||
up_blocks.0.resnets.2.time_emb_proj
|
||||
up_blocks.0.resnets.2.norm2
|
||||
up_blocks.0.resnets.2.dropout
|
||||
up_blocks.0.resnets.2.conv2
|
||||
up_blocks.0.resnets.2.conv_shortcut
|
||||
up_blocks.0.upsamplers
|
||||
up_blocks.0.upsamplers.0
|
||||
up_blocks.0.upsamplers.0.conv
|
||||
up_blocks.1
|
||||
up_blocks.1.attentions
|
||||
up_blocks.1.attentions.0
|
||||
up_blocks.1.attentions.0.norm
|
||||
up_blocks.1.attentions.0.proj_in
|
||||
up_blocks.1.attentions.0.transformer_blocks
|
||||
up_blocks.1.attentions.0.transformer_blocks.0
|
||||
up_blocks.1.attentions.0.transformer_blocks.0.norm1
|
||||
up_blocks.1.attentions.0.transformer_blocks.0.attn1
|
||||
up_blocks.1.attentions.0.transformer_blocks.0.attn1.to_q
|
||||
up_blocks.1.attentions.0.transformer_blocks.0.attn1.to_k
|
||||
up_blocks.1.attentions.0.transformer_blocks.0.attn1.to_v
|
||||
up_blocks.1.attentions.0.transformer_blocks.0.attn1.to_out
|
||||
up_blocks.1.attentions.0.transformer_blocks.0.attn1.to_out.0
|
||||
up_blocks.1.attentions.0.transformer_blocks.0.attn1.to_out.1
|
||||
up_blocks.1.attentions.0.transformer_blocks.0.norm2
|
||||
up_blocks.1.attentions.0.transformer_blocks.0.attn2
|
||||
up_blocks.1.attentions.0.transformer_blocks.0.attn2.to_q
|
||||
up_blocks.1.attentions.0.transformer_blocks.0.attn2.to_k
|
||||
up_blocks.1.attentions.0.transformer_blocks.0.attn2.to_v
|
||||
up_blocks.1.attentions.0.transformer_blocks.0.attn2.to_out
|
||||
up_blocks.1.attentions.0.transformer_blocks.0.attn2.to_out.0
|
||||
up_blocks.1.attentions.0.transformer_blocks.0.attn2.to_out.1
|
||||
up_blocks.1.attentions.0.transformer_blocks.0.norm3
|
||||
up_blocks.1.attentions.0.transformer_blocks.0.ff
|
||||
up_blocks.1.attentions.0.transformer_blocks.0.ff.net
|
||||
up_blocks.1.attentions.0.transformer_blocks.0.ff.net.0
|
||||
up_blocks.1.attentions.0.transformer_blocks.0.ff.net.0.proj
|
||||
up_blocks.1.attentions.0.transformer_blocks.0.ff.net.1
|
||||
up_blocks.1.attentions.0.transformer_blocks.0.ff.net.2
|
||||
up_blocks.1.attentions.0.proj_out
|
||||
up_blocks.1.attentions.1
|
||||
up_blocks.1.attentions.1.norm
|
||||
up_blocks.1.attentions.1.proj_in
|
||||
up_blocks.1.attentions.1.transformer_blocks
|
||||
up_blocks.1.attentions.1.transformer_blocks.0
|
||||
up_blocks.1.attentions.1.transformer_blocks.0.norm1
|
||||
up_blocks.1.attentions.1.transformer_blocks.0.attn1
|
||||
up_blocks.1.attentions.1.transformer_blocks.0.attn1.to_q
|
||||
up_blocks.1.attentions.1.transformer_blocks.0.attn1.to_k
|
||||
up_blocks.1.attentions.1.transformer_blocks.0.attn1.to_v
|
||||
up_blocks.1.attentions.1.transformer_blocks.0.attn1.to_out
|
||||
up_blocks.1.attentions.1.transformer_blocks.0.attn1.to_out.0
|
||||
up_blocks.1.attentions.1.transformer_blocks.0.attn1.to_out.1
|
||||
up_blocks.1.attentions.1.transformer_blocks.0.norm2
|
||||
up_blocks.1.attentions.1.transformer_blocks.0.attn2
|
||||
up_blocks.1.attentions.1.transformer_blocks.0.attn2.to_q
|
||||
up_blocks.1.attentions.1.transformer_blocks.0.attn2.to_k
|
||||
up_blocks.1.attentions.1.transformer_blocks.0.attn2.to_v
|
||||
up_blocks.1.attentions.1.transformer_blocks.0.attn2.to_out
|
||||
up_blocks.1.attentions.1.transformer_blocks.0.attn2.to_out.0
|
||||
up_blocks.1.attentions.1.transformer_blocks.0.attn2.to_out.1
|
||||
up_blocks.1.attentions.1.transformer_blocks.0.norm3
|
||||
up_blocks.1.attentions.1.transformer_blocks.0.ff
|
||||
up_blocks.1.attentions.1.transformer_blocks.0.ff.net
|
||||
up_blocks.1.attentions.1.transformer_blocks.0.ff.net.0
|
||||
up_blocks.1.attentions.1.transformer_blocks.0.ff.net.0.proj
|
||||
up_blocks.1.attentions.1.transformer_blocks.0.ff.net.1
|
||||
up_blocks.1.attentions.1.transformer_blocks.0.ff.net.2
|
||||
up_blocks.1.attentions.1.proj_out
|
||||
up_blocks.1.attentions.2
|
||||
up_blocks.1.attentions.2.norm
|
||||
up_blocks.1.attentions.2.proj_in
|
||||
up_blocks.1.attentions.2.transformer_blocks
|
||||
up_blocks.1.attentions.2.transformer_blocks.0
|
||||
up_blocks.1.attentions.2.transformer_blocks.0.norm1
|
||||
up_blocks.1.attentions.2.transformer_blocks.0.attn1
|
||||
up_blocks.1.attentions.2.transformer_blocks.0.attn1.to_q
|
||||
up_blocks.1.attentions.2.transformer_blocks.0.attn1.to_k
|
||||
up_blocks.1.attentions.2.transformer_blocks.0.attn1.to_v
|
||||
up_blocks.1.attentions.2.transformer_blocks.0.attn1.to_out
|
||||
up_blocks.1.attentions.2.transformer_blocks.0.attn1.to_out.0
|
||||
up_blocks.1.attentions.2.transformer_blocks.0.attn1.to_out.1
|
||||
up_blocks.1.attentions.2.transformer_blocks.0.norm2
|
||||
up_blocks.1.attentions.2.transformer_blocks.0.attn2
|
||||
up_blocks.1.attentions.2.transformer_blocks.0.attn2.to_q
|
||||
up_blocks.1.attentions.2.transformer_blocks.0.attn2.to_k
|
||||
up_blocks.1.attentions.2.transformer_blocks.0.attn2.to_v
|
||||
up_blocks.1.attentions.2.transformer_blocks.0.attn2.to_out
|
||||
up_blocks.1.attentions.2.transformer_blocks.0.attn2.to_out.0
|
||||
up_blocks.1.attentions.2.transformer_blocks.0.attn2.to_out.1
|
||||
up_blocks.1.attentions.2.transformer_blocks.0.norm3
|
||||
up_blocks.1.attentions.2.transformer_blocks.0.ff
|
||||
up_blocks.1.attentions.2.transformer_blocks.0.ff.net
|
||||
up_blocks.1.attentions.2.transformer_blocks.0.ff.net.0
|
||||
up_blocks.1.attentions.2.transformer_blocks.0.ff.net.0.proj
|
||||
up_blocks.1.attentions.2.transformer_blocks.0.ff.net.1
|
||||
up_blocks.1.attentions.2.transformer_blocks.0.ff.net.2
|
||||
up_blocks.1.attentions.2.proj_out
|
||||
up_blocks.1.resnets
|
||||
up_blocks.1.resnets.0
|
||||
up_blocks.1.resnets.0.norm1
|
||||
up_blocks.1.resnets.0.conv1
|
||||
up_blocks.1.resnets.0.time_emb_proj
|
||||
up_blocks.1.resnets.0.norm2
|
||||
up_blocks.1.resnets.0.dropout
|
||||
up_blocks.1.resnets.0.conv2
|
||||
up_blocks.1.resnets.0.conv_shortcut
|
||||
up_blocks.1.resnets.1
|
||||
up_blocks.1.resnets.1.norm1
|
||||
up_blocks.1.resnets.1.conv1
|
||||
up_blocks.1.resnets.1.time_emb_proj
|
||||
up_blocks.1.resnets.1.norm2
|
||||
up_blocks.1.resnets.1.dropout
|
||||
up_blocks.1.resnets.1.conv2
|
||||
up_blocks.1.resnets.1.conv_shortcut
|
||||
up_blocks.1.resnets.2
|
||||
up_blocks.1.resnets.2.norm1
|
||||
up_blocks.1.resnets.2.conv1
|
||||
up_blocks.1.resnets.2.time_emb_proj
|
||||
up_blocks.1.resnets.2.norm2
|
||||
up_blocks.1.resnets.2.dropout
|
||||
up_blocks.1.resnets.2.conv2
|
||||
up_blocks.1.resnets.2.conv_shortcut
|
||||
up_blocks.1.upsamplers
|
||||
up_blocks.1.upsamplers.0
|
||||
up_blocks.1.upsamplers.0.conv
|
||||
up_blocks.2
|
||||
up_blocks.2.attentions
|
||||
up_blocks.2.attentions.0
|
||||
up_blocks.2.attentions.0.norm
|
||||
up_blocks.2.attentions.0.proj_in
|
||||
up_blocks.2.attentions.0.transformer_blocks
|
||||
up_blocks.2.attentions.0.transformer_blocks.0
|
||||
up_blocks.2.attentions.0.transformer_blocks.0.norm1
|
||||
up_blocks.2.attentions.0.transformer_blocks.0.attn1
|
||||
up_blocks.2.attentions.0.transformer_blocks.0.attn1.to_q
|
||||
up_blocks.2.attentions.0.transformer_blocks.0.attn1.to_k
|
||||
up_blocks.2.attentions.0.transformer_blocks.0.attn1.to_v
|
||||
up_blocks.2.attentions.0.transformer_blocks.0.attn1.to_out
|
||||
up_blocks.2.attentions.0.transformer_blocks.0.attn1.to_out.0
|
||||
up_blocks.2.attentions.0.transformer_blocks.0.attn1.to_out.1
|
||||
up_blocks.2.attentions.0.transformer_blocks.0.norm2
|
||||
up_blocks.2.attentions.0.transformer_blocks.0.attn2
|
||||
up_blocks.2.attentions.0.transformer_blocks.0.attn2.to_q
|
||||
up_blocks.2.attentions.0.transformer_blocks.0.attn2.to_k
|
||||
up_blocks.2.attentions.0.transformer_blocks.0.attn2.to_v
|
||||
up_blocks.2.attentions.0.transformer_blocks.0.attn2.to_out
|
||||
up_blocks.2.attentions.0.transformer_blocks.0.attn2.to_out.0
|
||||
up_blocks.2.attentions.0.transformer_blocks.0.attn2.to_out.1
|
||||
up_blocks.2.attentions.0.transformer_blocks.0.norm3
|
||||
up_blocks.2.attentions.0.transformer_blocks.0.ff
|
||||
up_blocks.2.attentions.0.transformer_blocks.0.ff.net
|
||||
up_blocks.2.attentions.0.transformer_blocks.0.ff.net.0
|
||||
up_blocks.2.attentions.0.transformer_blocks.0.ff.net.0.proj
|
||||
up_blocks.2.attentions.0.transformer_blocks.0.ff.net.1
|
||||
up_blocks.2.attentions.0.transformer_blocks.0.ff.net.2
|
||||
up_blocks.2.attentions.0.proj_out
|
||||
up_blocks.2.attentions.1
|
||||
up_blocks.2.attentions.1.norm
|
||||
up_blocks.2.attentions.1.proj_in
|
||||
up_blocks.2.attentions.1.transformer_blocks
|
||||
up_blocks.2.attentions.1.transformer_blocks.0
|
||||
up_blocks.2.attentions.1.transformer_blocks.0.norm1
|
||||
up_blocks.2.attentions.1.transformer_blocks.0.attn1
|
||||
up_blocks.2.attentions.1.transformer_blocks.0.attn1.to_q
|
||||
up_blocks.2.attentions.1.transformer_blocks.0.attn1.to_k
|
||||
up_blocks.2.attentions.1.transformer_blocks.0.attn1.to_v
|
||||
up_blocks.2.attentions.1.transformer_blocks.0.attn1.to_out
|
||||
up_blocks.2.attentions.1.transformer_blocks.0.attn1.to_out.0
|
||||
up_blocks.2.attentions.1.transformer_blocks.0.attn1.to_out.1
|
||||
up_blocks.2.attentions.1.transformer_blocks.0.norm2
|
||||
up_blocks.2.attentions.1.transformer_blocks.0.attn2
|
||||
up_blocks.2.attentions.1.transformer_blocks.0.attn2.to_q
|
||||
up_blocks.2.attentions.1.transformer_blocks.0.attn2.to_k
|
||||
up_blocks.2.attentions.1.transformer_blocks.0.attn2.to_v
|
||||
up_blocks.2.attentions.1.transformer_blocks.0.attn2.to_out
|
||||
up_blocks.2.attentions.1.transformer_blocks.0.attn2.to_out.0
|
||||
up_blocks.2.attentions.1.transformer_blocks.0.attn2.to_out.1
|
||||
up_blocks.2.attentions.1.transformer_blocks.0.norm3
|
||||
up_blocks.2.attentions.1.transformer_blocks.0.ff
|
||||
up_blocks.2.attentions.1.transformer_blocks.0.ff.net
|
||||
up_blocks.2.attentions.1.transformer_blocks.0.ff.net.0
|
||||
up_blocks.2.attentions.1.transformer_blocks.0.ff.net.0.proj
|
||||
up_blocks.2.attentions.1.transformer_blocks.0.ff.net.1
|
||||
up_blocks.2.attentions.1.transformer_blocks.0.ff.net.2
|
||||
up_blocks.2.attentions.1.proj_out
|
||||
up_blocks.2.attentions.2
|
||||
up_blocks.2.attentions.2.norm
|
||||
up_blocks.2.attentions.2.proj_in
|
||||
up_blocks.2.attentions.2.transformer_blocks
|
||||
up_blocks.2.attentions.2.transformer_blocks.0
|
||||
up_blocks.2.attentions.2.transformer_blocks.0.norm1
|
||||
up_blocks.2.attentions.2.transformer_blocks.0.attn1
|
||||
up_blocks.2.attentions.2.transformer_blocks.0.attn1.to_q
|
||||
up_blocks.2.attentions.2.transformer_blocks.0.attn1.to_k
|
||||
up_blocks.2.attentions.2.transformer_blocks.0.attn1.to_v
|
||||
up_blocks.2.attentions.2.transformer_blocks.0.attn1.to_out
|
||||
up_blocks.2.attentions.2.transformer_blocks.0.attn1.to_out.0
|
||||
up_blocks.2.attentions.2.transformer_blocks.0.attn1.to_out.1
|
||||
up_blocks.2.attentions.2.transformer_blocks.0.norm2
|
||||
up_blocks.2.attentions.2.transformer_blocks.0.attn2
|
||||
up_blocks.2.attentions.2.transformer_blocks.0.attn2.to_q
|
||||
up_blocks.2.attentions.2.transformer_blocks.0.attn2.to_k
|
||||
up_blocks.2.attentions.2.transformer_blocks.0.attn2.to_v
|
||||
up_blocks.2.attentions.2.transformer_blocks.0.attn2.to_out
|
||||
up_blocks.2.attentions.2.transformer_blocks.0.attn2.to_out.0
|
||||
up_blocks.2.attentions.2.transformer_blocks.0.attn2.to_out.1
|
||||
up_blocks.2.attentions.2.transformer_blocks.0.norm3
|
||||
up_blocks.2.attentions.2.transformer_blocks.0.ff
|
||||
up_blocks.2.attentions.2.transformer_blocks.0.ff.net
|
||||
up_blocks.2.attentions.2.transformer_blocks.0.ff.net.0
|
||||
up_blocks.2.attentions.2.transformer_blocks.0.ff.net.0.proj
|
||||
up_blocks.2.attentions.2.transformer_blocks.0.ff.net.1
|
||||
up_blocks.2.attentions.2.transformer_blocks.0.ff.net.2
|
||||
up_blocks.2.attentions.2.proj_out
|
||||
up_blocks.2.resnets
|
||||
up_blocks.2.resnets.0
|
||||
up_blocks.2.resnets.0.norm1
|
||||
up_blocks.2.resnets.0.conv1
|
||||
up_blocks.2.resnets.0.time_emb_proj
|
||||
up_blocks.2.resnets.0.norm2
|
||||
up_blocks.2.resnets.0.dropout
|
||||
up_blocks.2.resnets.0.conv2
|
||||
up_blocks.2.resnets.0.conv_shortcut
|
||||
up_blocks.2.resnets.1
|
||||
up_blocks.2.resnets.1.norm1
|
||||
up_blocks.2.resnets.1.conv1
|
||||
up_blocks.2.resnets.1.time_emb_proj
|
||||
up_blocks.2.resnets.1.norm2
|
||||
up_blocks.2.resnets.1.dropout
|
||||
up_blocks.2.resnets.1.conv2
|
||||
up_blocks.2.resnets.1.conv_shortcut
|
||||
up_blocks.2.resnets.2
|
||||
up_blocks.2.resnets.2.norm1
|
||||
up_blocks.2.resnets.2.conv1
|
||||
up_blocks.2.resnets.2.time_emb_proj
|
||||
up_blocks.2.resnets.2.norm2
|
||||
up_blocks.2.resnets.2.dropout
|
||||
up_blocks.2.resnets.2.conv2
|
||||
up_blocks.2.resnets.2.conv_shortcut
|
||||
up_blocks.2.upsamplers
|
||||
up_blocks.2.upsamplers.0
|
||||
up_blocks.2.upsamplers.0.conv
|
||||
up_blocks.3
|
||||
up_blocks.3.attentions
|
||||
up_blocks.3.attentions.0
|
||||
up_blocks.3.attentions.0.norm
|
||||
up_blocks.3.attentions.0.proj_in
|
||||
up_blocks.3.attentions.0.transformer_blocks
|
||||
up_blocks.3.attentions.0.transformer_blocks.0
|
||||
up_blocks.3.attentions.0.transformer_blocks.0.norm1
|
||||
up_blocks.3.attentions.0.transformer_blocks.0.attn1
|
||||
up_blocks.3.attentions.0.transformer_blocks.0.attn1.to_q
|
||||
up_blocks.3.attentions.0.transformer_blocks.0.attn1.to_k
|
||||
up_blocks.3.attentions.0.transformer_blocks.0.attn1.to_v
|
||||
up_blocks.3.attentions.0.transformer_blocks.0.attn1.to_out
|
||||
up_blocks.3.attentions.0.transformer_blocks.0.attn1.to_out.0
|
||||
up_blocks.3.attentions.0.transformer_blocks.0.attn1.to_out.1
|
||||
up_blocks.3.attentions.0.transformer_blocks.0.norm2
|
||||
up_blocks.3.attentions.0.transformer_blocks.0.attn2
|
||||
up_blocks.3.attentions.0.transformer_blocks.0.attn2.to_q
|
||||
up_blocks.3.attentions.0.transformer_blocks.0.attn2.to_k
|
||||
up_blocks.3.attentions.0.transformer_blocks.0.attn2.to_v
|
||||
up_blocks.3.attentions.0.transformer_blocks.0.attn2.to_out
|
||||
up_blocks.3.attentions.0.transformer_blocks.0.attn2.to_out.0
|
||||
up_blocks.3.attentions.0.transformer_blocks.0.attn2.to_out.1
|
||||
up_blocks.3.attentions.0.transformer_blocks.0.norm3
|
||||
up_blocks.3.attentions.0.transformer_blocks.0.ff
|
||||
up_blocks.3.attentions.0.transformer_blocks.0.ff.net
|
||||
up_blocks.3.attentions.0.transformer_blocks.0.ff.net.0
|
||||
up_blocks.3.attentions.0.transformer_blocks.0.ff.net.0.proj
|
||||
up_blocks.3.attentions.0.transformer_blocks.0.ff.net.1
|
||||
up_blocks.3.attentions.0.transformer_blocks.0.ff.net.2
|
||||
up_blocks.3.attentions.0.proj_out
|
||||
up_blocks.3.attentions.1
|
||||
up_blocks.3.attentions.1.norm
|
||||
up_blocks.3.attentions.1.proj_in
|
||||
up_blocks.3.attentions.1.transformer_blocks
|
||||
up_blocks.3.attentions.1.transformer_blocks.0
|
||||
up_blocks.3.attentions.1.transformer_blocks.0.norm1
|
||||
up_blocks.3.attentions.1.transformer_blocks.0.attn1
|
||||
up_blocks.3.attentions.1.transformer_blocks.0.attn1.to_q
|
||||
up_blocks.3.attentions.1.transformer_blocks.0.attn1.to_k
|
||||
up_blocks.3.attentions.1.transformer_blocks.0.attn1.to_v
|
||||
up_blocks.3.attentions.1.transformer_blocks.0.attn1.to_out
|
||||
up_blocks.3.attentions.1.transformer_blocks.0.attn1.to_out.0
|
||||
up_blocks.3.attentions.1.transformer_blocks.0.attn1.to_out.1
|
||||
up_blocks.3.attentions.1.transformer_blocks.0.norm2
|
||||
up_blocks.3.attentions.1.transformer_blocks.0.attn2
|
||||
up_blocks.3.attentions.1.transformer_blocks.0.attn2.to_q
|
||||
up_blocks.3.attentions.1.transformer_blocks.0.attn2.to_k
|
||||
up_blocks.3.attentions.1.transformer_blocks.0.attn2.to_v
|
||||
up_blocks.3.attentions.1.transformer_blocks.0.attn2.to_out
|
||||
up_blocks.3.attentions.1.transformer_blocks.0.attn2.to_out.0
|
||||
up_blocks.3.attentions.1.transformer_blocks.0.attn2.to_out.1
|
||||
up_blocks.3.attentions.1.transformer_blocks.0.norm3
|
||||
up_blocks.3.attentions.1.transformer_blocks.0.ff
|
||||
up_blocks.3.attentions.1.transformer_blocks.0.ff.net
|
||||
up_blocks.3.attentions.1.transformer_blocks.0.ff.net.0
|
||||
up_blocks.3.attentions.1.transformer_blocks.0.ff.net.0.proj
|
||||
up_blocks.3.attentions.1.transformer_blocks.0.ff.net.1
|
||||
up_blocks.3.attentions.1.transformer_blocks.0.ff.net.2
|
||||
up_blocks.3.attentions.1.proj_out
|
||||
up_blocks.3.attentions.2
|
||||
up_blocks.3.attentions.2.norm
|
||||
up_blocks.3.attentions.2.proj_in
|
||||
up_blocks.3.attentions.2.transformer_blocks
|
||||
up_blocks.3.attentions.2.transformer_blocks.0
|
||||
up_blocks.3.attentions.2.transformer_blocks.0.norm1
|
||||
up_blocks.3.attentions.2.transformer_blocks.0.attn1
|
||||
up_blocks.3.attentions.2.transformer_blocks.0.attn1.to_q
|
||||
up_blocks.3.attentions.2.transformer_blocks.0.attn1.to_k
|
||||
up_blocks.3.attentions.2.transformer_blocks.0.attn1.to_v
|
||||
up_blocks.3.attentions.2.transformer_blocks.0.attn1.to_out
|
||||
up_blocks.3.attentions.2.transformer_blocks.0.attn1.to_out.0
|
||||
up_blocks.3.attentions.2.transformer_blocks.0.attn1.to_out.1
|
||||
up_blocks.3.attentions.2.transformer_blocks.0.norm2
|
||||
up_blocks.3.attentions.2.transformer_blocks.0.attn2
|
||||
up_blocks.3.attentions.2.transformer_blocks.0.attn2.to_q
|
||||
up_blocks.3.attentions.2.transformer_blocks.0.attn2.to_k
|
||||
up_blocks.3.attentions.2.transformer_blocks.0.attn2.to_v
|
||||
up_blocks.3.attentions.2.transformer_blocks.0.attn2.to_out
|
||||
up_blocks.3.attentions.2.transformer_blocks.0.attn2.to_out.0
|
||||
up_blocks.3.attentions.2.transformer_blocks.0.attn2.to_out.1
|
||||
up_blocks.3.attentions.2.transformer_blocks.0.norm3
|
||||
up_blocks.3.attentions.2.transformer_blocks.0.ff
|
||||
up_blocks.3.attentions.2.transformer_blocks.0.ff.net
|
||||
up_blocks.3.attentions.2.transformer_blocks.0.ff.net.0
|
||||
up_blocks.3.attentions.2.transformer_blocks.0.ff.net.0.proj
|
||||
up_blocks.3.attentions.2.transformer_blocks.0.ff.net.1
|
||||
up_blocks.3.attentions.2.transformer_blocks.0.ff.net.2
|
||||
up_blocks.3.attentions.2.proj_out
|
||||
up_blocks.3.resnets
|
||||
up_blocks.3.resnets.0
|
||||
up_blocks.3.resnets.0.norm1
|
||||
up_blocks.3.resnets.0.conv1
|
||||
up_blocks.3.resnets.0.time_emb_proj
|
||||
up_blocks.3.resnets.0.norm2
|
||||
up_blocks.3.resnets.0.dropout
|
||||
up_blocks.3.resnets.0.conv2
|
||||
up_blocks.3.resnets.0.conv_shortcut
|
||||
up_blocks.3.resnets.1
|
||||
up_blocks.3.resnets.1.norm1
|
||||
up_blocks.3.resnets.1.conv1
|
||||
up_blocks.3.resnets.1.time_emb_proj
|
||||
up_blocks.3.resnets.1.norm2
|
||||
up_blocks.3.resnets.1.dropout
|
||||
up_blocks.3.resnets.1.conv2
|
||||
up_blocks.3.resnets.1.conv_shortcut
|
||||
up_blocks.3.resnets.2
|
||||
up_blocks.3.resnets.2.norm1
|
||||
up_blocks.3.resnets.2.conv1
|
||||
up_blocks.3.resnets.2.time_emb_proj
|
||||
up_blocks.3.resnets.2.norm2
|
||||
up_blocks.3.resnets.2.dropout
|
||||
up_blocks.3.resnets.2.conv2
|
||||
up_blocks.3.resnets.2.conv_shortcut
|
||||
mid_block
|
||||
mid_block.attentions
|
||||
mid_block.attentions.0
|
||||
mid_block.attentions.0.norm
|
||||
mid_block.attentions.0.proj_in
|
||||
mid_block.attentions.0.transformer_blocks
|
||||
mid_block.attentions.0.transformer_blocks.0
|
||||
mid_block.attentions.0.transformer_blocks.0.norm1
|
||||
mid_block.attentions.0.transformer_blocks.0.attn1
|
||||
mid_block.attentions.0.transformer_blocks.0.attn1.to_q
|
||||
mid_block.attentions.0.transformer_blocks.0.attn1.to_k
|
||||
mid_block.attentions.0.transformer_blocks.0.attn1.to_v
|
||||
mid_block.attentions.0.transformer_blocks.0.attn1.to_out
|
||||
mid_block.attentions.0.transformer_blocks.0.attn1.to_out.0
|
||||
mid_block.attentions.0.transformer_blocks.0.attn1.to_out.1
|
||||
mid_block.attentions.0.transformer_blocks.0.norm2
|
||||
mid_block.attentions.0.transformer_blocks.0.attn2
|
||||
mid_block.attentions.0.transformer_blocks.0.attn2.to_q
|
||||
mid_block.attentions.0.transformer_blocks.0.attn2.to_k
|
||||
mid_block.attentions.0.transformer_blocks.0.attn2.to_v
|
||||
mid_block.attentions.0.transformer_blocks.0.attn2.to_out
|
||||
mid_block.attentions.0.transformer_blocks.0.attn2.to_out.0
|
||||
mid_block.attentions.0.transformer_blocks.0.attn2.to_out.1
|
||||
mid_block.attentions.0.transformer_blocks.0.norm3
|
||||
mid_block.attentions.0.transformer_blocks.0.ff
|
||||
mid_block.attentions.0.transformer_blocks.0.ff.net
|
||||
mid_block.attentions.0.transformer_blocks.0.ff.net.0
|
||||
mid_block.attentions.0.transformer_blocks.0.ff.net.0.proj
|
||||
mid_block.attentions.0.transformer_blocks.0.ff.net.1
|
||||
mid_block.attentions.0.transformer_blocks.0.ff.net.2
|
||||
mid_block.attentions.0.proj_out
|
||||
mid_block.resnets
|
||||
mid_block.resnets.0
|
||||
mid_block.resnets.0.norm1
|
||||
mid_block.resnets.0.conv1
|
||||
mid_block.resnets.0.time_emb_proj
|
||||
mid_block.resnets.0.norm2
|
||||
mid_block.resnets.0.dropout
|
||||
mid_block.resnets.0.conv2
|
||||
mid_block.resnets.1
|
||||
mid_block.resnets.1.norm1
|
||||
mid_block.resnets.1.conv1
|
||||
mid_block.resnets.1.time_emb_proj
|
||||
mid_block.resnets.1.norm2
|
||||
mid_block.resnets.1.dropout
|
||||
mid_block.resnets.1.conv2
|
||||
conv_norm_out
|
||||
conv_out
|
||||
File diff suppressed because it is too large
Load Diff
@@ -0,0 +1,32 @@
|
||||
import torch
|
||||
|
||||
|
||||
def isinstance_str(x: object, cls_name: str):
|
||||
"""
|
||||
Checks whether x has any class *named* cls_name in its ancestry.
|
||||
Doesn't require access to the class's implementation.
|
||||
|
||||
Useful for patching!
|
||||
"""
|
||||
|
||||
for _cls in x.__class__.__mro__:
|
||||
if _cls.__name__ == cls_name:
|
||||
return True
|
||||
|
||||
return False
|
||||
|
||||
|
||||
def init_generator(device: torch.device, fallback: torch.Generator=None):
|
||||
"""
|
||||
Forks the current default random generator given device.
|
||||
"""
|
||||
if device.type == "cpu":
|
||||
return torch.Generator(device="cpu").set_state(torch.get_rng_state())
|
||||
elif device.type == "cuda":
|
||||
return torch.Generator(device=device).set_state(torch.cuda.get_rng_state())
|
||||
else:
|
||||
if fallback is None:
|
||||
return init_generator(torch.device("cpu"))
|
||||
else:
|
||||
return fallback
|
||||
|
||||
@@ -0,0 +1,358 @@
|
||||
import os
|
||||
from contextlib import nullcontext
|
||||
import torch
|
||||
try:
|
||||
from diffusers import (
|
||||
DPMSolverMultistepScheduler,
|
||||
StableDiffusionPipeline,
|
||||
StableDiffusionImg2ImgPipeline,
|
||||
EulerDiscreteScheduler,
|
||||
EulerAncestralDiscreteScheduler,
|
||||
AutoencoderKL,
|
||||
UNet2DConditionModel,
|
||||
LCMScheduler,
|
||||
DDPMScheduler,
|
||||
DEISMultistepScheduler,
|
||||
PNDMScheduler,
|
||||
UniPCMultistepScheduler
|
||||
)
|
||||
from diffusers.loaders.single_file_utils import (
|
||||
convert_ldm_vae_checkpoint,
|
||||
convert_ldm_unet_checkpoint,
|
||||
create_vae_diffusers_config,
|
||||
create_unet_diffusers_config,
|
||||
create_text_encoder_from_ldm_clip_checkpoint
|
||||
)
|
||||
except:
|
||||
raise ImportError("Diffusers version too old. Please update to 0.26.0 minimum.")
|
||||
from .scheduling_tcd import TCDScheduler
|
||||
from contextlib import nullcontext
|
||||
from diffusers.utils import is_accelerate_available
|
||||
if is_accelerate_available():
|
||||
from accelerate import init_empty_weights
|
||||
from accelerate.utils import set_module_tensor_to_device
|
||||
|
||||
from .hidiffusion import apply_hidiffusion, remove_hidiffusion
|
||||
|
||||
from omegaconf import OmegaConf
|
||||
from transformers import CLIPTokenizer
|
||||
import comfy.model_management as mm
|
||||
import comfy.utils
|
||||
import folder_paths
|
||||
|
||||
script_directory = os.path.dirname(os.path.abspath(__file__))
|
||||
|
||||
class diffusers_model_loader:
|
||||
@classmethod
|
||||
def INPUT_TYPES(s):
|
||||
return {"required": {
|
||||
"model": ("MODEL",),
|
||||
"clip": ("CLIP",),
|
||||
"vae": ("VAE",),
|
||||
},
|
||||
}
|
||||
|
||||
RETURN_TYPES = ("DIFFUSERSMODEL",)
|
||||
RETURN_NAMES = ("diffusers_model",)
|
||||
FUNCTION = "loadmodel"
|
||||
CATEGORY = "IC-Light-Wrapper"
|
||||
|
||||
def loadmodel(self, model, clip, vae):
|
||||
mm.soft_empty_cache()
|
||||
dtype = mm.unet_dtype()
|
||||
vae_dtype = mm.vae_dtype()
|
||||
device = mm.get_torch_device()
|
||||
|
||||
custom_config = {
|
||||
'model': model,
|
||||
'vae': vae,
|
||||
}
|
||||
if not hasattr(self, 'model') or self.model == None or custom_config != self.current_config:
|
||||
pbar = comfy.utils.ProgressBar(5)
|
||||
self.current_config = custom_config
|
||||
# setup pretrained models
|
||||
original_config = OmegaConf.load(os.path.join(script_directory, f"configs/v1-inference.yaml"))
|
||||
|
||||
print("loading ELLA")
|
||||
checkpoint_path = os.path.join(folder_paths.models_dir,'ella')
|
||||
ella_path = os.path.join(checkpoint_path, 'ella-sd1.5-tsc-t5xl.safetensors')
|
||||
if not os.path.exists(ella_path):
|
||||
from huggingface_hub import snapshot_download
|
||||
snapshot_download(repo_id="QQGYLab/ELLA", local_dir=checkpoint_path, local_dir_use_symlinks=False)
|
||||
|
||||
with (init_empty_weights() if is_accelerate_available() else nullcontext()):
|
||||
converted_vae_config = create_vae_diffusers_config(original_config, image_size=512)
|
||||
new_vae = AutoencoderKL(**converted_vae_config)
|
||||
|
||||
converted_unet_config = create_unet_diffusers_config(original_config, image_size=512)
|
||||
unet = UNet2DConditionModel(**converted_unet_config)
|
||||
|
||||
clip_sd = None
|
||||
load_models = [model]
|
||||
load_models.append(clip.load_model())
|
||||
clip_sd = clip.get_sd()
|
||||
comfy.model_management.load_models_gpu(load_models)
|
||||
sd = model.model.state_dict_for_saving(clip_sd, vae.get_sd(), None)
|
||||
|
||||
converted_vae = convert_ldm_vae_checkpoint(sd, converted_vae_config)
|
||||
if is_accelerate_available():
|
||||
for key in converted_vae:
|
||||
set_module_tensor_to_device(new_vae, key, device=device, dtype=dtype, value=converted_vae[key])
|
||||
else:
|
||||
new_vae.load_state_dict(converted_vae, strict=False)
|
||||
del converted_vae
|
||||
pbar.update(1)
|
||||
|
||||
converted_unet = convert_ldm_unet_checkpoint(sd, converted_unet_config)
|
||||
if is_accelerate_available():
|
||||
for key in converted_unet:
|
||||
set_module_tensor_to_device(unet, key, device=device, dtype=dtype, value=converted_unet[key])
|
||||
else:
|
||||
unet.load_state_dict(converted_unet, strict=False)
|
||||
del converted_unet
|
||||
|
||||
pbar.update(1)
|
||||
# 3. text_model
|
||||
print("loading text model")
|
||||
text_encoder = create_text_encoder_from_ldm_clip_checkpoint("openai/clip-vit-large-patch14",sd)
|
||||
scheduler_config = {
|
||||
'num_train_timesteps': 1000,
|
||||
'beta_start': 0.00085,
|
||||
'beta_end': 0.012,
|
||||
'beta_schedule': "scaled_linear",
|
||||
'steps_offset': 1
|
||||
}
|
||||
# 4. tokenizer
|
||||
tokenizer_path = os.path.join(script_directory, "configs/tokenizer")
|
||||
tokenizer = CLIPTokenizer.from_pretrained(tokenizer_path)
|
||||
|
||||
scheduler=DPMSolverMultistepScheduler(**scheduler_config)
|
||||
pbar.update(1)
|
||||
del sd
|
||||
|
||||
pbar.update(1)
|
||||
|
||||
print("creating pipeline")
|
||||
self.pipe = StableDiffusionImg2ImgPipeline(
|
||||
unet=unet,
|
||||
vae=new_vae,
|
||||
text_encoder=text_encoder,
|
||||
tokenizer=tokenizer,
|
||||
scheduler=scheduler,
|
||||
safety_checker=None,
|
||||
feature_extractor=None,
|
||||
requires_safety_checker=False,
|
||||
image_encoder=None
|
||||
)
|
||||
print("pipeline created")
|
||||
pbar.update(1)
|
||||
#self.pipe.enable_model_cpu_offload()
|
||||
diffusers_model = {
|
||||
'pipe': self.pipe,
|
||||
}
|
||||
|
||||
return (diffusers_model,)
|
||||
|
||||
class LoadICLightUnetDiffusers:
|
||||
@classmethod
|
||||
def INPUT_TYPES(s):
|
||||
return {
|
||||
"required": {
|
||||
"diffusersmodel": ("DIFFUSERSMODEL",),
|
||||
"model_path": (folder_paths.get_filename_list("unet"), )
|
||||
}
|
||||
}
|
||||
|
||||
RETURN_TYPES = ("DIFFUSERSMODEL",)
|
||||
FUNCTION = "load"
|
||||
CATEGORY = "IC-Light-Wrapper"
|
||||
|
||||
def load(self, diffusersmodel, model_path):
|
||||
unet = diffusersmodel["pipe"].unet
|
||||
device = mm.get_torch_device()
|
||||
|
||||
unet_original_forward = unet.forward
|
||||
|
||||
new_conv_in = torch.nn.Conv2d(8, unet.conv_in.out_channels, unet.conv_in.kernel_size, unet.conv_in.stride, unet.conv_in.padding)
|
||||
new_conv_in.weight.zero_()
|
||||
new_conv_in.weight[:, :4, :, :].copy_(unet.conv_in.weight)
|
||||
new_conv_in.bias = unet.conv_in.bias
|
||||
unet.conv_in = new_conv_in
|
||||
|
||||
def hooked_unet_forward(sample, timestep, encoder_hidden_states, **kwargs):
|
||||
c_concat = kwargs['cross_attention_kwargs']['concat_conds'].to(sample)
|
||||
c_concat = torch.cat([c_concat] * (sample.shape[0] // c_concat.shape[0]), dim=0)
|
||||
new_sample = torch.cat([sample, c_concat], dim=1)
|
||||
kwargs['cross_attention_kwargs'] = {}
|
||||
return unet_original_forward(new_sample, timestep, encoder_hidden_states, **kwargs)
|
||||
|
||||
unet.forward = hooked_unet_forward
|
||||
|
||||
model_full_path = folder_paths.get_full_path("unet", model_path)
|
||||
if not os.path.exists(model_full_path):
|
||||
raise Exception("Invalid model path")
|
||||
else:
|
||||
print("LoadICLightUnet: Loading LoadICLightUnet weights")
|
||||
from comfy.utils import load_torch_file
|
||||
sd_offset = load_torch_file(model_full_path, device=mm.get_torch_device())
|
||||
sd_origin = unet.state_dict()
|
||||
keys = sd_origin.keys()
|
||||
sd_merged = {k: sd_origin[k].to(device) + sd_offset[k].to(device) for k in sd_origin.keys()}
|
||||
unet.load_state_dict(sd_merged, strict=True)
|
||||
del sd_offset, sd_origin, sd_merged, keys
|
||||
|
||||
return diffusersmodel,
|
||||
|
||||
class iclight_diffusers_sampler:
|
||||
@classmethod
|
||||
def INPUT_TYPES(s):
|
||||
return {"required": {
|
||||
"diffusers_model": ("DIFFUSERSMODEL",),
|
||||
"latent": ("LATENT",),
|
||||
"width": ("INT", {"default": 512, "min": 64, "max": 2048, "step": 64}),
|
||||
"height": ("INT", {"default": 512, "min": 64, "max": 2048, "step": 64}),
|
||||
"steps": ("INT", {"default": 25, "min": 1, "max": 200, "step": 1}),
|
||||
"guidance_scale": ("FLOAT", {"default": 2.0, "min": 1.01, "max": 20.0, "step": 0.01}),
|
||||
"denoise_strength": ("FLOAT", {"default": 0.9, "min": 0.01, "max": 1.0, "step": 0.01}),
|
||||
"seed": ("INT", {"default": 0, "min": 0, "max": 0xffffffffffffffff}),
|
||||
"scheduler": (
|
||||
[
|
||||
'DPMSolverMultistepScheduler',
|
||||
'DPMSolverMultistepScheduler_SDE_karras',
|
||||
'DDPMScheduler',
|
||||
'LCMScheduler',
|
||||
'PNDMScheduler',
|
||||
'DEISMultistepScheduler',
|
||||
'EulerDiscreteScheduler',
|
||||
'EulerAncestralDiscreteScheduler',
|
||||
'UniPCMultistepScheduler',
|
||||
'TCDScheduler'
|
||||
], {
|
||||
"default": 'DPMSolverMultistepScheduler'
|
||||
}),
|
||||
"prompt": ("STRING", {"default": "positive", "multiline": True}),
|
||||
"n_prompt": ("STRING", {"default": "negative", "multiline": True}),
|
||||
"hidiffusion": ("BOOLEAN", {"default": False}),
|
||||
},
|
||||
"optional" : {
|
||||
"bg_latent": ("LATENT",),
|
||||
"fixed_seed": ("BOOLEAN", {"default": True}),
|
||||
}
|
||||
}
|
||||
|
||||
RETURN_TYPES = ("LATENT",)
|
||||
RETURN_NAMES = ("samples",)
|
||||
FUNCTION = "process"
|
||||
CATEGORY = "IC-Light-Wrapper"
|
||||
|
||||
def process(self, latent, diffusers_model, width, height, steps, guidance_scale, denoise_strength, seed, scheduler, prompt, n_prompt, hidiffusion, bg_latent=None, fixed_seed=True):
|
||||
device = mm.get_torch_device()
|
||||
mm.unload_all_models()
|
||||
mm.soft_empty_cache()
|
||||
dtype = mm.unet_dtype()
|
||||
pipe=diffusers_model['pipe']
|
||||
pipe.to(device, dtype=dtype)
|
||||
scale_factor = pipe.vae.config.scaling_factor
|
||||
|
||||
scheduler_config = {
|
||||
'num_train_timesteps': 1000,
|
||||
'beta_start': 0.00085,
|
||||
'beta_end': 0.012,
|
||||
'beta_schedule': "scaled_linear",
|
||||
'steps_offset': 1,
|
||||
}
|
||||
if scheduler == 'DPMSolverMultistepScheduler':
|
||||
noise_scheduler = DPMSolverMultistepScheduler(**scheduler_config)
|
||||
elif scheduler == 'DPMSolverMultistepScheduler_SDE_karras':
|
||||
scheduler_config.update({"algorithm_type": "sde-dpmsolver++"})
|
||||
scheduler_config.update({"use_karras_sigmas": True})
|
||||
noise_scheduler = DPMSolverMultistepScheduler(**scheduler_config)
|
||||
elif scheduler == 'DDPMScheduler':
|
||||
noise_scheduler = DDPMScheduler(**scheduler_config)
|
||||
elif scheduler == 'LCMScheduler':
|
||||
noise_scheduler = LCMScheduler(**scheduler_config)
|
||||
elif scheduler == 'PNDMScheduler':
|
||||
scheduler_config.update({"set_alpha_to_one": False})
|
||||
scheduler_config.update({"trained_betas": None})
|
||||
noise_scheduler = PNDMScheduler(**scheduler_config)
|
||||
elif scheduler == 'DEISMultistepScheduler':
|
||||
noise_scheduler = DEISMultistepScheduler(**scheduler_config)
|
||||
elif scheduler == 'EulerDiscreteScheduler':
|
||||
noise_scheduler = EulerDiscreteScheduler(**scheduler_config)
|
||||
elif scheduler == 'EulerAncestralDiscreteScheduler':
|
||||
noise_scheduler = EulerAncestralDiscreteScheduler(**scheduler_config)
|
||||
elif scheduler == 'UniPCMultistepScheduler':
|
||||
noise_scheduler = UniPCMultistepScheduler(**scheduler_config)
|
||||
elif scheduler == 'TCDScheduler':
|
||||
noise_scheduler = TCDScheduler(**scheduler_config)
|
||||
|
||||
pipe.scheduler = noise_scheduler
|
||||
if hidiffusion:
|
||||
apply_hidiffusion(pipe)
|
||||
else:
|
||||
remove_hidiffusion(pipe)
|
||||
|
||||
if bg_latent is not None:
|
||||
bg_latent = bg_latent["samples"]
|
||||
bg_latent = bg_latent * pipe.vae.config.scaling_factor
|
||||
else:
|
||||
bg_latent = None
|
||||
|
||||
concat_conds = latent["samples"]
|
||||
concat_conds = concat_conds * pipe.vae.config.scaling_factor
|
||||
B, H, W, C = latent["samples"].shape
|
||||
prompt_list = []
|
||||
prompt_list.append(prompt)
|
||||
if len(prompt_list) < B:
|
||||
prompt_list += [prompt_list[-1]] * (B - len(prompt_list))
|
||||
|
||||
n_prompt_list = []
|
||||
n_prompt_list.append(n_prompt)
|
||||
if len(n_prompt_list) < B:
|
||||
|
||||
n_prompt_list += [n_prompt_list[-1]] * (B - len(n_prompt_list))
|
||||
|
||||
if fixed_seed:
|
||||
generator = [torch.Generator(device=device).manual_seed(seed) for _ in range(B)]
|
||||
else:
|
||||
generator= [torch.Generator(device="cuda").manual_seed(i) for i in range(B)]
|
||||
|
||||
pbar = comfy.utils.ProgressBar(steps)
|
||||
def progress_counter_callback(pipeline, step, timestep, callback_kwargs):
|
||||
pbar.update(1)
|
||||
return callback_kwargs or {}
|
||||
|
||||
autocast_condition = (dtype != torch.float32) and not mm.is_device_mps(device)
|
||||
with torch.autocast(mm.get_autocast_device(device), dtype=dtype) if autocast_condition else nullcontext():
|
||||
|
||||
images = pipe(
|
||||
image=bg_latent,
|
||||
prompt = prompt_list,
|
||||
strength = denoise_strength,
|
||||
negative_prompt = n_prompt_list,
|
||||
prompt_embeds=None,
|
||||
negative_prompt_embeds=None,
|
||||
guidance_scale=guidance_scale,
|
||||
num_inference_steps=int(round(steps / denoise_strength)),
|
||||
height=height,
|
||||
width=width,
|
||||
cross_attention_kwargs={'concat_conds': concat_conds},
|
||||
generator=generator,
|
||||
output_type="latent",
|
||||
callback_on_step_end=progress_counter_callback,
|
||||
#callback_on_step_end_tensor_inputs=["latents", "prompt_embeds", "negative_prompt_embeds"],
|
||||
).images
|
||||
images = images / scale_factor
|
||||
#image_out = images.permute(0, 2, 3, 1).cpu().float()
|
||||
return ({"samples": images},)
|
||||
|
||||
NODE_CLASS_MAPPINGS = {
|
||||
"diffusers_model_loader": diffusers_model_loader,
|
||||
"LoadICLightUnetDiffusers": LoadICLightUnetDiffusers,
|
||||
"iclight_diffusers_sampler": iclight_diffusers_sampler
|
||||
}
|
||||
NODE_DISPLAY_NAME_MAPPINGS = {
|
||||
"diffusers_model_loader": "Diffusers Model Loader",
|
||||
"LoadICLightUnetDiffusers": "LoadICLightUnetDiffusers",
|
||||
"iclight_diffusers_sampler": "IC-Light Diffusers Sampler"
|
||||
}
|
||||
@@ -0,0 +1,3 @@
|
||||
diffusers>=0.27.2
|
||||
accelerate>=0.29.2
|
||||
omegaconf
|
||||
@@ -0,0 +1,697 @@
|
||||
# Copyright 2024 Stanford University Team and The HuggingFace Team. All rights reserved.
|
||||
#
|
||||
# Licensed under the Apache License, Version 2.0 (the "License");
|
||||
# you may not use this file except in compliance with the License.
|
||||
# You may obtain a copy of the License at
|
||||
#
|
||||
# http://www.apache.org/licenses/LICENSE-2.0
|
||||
#
|
||||
# Unless required by applicable law or agreed to in writing, software
|
||||
# distributed under the License is distributed on an "AS IS" BASIS,
|
||||
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
# See the License for the specific language governing permissions and
|
||||
# limitations under the License.
|
||||
|
||||
# DISCLAIMER: This code is strongly influenced by https://github.com/pesser/pytorch_diffusion
|
||||
# and https://github.com/hojonathanho/diffusion
|
||||
|
||||
import math
|
||||
from dataclasses import dataclass
|
||||
from typing import List, Optional, Tuple, Union
|
||||
|
||||
import numpy as np
|
||||
import torch
|
||||
|
||||
from diffusers.configuration_utils import ConfigMixin, register_to_config
|
||||
from diffusers.schedulers.scheduling_utils import SchedulerMixin
|
||||
from diffusers.utils import BaseOutput, logging
|
||||
from diffusers.utils.torch_utils import randn_tensor
|
||||
|
||||
|
||||
logger = logging.get_logger(__name__) # pylint: disable=invalid-name
|
||||
|
||||
|
||||
@dataclass
|
||||
class TCDSchedulerOutput(BaseOutput):
|
||||
"""
|
||||
Output class for the scheduler's `step` function output.
|
||||
|
||||
Args:
|
||||
prev_sample (`torch.FloatTensor` of shape `(batch_size, num_channels, height, width)` for images):
|
||||
Computed sample `(x_{t-1})` of previous timestep. `prev_sample` should be used as next model input in the
|
||||
denoising loop.
|
||||
pred_noised_sample (`torch.FloatTensor` of shape `(batch_size, num_channels, height, width)` for images):
|
||||
The predicted noised sample `(x_{s})` based on the model output from the current timestep.
|
||||
"""
|
||||
|
||||
prev_sample: torch.FloatTensor
|
||||
pred_noised_sample: Optional[torch.FloatTensor] = None
|
||||
|
||||
|
||||
# Copied from diffusers.schedulers.scheduling_ddpm.betas_for_alpha_bar
|
||||
def betas_for_alpha_bar(
|
||||
num_diffusion_timesteps,
|
||||
max_beta=0.999,
|
||||
alpha_transform_type="cosine",
|
||||
):
|
||||
"""
|
||||
Create a beta schedule that discretizes the given alpha_t_bar function, which defines the cumulative product of
|
||||
(1-beta) over time from t = [0,1].
|
||||
|
||||
Contains a function alpha_bar that takes an argument t and transforms it to the cumulative product of (1-beta) up
|
||||
to that part of the diffusion process.
|
||||
|
||||
|
||||
Args:
|
||||
num_diffusion_timesteps (`int`): the number of betas to produce.
|
||||
max_beta (`float`): the maximum beta to use; use values lower than 1 to
|
||||
prevent singularities.
|
||||
alpha_transform_type (`str`, *optional*, default to `cosine`): the type of noise schedule for alpha_bar.
|
||||
Choose from `cosine` or `exp`
|
||||
|
||||
Returns:
|
||||
betas (`np.ndarray`): the betas used by the scheduler to step the model outputs
|
||||
"""
|
||||
if alpha_transform_type == "cosine":
|
||||
|
||||
def alpha_bar_fn(t):
|
||||
return math.cos((t + 0.008) / 1.008 * math.pi / 2) ** 2
|
||||
|
||||
elif alpha_transform_type == "exp":
|
||||
|
||||
def alpha_bar_fn(t):
|
||||
return math.exp(t * -12.0)
|
||||
|
||||
else:
|
||||
raise ValueError(f"Unsupported alpha_transform_type: {alpha_transform_type}")
|
||||
|
||||
betas = []
|
||||
for i in range(num_diffusion_timesteps):
|
||||
t1 = i / num_diffusion_timesteps
|
||||
t2 = (i + 1) / num_diffusion_timesteps
|
||||
betas.append(min(1 - alpha_bar_fn(t2) / alpha_bar_fn(t1), max_beta))
|
||||
return torch.tensor(betas, dtype=torch.float32)
|
||||
|
||||
|
||||
# Copied from diffusers.schedulers.scheduling_ddim.rescale_zero_terminal_snr
|
||||
def rescale_zero_terminal_snr(betas: torch.FloatTensor) -> torch.FloatTensor:
|
||||
"""
|
||||
Rescales betas to have zero terminal SNR Based on https://arxiv.org/pdf/2305.08891.pdf (Algorithm 1)
|
||||
|
||||
|
||||
Args:
|
||||
betas (`torch.FloatTensor`):
|
||||
the betas that the scheduler is being initialized with.
|
||||
|
||||
Returns:
|
||||
`torch.FloatTensor`: rescaled betas with zero terminal SNR
|
||||
"""
|
||||
# Convert betas to alphas_bar_sqrt
|
||||
alphas = 1.0 - betas
|
||||
alphas_cumprod = torch.cumprod(alphas, dim=0)
|
||||
alphas_bar_sqrt = alphas_cumprod.sqrt()
|
||||
|
||||
# Store old values.
|
||||
alphas_bar_sqrt_0 = alphas_bar_sqrt[0].clone()
|
||||
alphas_bar_sqrt_T = alphas_bar_sqrt[-1].clone()
|
||||
|
||||
# Shift so the last timestep is zero.
|
||||
alphas_bar_sqrt -= alphas_bar_sqrt_T
|
||||
|
||||
# Scale so the first timestep is back to the old value.
|
||||
alphas_bar_sqrt *= alphas_bar_sqrt_0 / (alphas_bar_sqrt_0 - alphas_bar_sqrt_T)
|
||||
|
||||
# Convert alphas_bar_sqrt to betas
|
||||
alphas_bar = alphas_bar_sqrt**2 # Revert sqrt
|
||||
alphas = alphas_bar[1:] / alphas_bar[:-1] # Revert cumprod
|
||||
alphas = torch.cat([alphas_bar[0:1], alphas])
|
||||
betas = 1 - alphas
|
||||
|
||||
return betas
|
||||
|
||||
|
||||
class TCDScheduler(SchedulerMixin, ConfigMixin):
|
||||
"""
|
||||
`TCDScheduler` incorporates the `Strategic Stochastic Sampling` introduced by the paper `Trajectory Consistency
|
||||
Distillation`, extending the original Multistep Consistency Sampling to enable unrestricted trajectory traversal.
|
||||
|
||||
This code is based on the official repo of TCD(https://github.com/jabir-zheng/TCD).
|
||||
|
||||
This model inherits from [`SchedulerMixin`] and [`ConfigMixin`]. [`~ConfigMixin`] takes care of storing all config
|
||||
attributes that are passed in the scheduler's `__init__` function, such as `num_train_timesteps`. They can be
|
||||
accessed via `scheduler.config.num_train_timesteps`. [`SchedulerMixin`] provides general loading and saving
|
||||
functionality via the [`SchedulerMixin.save_pretrained`] and [`~SchedulerMixin.from_pretrained`] functions.
|
||||
|
||||
Args:
|
||||
num_train_timesteps (`int`, defaults to 1000):
|
||||
The number of diffusion steps to train the model.
|
||||
beta_start (`float`, defaults to 0.0001):
|
||||
The starting `beta` value of inference.
|
||||
beta_end (`float`, defaults to 0.02):
|
||||
The final `beta` value.
|
||||
beta_schedule (`str`, defaults to `"linear"`):
|
||||
The beta schedule, a mapping from a beta range to a sequence of betas for stepping the model. Choose from
|
||||
`linear`, `scaled_linear`, or `squaredcos_cap_v2`.
|
||||
trained_betas (`np.ndarray`, *optional*):
|
||||
Pass an array of betas directly to the constructor to bypass `beta_start` and `beta_end`.
|
||||
original_inference_steps (`int`, *optional*, defaults to 50):
|
||||
The default number of inference steps used to generate a linearly-spaced timestep schedule, from which we
|
||||
will ultimately take `num_inference_steps` evenly spaced timesteps to form the final timestep schedule.
|
||||
clip_sample (`bool`, defaults to `True`):
|
||||
Clip the predicted sample for numerical stability.
|
||||
clip_sample_range (`float`, defaults to 1.0):
|
||||
The maximum magnitude for sample clipping. Valid only when `clip_sample=True`.
|
||||
set_alpha_to_one (`bool`, defaults to `True`):
|
||||
Each diffusion step uses the alphas product value at that step and at the previous one. For the final step
|
||||
there is no previous alpha. When this option is `True` the previous alpha product is fixed to `1`,
|
||||
otherwise it uses the alpha value at step 0.
|
||||
steps_offset (`int`, defaults to 0):
|
||||
An offset added to the inference steps, as required by some model families.
|
||||
prediction_type (`str`, defaults to `epsilon`, *optional*):
|
||||
Prediction type of the scheduler function; can be `epsilon` (predicts the noise of the diffusion process),
|
||||
`sample` (directly predicts the noisy sample`) or `v_prediction` (see section 2.4 of [Imagen
|
||||
Video](https://imagen.research.google/video/paper.pdf) paper).
|
||||
thresholding (`bool`, defaults to `False`):
|
||||
Whether to use the "dynamic thresholding" method. This is unsuitable for latent-space diffusion models such
|
||||
as Stable Diffusion.
|
||||
dynamic_thresholding_ratio (`float`, defaults to 0.995):
|
||||
The ratio for the dynamic thresholding method. Valid only when `thresholding=True`.
|
||||
sample_max_value (`float`, defaults to 1.0):
|
||||
The threshold value for dynamic thresholding. Valid only when `thresholding=True`.
|
||||
timestep_spacing (`str`, defaults to `"leading"`):
|
||||
The way the timesteps should be scaled. Refer to Table 2 of the [Common Diffusion Noise Schedules and
|
||||
Sample Steps are Flawed](https://huggingface.co/papers/2305.08891) for more information.
|
||||
timestep_scaling (`float`, defaults to 10.0):
|
||||
The factor the timesteps will be multiplied by when calculating the consistency model boundary conditions
|
||||
`c_skip` and `c_out`. Increasing this will decrease the approximation error (although the approximation
|
||||
error at the default of `10.0` is already pretty small).
|
||||
rescale_betas_zero_snr (`bool`, defaults to `False`):
|
||||
Whether to rescale the betas to have zero terminal SNR. This enables the model to generate very bright and
|
||||
dark samples instead of limiting it to samples with medium brightness. Loosely related to
|
||||
[`--offset_noise`](https://github.com/huggingface/diffusers/blob/74fd735eb073eb1d774b1ab4154a0876eb82f055/examples/dreambooth/train_dreambooth.py#L506).
|
||||
"""
|
||||
|
||||
order = 1
|
||||
|
||||
@register_to_config
|
||||
def __init__(
|
||||
self,
|
||||
num_train_timesteps: int = 1000,
|
||||
beta_start: float = 0.00085,
|
||||
beta_end: float = 0.012,
|
||||
beta_schedule: str = "scaled_linear",
|
||||
trained_betas: Optional[Union[np.ndarray, List[float]]] = None,
|
||||
original_inference_steps: int = 50,
|
||||
clip_sample: bool = False,
|
||||
clip_sample_range: float = 1.0,
|
||||
set_alpha_to_one: bool = True,
|
||||
steps_offset: int = 0,
|
||||
prediction_type: str = "epsilon",
|
||||
thresholding: bool = False,
|
||||
dynamic_thresholding_ratio: float = 0.995,
|
||||
sample_max_value: float = 1.0,
|
||||
timestep_spacing: str = "leading",
|
||||
timestep_scaling: float = 10.0,
|
||||
rescale_betas_zero_snr: bool = False,
|
||||
):
|
||||
if trained_betas is not None:
|
||||
self.betas = torch.tensor(trained_betas, dtype=torch.float32)
|
||||
elif beta_schedule == "linear":
|
||||
self.betas = torch.linspace(beta_start, beta_end, num_train_timesteps, dtype=torch.float32)
|
||||
elif beta_schedule == "scaled_linear":
|
||||
# this schedule is very specific to the latent diffusion model.
|
||||
self.betas = torch.linspace(beta_start**0.5, beta_end**0.5, num_train_timesteps, dtype=torch.float32) ** 2
|
||||
elif beta_schedule == "squaredcos_cap_v2":
|
||||
# Glide cosine schedule
|
||||
self.betas = betas_for_alpha_bar(num_train_timesteps)
|
||||
else:
|
||||
raise NotImplementedError(f"{beta_schedule} does is not implemented for {self.__class__}")
|
||||
|
||||
# Rescale for zero SNR
|
||||
if rescale_betas_zero_snr:
|
||||
self.betas = rescale_zero_terminal_snr(self.betas)
|
||||
|
||||
self.alphas = 1.0 - self.betas
|
||||
self.alphas_cumprod = torch.cumprod(self.alphas, dim=0)
|
||||
|
||||
# At every step in ddim, we are looking into the previous alphas_cumprod
|
||||
# For the final step, there is no previous alphas_cumprod because we are already at 0
|
||||
# `set_alpha_to_one` decides whether we set this parameter simply to one or
|
||||
# whether we use the final alpha of the "non-previous" one.
|
||||
self.final_alpha_cumprod = torch.tensor(1.0) if set_alpha_to_one else self.alphas_cumprod[0]
|
||||
|
||||
# standard deviation of the initial noise distribution
|
||||
self.init_noise_sigma = 1.0
|
||||
|
||||
# setable values
|
||||
self.num_inference_steps = None
|
||||
self.timesteps = torch.from_numpy(np.arange(0, num_train_timesteps)[::-1].copy().astype(np.int64))
|
||||
self.custom_timesteps = False
|
||||
|
||||
self._step_index = None
|
||||
self._begin_index = None
|
||||
|
||||
# Copied from diffusers.schedulers.scheduling_euler_discrete.EulerDiscreteScheduler.index_for_timestep
|
||||
def index_for_timestep(self, timestep, schedule_timesteps=None):
|
||||
if schedule_timesteps is None:
|
||||
schedule_timesteps = self.timesteps
|
||||
|
||||
indices = (schedule_timesteps == timestep).nonzero()
|
||||
|
||||
# The sigma index that is taken for the **very** first `step`
|
||||
# is always the second index (or the last index if there is only 1)
|
||||
# This way we can ensure we don't accidentally skip a sigma in
|
||||
# case we start in the middle of the denoising schedule (e.g. for image-to-image)
|
||||
pos = 1 if len(indices) > 1 else 0
|
||||
|
||||
return indices[pos].item()
|
||||
|
||||
# Copied from diffusers.schedulers.scheduling_euler_discrete.EulerDiscreteScheduler._init_step_index
|
||||
def _init_step_index(self, timestep):
|
||||
if self.begin_index is None:
|
||||
if isinstance(timestep, torch.Tensor):
|
||||
timestep = timestep.to(self.timesteps.device)
|
||||
self._step_index = self.index_for_timestep(timestep)
|
||||
else:
|
||||
self._step_index = self._begin_index
|
||||
|
||||
@property
|
||||
def step_index(self):
|
||||
return self._step_index
|
||||
|
||||
@property
|
||||
def begin_index(self):
|
||||
"""
|
||||
The index for the first timestep. It should be set from pipeline with `set_begin_index` method.
|
||||
"""
|
||||
return self._begin_index
|
||||
|
||||
# Copied from diffusers.schedulers.scheduling_dpmsolver_multistep.DPMSolverMultistepScheduler.set_begin_index
|
||||
def set_begin_index(self, begin_index: int = 0):
|
||||
"""
|
||||
Sets the begin index for the scheduler. This function should be run from pipeline before the inference.
|
||||
|
||||
Args:
|
||||
begin_index (`int`):
|
||||
The begin index for the scheduler.
|
||||
"""
|
||||
self._begin_index = begin_index
|
||||
|
||||
def scale_model_input(self, sample: torch.FloatTensor, timestep: Optional[int] = None) -> torch.FloatTensor:
|
||||
"""
|
||||
Ensures interchangeability with schedulers that need to scale the denoising model input depending on the
|
||||
current timestep.
|
||||
|
||||
Args:
|
||||
sample (`torch.FloatTensor`):
|
||||
The input sample.
|
||||
timestep (`int`, *optional*):
|
||||
The current timestep in the diffusion chain.
|
||||
|
||||
Returns:
|
||||
`torch.FloatTensor`:
|
||||
A scaled input sample.
|
||||
"""
|
||||
return sample
|
||||
|
||||
# Copied from diffusers.schedulers.scheduling_ddim.DDIMScheduler._get_variance
|
||||
def _get_variance(self, timestep, prev_timestep):
|
||||
alpha_prod_t = self.alphas_cumprod[timestep]
|
||||
alpha_prod_t_prev = self.alphas_cumprod[prev_timestep] if prev_timestep >= 0 else self.final_alpha_cumprod
|
||||
beta_prod_t = 1 - alpha_prod_t
|
||||
beta_prod_t_prev = 1 - alpha_prod_t_prev
|
||||
|
||||
variance = (beta_prod_t_prev / beta_prod_t) * (1 - alpha_prod_t / alpha_prod_t_prev)
|
||||
|
||||
return variance
|
||||
|
||||
# Copied from diffusers.schedulers.scheduling_ddpm.DDPMScheduler._threshold_sample
|
||||
def _threshold_sample(self, sample: torch.FloatTensor) -> torch.FloatTensor:
|
||||
"""
|
||||
"Dynamic thresholding: At each sampling step we set s to a certain percentile absolute pixel value in xt0 (the
|
||||
prediction of x_0 at timestep t), and if s > 1, then we threshold xt0 to the range [-s, s] and then divide by
|
||||
s. Dynamic thresholding pushes saturated pixels (those near -1 and 1) inwards, thereby actively preventing
|
||||
pixels from saturation at each step. We find that dynamic thresholding results in significantly better
|
||||
photorealism as well as better image-text alignment, especially when using very large guidance weights."
|
||||
|
||||
https://arxiv.org/abs/2205.11487
|
||||
"""
|
||||
dtype = sample.dtype
|
||||
batch_size, channels, *remaining_dims = sample.shape
|
||||
|
||||
if dtype not in (torch.float32, torch.float64):
|
||||
sample = sample.float() # upcast for quantile calculation, and clamp not implemented for cpu half
|
||||
|
||||
# Flatten sample for doing quantile calculation along each image
|
||||
sample = sample.reshape(batch_size, channels * np.prod(remaining_dims))
|
||||
|
||||
abs_sample = sample.abs() # "a certain percentile absolute pixel value"
|
||||
|
||||
s = torch.quantile(abs_sample, self.config.dynamic_thresholding_ratio, dim=1)
|
||||
s = torch.clamp(
|
||||
s, min=1, max=self.config.sample_max_value
|
||||
) # When clamped to min=1, equivalent to standard clipping to [-1, 1]
|
||||
s = s.unsqueeze(1) # (batch_size, 1) because clamp will broadcast along dim=0
|
||||
sample = torch.clamp(sample, -s, s) / s # "we threshold xt0 to the range [-s, s] and then divide by s"
|
||||
|
||||
sample = sample.reshape(batch_size, channels, *remaining_dims)
|
||||
sample = sample.to(dtype)
|
||||
|
||||
return sample
|
||||
|
||||
def set_timesteps(
|
||||
self,
|
||||
num_inference_steps: Optional[int] = None,
|
||||
device: Union[str, torch.device] = None,
|
||||
original_inference_steps: Optional[int] = None,
|
||||
timesteps: Optional[List[int]] = None,
|
||||
strength: float = 1.0,
|
||||
):
|
||||
"""
|
||||
Sets the discrete timesteps used for the diffusion chain (to be run before inference).
|
||||
|
||||
Args:
|
||||
num_inference_steps (`int`, *optional*):
|
||||
The number of diffusion steps used when generating samples with a pre-trained model. If used,
|
||||
`timesteps` must be `None`.
|
||||
device (`str` or `torch.device`, *optional*):
|
||||
The device to which the timesteps should be moved to. If `None`, the timesteps are not moved.
|
||||
original_inference_steps (`int`, *optional*):
|
||||
The original number of inference steps, which will be used to generate a linearly-spaced timestep
|
||||
schedule (which is different from the standard `diffusers` implementation). We will then take
|
||||
`num_inference_steps` timesteps from this schedule, evenly spaced in terms of indices, and use that as
|
||||
our final timestep schedule. If not set, this will default to the `original_inference_steps` attribute.
|
||||
timesteps (`List[int]`, *optional*):
|
||||
Custom timesteps used to support arbitrary spacing between timesteps. If `None`, then the default
|
||||
timestep spacing strategy of equal spacing between timesteps on the training/distillation timestep
|
||||
schedule is used. If `timesteps` is passed, `num_inference_steps` must be `None`.
|
||||
strength (`float`, *optional*, defaults to 1.0):
|
||||
Used to determine the number of timesteps used for inference when using img2img, inpaint, etc.
|
||||
"""
|
||||
# 0. Check inputs
|
||||
if num_inference_steps is None and timesteps is None:
|
||||
raise ValueError("Must pass exactly one of `num_inference_steps` or `custom_timesteps`.")
|
||||
|
||||
if num_inference_steps is not None and timesteps is not None:
|
||||
raise ValueError("Can only pass one of `num_inference_steps` or `custom_timesteps`.")
|
||||
|
||||
# 1. Calculate the TCD original training/distillation timestep schedule.
|
||||
original_steps = (
|
||||
original_inference_steps if original_inference_steps is not None else self.config.original_inference_steps
|
||||
)
|
||||
|
||||
if original_inference_steps is None:
|
||||
# default option, timesteps align with discrete inference steps
|
||||
if original_steps > self.config.num_train_timesteps:
|
||||
raise ValueError(
|
||||
f"`original_steps`: {original_steps} cannot be larger than `self.config.train_timesteps`:"
|
||||
f" {self.config.num_train_timesteps} as the unet model trained with this scheduler can only handle"
|
||||
f" maximal {self.config.num_train_timesteps} timesteps."
|
||||
)
|
||||
# TCD Timesteps Setting
|
||||
# The skipping step parameter k from the paper.
|
||||
k = self.config.num_train_timesteps // original_steps
|
||||
# TCD Training/Distillation Steps Schedule
|
||||
tcd_origin_timesteps = np.asarray(list(range(1, int(original_steps * strength) + 1))) * k - 1
|
||||
else:
|
||||
# customised option, sampled timesteps can be any arbitrary value
|
||||
tcd_origin_timesteps = np.asarray(list(range(0, int(self.config.num_train_timesteps * strength))))
|
||||
|
||||
# 2. Calculate the TCD inference timestep schedule.
|
||||
if timesteps is not None:
|
||||
# 2.1 Handle custom timestep schedules.
|
||||
train_timesteps = set(tcd_origin_timesteps)
|
||||
non_train_timesteps = []
|
||||
for i in range(1, len(timesteps)):
|
||||
if timesteps[i] >= timesteps[i - 1]:
|
||||
raise ValueError("`custom_timesteps` must be in descending order.")
|
||||
|
||||
if timesteps[i] not in train_timesteps:
|
||||
non_train_timesteps.append(timesteps[i])
|
||||
|
||||
if timesteps[0] >= self.config.num_train_timesteps:
|
||||
raise ValueError(
|
||||
f"`timesteps` must start before `self.config.train_timesteps`:"
|
||||
f" {self.config.num_train_timesteps}."
|
||||
)
|
||||
|
||||
# Raise warning if timestep schedule does not start with self.config.num_train_timesteps - 1
|
||||
if strength == 1.0 and timesteps[0] != self.config.num_train_timesteps - 1:
|
||||
logger.warning(
|
||||
f"The first timestep on the custom timestep schedule is {timesteps[0]}, not"
|
||||
f" `self.config.num_train_timesteps - 1`: {self.config.num_train_timesteps - 1}. You may get"
|
||||
f" unexpected results when using this timestep schedule."
|
||||
)
|
||||
|
||||
# Raise warning if custom timestep schedule contains timesteps not on original timestep schedule
|
||||
if non_train_timesteps:
|
||||
logger.warning(
|
||||
f"The custom timestep schedule contains the following timesteps which are not on the original"
|
||||
f" training/distillation timestep schedule: {non_train_timesteps}. You may get unexpected results"
|
||||
f" when using this timestep schedule."
|
||||
)
|
||||
|
||||
# Raise warning if custom timestep schedule is longer than original_steps
|
||||
if original_steps is not None:
|
||||
if len(timesteps) > original_steps:
|
||||
logger.warning(
|
||||
f"The number of timesteps in the custom timestep schedule is {len(timesteps)}, which exceeds the"
|
||||
f" the length of the timestep schedule used for training: {original_steps}. You may get some"
|
||||
f" unexpected results when using this timestep schedule."
|
||||
)
|
||||
else:
|
||||
if len(timesteps) > self.config.num_train_timesteps:
|
||||
logger.warning(
|
||||
f"The number of timesteps in the custom timestep schedule is {len(timesteps)}, which exceeds the"
|
||||
f" the length of the timestep schedule used for training: {self.config.num_train_timesteps}. You may get some"
|
||||
f" unexpected results when using this timestep schedule."
|
||||
)
|
||||
|
||||
timesteps = np.array(timesteps, dtype=np.int64)
|
||||
self.num_inference_steps = len(timesteps)
|
||||
self.custom_timesteps = True
|
||||
|
||||
# Apply strength (e.g. for img2img pipelines) (see StableDiffusionImg2ImgPipeline.get_timesteps)
|
||||
init_timestep = min(int(self.num_inference_steps * strength), self.num_inference_steps)
|
||||
t_start = max(self.num_inference_steps - init_timestep, 0)
|
||||
timesteps = timesteps[t_start * self.order :]
|
||||
# TODO: also reset self.num_inference_steps?
|
||||
else:
|
||||
# 2.2 Create the "standard" TCD inference timestep schedule.
|
||||
if num_inference_steps > self.config.num_train_timesteps:
|
||||
raise ValueError(
|
||||
f"`num_inference_steps`: {num_inference_steps} cannot be larger than `self.config.train_timesteps`:"
|
||||
f" {self.config.num_train_timesteps} as the unet model trained with this scheduler can only handle"
|
||||
f" maximal {self.config.num_train_timesteps} timesteps."
|
||||
)
|
||||
|
||||
if original_steps is not None:
|
||||
skipping_step = len(tcd_origin_timesteps) // num_inference_steps
|
||||
|
||||
if skipping_step < 1:
|
||||
raise ValueError(
|
||||
f"The combination of `original_steps x strength`: {original_steps} x {strength} is smaller than `num_inference_steps`: {num_inference_steps}. Make sure to either reduce `num_inference_steps` to a value smaller than {int(original_steps * strength)} or increase `strength` to a value higher than {float(num_inference_steps / original_steps)}."
|
||||
)
|
||||
|
||||
self.num_inference_steps = num_inference_steps
|
||||
|
||||
if original_steps is not None:
|
||||
if num_inference_steps > original_steps:
|
||||
raise ValueError(
|
||||
f"`num_inference_steps`: {num_inference_steps} cannot be larger than `original_inference_steps`:"
|
||||
f" {original_steps} because the final timestep schedule will be a subset of the"
|
||||
f" `original_inference_steps`-sized initial timestep schedule."
|
||||
)
|
||||
else:
|
||||
if num_inference_steps > self.config.num_train_timesteps:
|
||||
raise ValueError(
|
||||
f"`num_inference_steps`: {num_inference_steps} cannot be larger than `num_train_timesteps`:"
|
||||
f" {self.config.num_train_timesteps} because the final timestep schedule will be a subset of the"
|
||||
f" `num_train_timesteps`-sized initial timestep schedule."
|
||||
)
|
||||
|
||||
# TCD Inference Steps Schedule
|
||||
tcd_origin_timesteps = tcd_origin_timesteps[::-1].copy()
|
||||
# Select (approximately) evenly spaced indices from tcd_origin_timesteps.
|
||||
inference_indices = np.linspace(0, len(tcd_origin_timesteps), num=num_inference_steps, endpoint=False)
|
||||
inference_indices = np.floor(inference_indices).astype(np.int64)
|
||||
timesteps = tcd_origin_timesteps[inference_indices]
|
||||
|
||||
self.timesteps = torch.from_numpy(timesteps).to(device=device, dtype=torch.long)
|
||||
|
||||
self._step_index = None
|
||||
self._begin_index = None
|
||||
|
||||
def step(
|
||||
self,
|
||||
model_output: torch.FloatTensor,
|
||||
timestep: int,
|
||||
sample: torch.FloatTensor,
|
||||
eta: float = 0.3,
|
||||
generator: Optional[torch.Generator] = None,
|
||||
return_dict: bool = True,
|
||||
) -> Union[TCDSchedulerOutput, Tuple]:
|
||||
"""
|
||||
Predict the sample from the previous timestep by reversing the SDE. This function propagates the diffusion
|
||||
process from the learned model outputs (most often the predicted noise).
|
||||
|
||||
Args:
|
||||
model_output (`torch.FloatTensor`):
|
||||
The direct output from learned diffusion model.
|
||||
timestep (`int`):
|
||||
The current discrete timestep in the diffusion chain.
|
||||
sample (`torch.FloatTensor`):
|
||||
A current instance of a sample created by the diffusion process.
|
||||
eta (`float`):
|
||||
A stochastic parameter (referred to as `gamma` in the paper) used to control the stochasticity in every
|
||||
step. When eta = 0, it represents deterministic sampling, whereas eta = 1 indicates full stochastic
|
||||
sampling.
|
||||
generator (`torch.Generator`, *optional*):
|
||||
A random number generator.
|
||||
return_dict (`bool`, *optional*, defaults to `True`):
|
||||
Whether or not to return a [`~schedulers.scheduling_tcd.TCDSchedulerOutput`] or `tuple`.
|
||||
Returns:
|
||||
[`~schedulers.scheduling_utils.TCDSchedulerOutput`] or `tuple`:
|
||||
If return_dict is `True`, [`~schedulers.scheduling_tcd.TCDSchedulerOutput`] is returned, otherwise a
|
||||
tuple is returned where the first element is the sample tensor.
|
||||
"""
|
||||
if self.num_inference_steps is None:
|
||||
raise ValueError(
|
||||
"Number of inference steps is 'None', you need to run 'set_timesteps' after creating the scheduler"
|
||||
)
|
||||
|
||||
if self.step_index is None:
|
||||
self._init_step_index(timestep)
|
||||
|
||||
assert 0 <= eta <= 1.0, "gamma must be less than or equal to 1.0"
|
||||
|
||||
# 1. get previous step value
|
||||
prev_step_index = self.step_index + 1
|
||||
if prev_step_index < len(self.timesteps):
|
||||
prev_timestep = self.timesteps[prev_step_index]
|
||||
else:
|
||||
prev_timestep = torch.tensor(0)
|
||||
|
||||
timestep_s = torch.floor((1 - eta) * prev_timestep).to(dtype=torch.long)
|
||||
|
||||
# 2. compute alphas, betas
|
||||
alpha_prod_t = self.alphas_cumprod[timestep]
|
||||
beta_prod_t = 1 - alpha_prod_t
|
||||
|
||||
alpha_prod_t_prev = self.alphas_cumprod[prev_timestep] if prev_timestep >= 0 else self.final_alpha_cumprod
|
||||
|
||||
alpha_prod_s = self.alphas_cumprod[timestep_s]
|
||||
beta_prod_s = 1 - alpha_prod_s
|
||||
|
||||
# 3. Compute the predicted noised sample x_s based on the model parameterization
|
||||
if self.config.prediction_type == "epsilon": # noise-prediction
|
||||
pred_original_sample = (sample - beta_prod_t.sqrt() * model_output) / alpha_prod_t.sqrt()
|
||||
pred_epsilon = model_output
|
||||
pred_noised_sample = alpha_prod_s.sqrt() * pred_original_sample + beta_prod_s.sqrt() * pred_epsilon
|
||||
elif self.config.prediction_type == "sample": # x-prediction
|
||||
pred_original_sample = model_output
|
||||
pred_epsilon = (sample - alpha_prod_t ** (0.5) * pred_original_sample) / beta_prod_t ** (0.5)
|
||||
pred_noised_sample = alpha_prod_s.sqrt() * pred_original_sample + beta_prod_s.sqrt() * pred_epsilon
|
||||
elif self.config.prediction_type == "v_prediction": # v-prediction
|
||||
pred_original_sample = (alpha_prod_t**0.5) * sample - (beta_prod_t**0.5) * model_output
|
||||
pred_epsilon = (alpha_prod_t**0.5) * model_output + (beta_prod_t**0.5) * sample
|
||||
pred_noised_sample = alpha_prod_s.sqrt() * pred_original_sample + beta_prod_s.sqrt() * pred_epsilon
|
||||
else:
|
||||
raise ValueError(
|
||||
f"prediction_type given as {self.config.prediction_type} must be one of `epsilon`, `sample` or"
|
||||
" `v_prediction` for `TCDScheduler`."
|
||||
)
|
||||
|
||||
# 4. Sample and inject noise z ~ N(0, I) for MultiStep Inference
|
||||
# Noise is not used on the final timestep of the timestep schedule.
|
||||
# This also means that noise is not used for one-step sampling.
|
||||
# Eta (referred to as "gamma" in the paper) was introduced to control the stochasticity in every step.
|
||||
# When eta = 0, it represents deterministic sampling, whereas eta = 1 indicates full stochastic sampling.
|
||||
if eta > 0:
|
||||
if self.step_index != self.num_inference_steps - 1:
|
||||
noise = randn_tensor(
|
||||
model_output.shape, generator=generator, device=model_output.device, dtype=pred_noised_sample.dtype
|
||||
)
|
||||
prev_sample = (alpha_prod_t_prev / alpha_prod_s).sqrt() * pred_noised_sample + (
|
||||
1 - alpha_prod_t_prev / alpha_prod_s
|
||||
).sqrt() * noise
|
||||
else:
|
||||
prev_sample = pred_noised_sample
|
||||
else:
|
||||
prev_sample = pred_noised_sample
|
||||
|
||||
# upon completion increase step index by one
|
||||
self._step_index += 1
|
||||
|
||||
if not return_dict:
|
||||
return (prev_sample, pred_noised_sample)
|
||||
|
||||
return TCDSchedulerOutput(prev_sample=prev_sample, pred_noised_sample=pred_noised_sample)
|
||||
|
||||
# Copied from diffusers.schedulers.scheduling_ddpm.DDPMScheduler.add_noise
|
||||
def add_noise(
|
||||
self,
|
||||
original_samples: torch.FloatTensor,
|
||||
noise: torch.FloatTensor,
|
||||
timesteps: torch.IntTensor,
|
||||
) -> torch.FloatTensor:
|
||||
# Make sure alphas_cumprod and timestep have same device and dtype as original_samples
|
||||
# Move the self.alphas_cumprod to device to avoid redundant CPU to GPU data movement
|
||||
# for the subsequent add_noise calls
|
||||
self.alphas_cumprod = self.alphas_cumprod.to(device=original_samples.device)
|
||||
alphas_cumprod = self.alphas_cumprod.to(dtype=original_samples.dtype)
|
||||
timesteps = timesteps.to(original_samples.device)
|
||||
|
||||
sqrt_alpha_prod = alphas_cumprod[timesteps] ** 0.5
|
||||
sqrt_alpha_prod = sqrt_alpha_prod.flatten()
|
||||
while len(sqrt_alpha_prod.shape) < len(original_samples.shape):
|
||||
sqrt_alpha_prod = sqrt_alpha_prod.unsqueeze(-1)
|
||||
|
||||
sqrt_one_minus_alpha_prod = (1 - alphas_cumprod[timesteps]) ** 0.5
|
||||
sqrt_one_minus_alpha_prod = sqrt_one_minus_alpha_prod.flatten()
|
||||
while len(sqrt_one_minus_alpha_prod.shape) < len(original_samples.shape):
|
||||
sqrt_one_minus_alpha_prod = sqrt_one_minus_alpha_prod.unsqueeze(-1)
|
||||
|
||||
noisy_samples = sqrt_alpha_prod * original_samples + sqrt_one_minus_alpha_prod * noise
|
||||
return noisy_samples
|
||||
|
||||
# Copied from diffusers.schedulers.scheduling_ddpm.DDPMScheduler.get_velocity
|
||||
def get_velocity(
|
||||
self, sample: torch.FloatTensor, noise: torch.FloatTensor, timesteps: torch.IntTensor
|
||||
) -> torch.FloatTensor:
|
||||
# Make sure alphas_cumprod and timestep have same device and dtype as sample
|
||||
self.alphas_cumprod = self.alphas_cumprod.to(device=sample.device)
|
||||
alphas_cumprod = self.alphas_cumprod.to(dtype=sample.dtype)
|
||||
timesteps = timesteps.to(sample.device)
|
||||
|
||||
sqrt_alpha_prod = alphas_cumprod[timesteps] ** 0.5
|
||||
sqrt_alpha_prod = sqrt_alpha_prod.flatten()
|
||||
while len(sqrt_alpha_prod.shape) < len(sample.shape):
|
||||
sqrt_alpha_prod = sqrt_alpha_prod.unsqueeze(-1)
|
||||
|
||||
sqrt_one_minus_alpha_prod = (1 - alphas_cumprod[timesteps]) ** 0.5
|
||||
sqrt_one_minus_alpha_prod = sqrt_one_minus_alpha_prod.flatten()
|
||||
while len(sqrt_one_minus_alpha_prod.shape) < len(sample.shape):
|
||||
sqrt_one_minus_alpha_prod = sqrt_one_minus_alpha_prod.unsqueeze(-1)
|
||||
|
||||
velocity = sqrt_alpha_prod * noise - sqrt_one_minus_alpha_prod * sample
|
||||
return velocity
|
||||
|
||||
def __len__(self):
|
||||
return self.config.num_train_timesteps
|
||||
|
||||
# Copied from diffusers.schedulers.scheduling_ddpm.DDPMScheduler.previous_timestep
|
||||
def previous_timestep(self, timestep):
|
||||
if self.custom_timesteps:
|
||||
index = (self.timesteps == timestep).nonzero(as_tuple=True)[0][0]
|
||||
if index == self.timesteps.shape[0] - 1:
|
||||
prev_t = torch.tensor(-1)
|
||||
else:
|
||||
prev_t = self.timesteps[index + 1]
|
||||
else:
|
||||
num_inference_steps = (
|
||||
self.num_inference_steps if self.num_inference_steps else self.config.num_train_timesteps
|
||||
)
|
||||
prev_t = timestep - self.config.num_train_timesteps // num_inference_steps
|
||||
|
||||
return prev_t
|
||||
Reference in New Issue
Block a user