Spaces:
Running on Zero
Running on Zero
File size: 24,770 Bytes
b701455 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 | import logging
import math
from typing import Any, Dict, List, Optional
import torch.nn as nn
import torch
from src.Utilities import util
from src.AutoEncoders import ResBlock
from src.NeuralNetwork import transformer
from src.cond import cast
from src.sample import sampling, sampling_util
UNET_MAP_ATTENTIONS = {"proj_in.weight", "proj_in.bias", "proj_out.weight", "proj_out.bias", "norm.weight", "norm.bias"}
TRANSFORMER_BLOCKS = {
"norm1.weight", "norm1.bias", "norm2.weight", "norm2.bias", "norm3.weight", "norm3.bias",
"attn1.to_q.weight", "attn1.to_k.weight", "attn1.to_v.weight", "attn1.to_out.0.weight", "attn1.to_out.0.bias",
"attn2.to_q.weight", "attn2.to_k.weight", "attn2.to_v.weight", "attn2.to_out.0.weight", "attn2.to_out.0.bias",
"ff.net.0.proj.weight", "ff.net.0.proj.bias", "ff.net.2.weight", "ff.net.2.bias",
}
UNET_MAP_RESNET = {
"in_layers.2.weight": "conv1.weight", "in_layers.2.bias": "conv1.bias",
"emb_layers.1.weight": "time_emb_proj.weight", "emb_layers.1.bias": "time_emb_proj.bias",
"out_layers.3.weight": "conv2.weight", "out_layers.3.bias": "conv2.bias",
"skip_connection.weight": "conv_shortcut.weight", "skip_connection.bias": "conv_shortcut.bias",
"in_layers.0.weight": "norm1.weight", "in_layers.0.bias": "norm1.bias",
"out_layers.0.weight": "norm2.weight", "out_layers.0.bias": "norm2.bias",
}
UNET_MAP_BASIC = {
("label_emb.0.0.weight", "class_embedding.linear_1.weight"), ("label_emb.0.0.bias", "class_embedding.linear_1.bias"),
("label_emb.0.2.weight", "class_embedding.linear_2.weight"), ("label_emb.0.2.bias", "class_embedding.linear_2.bias"),
("label_emb.0.0.weight", "add_embedding.linear_1.weight"), ("label_emb.0.0.bias", "add_embedding.linear_1.bias"),
("label_emb.0.2.weight", "add_embedding.linear_2.weight"), ("label_emb.0.2.bias", "add_embedding.linear_2.bias"),
("input_blocks.0.0.weight", "conv_in.weight"), ("input_blocks.0.0.bias", "conv_in.bias"),
("out.0.weight", "conv_norm_out.weight"), ("out.0.bias", "conv_norm_out.bias"),
("out.2.weight", "conv_out.weight"), ("out.2.bias", "conv_out.bias"),
("time_embed.0.weight", "time_embedding.linear_1.weight"), ("time_embed.0.bias", "time_embedding.linear_1.bias"),
("time_embed.2.weight", "time_embedding.linear_2.weight"), ("time_embed.2.bias", "time_embedding.linear_2.bias"),
}
oai_ops = cast.disable_weight_init
def unet_to_diffusers(unet_config: dict) -> dict:
if "num_res_blocks" not in unet_config:
return {}
num_res_blocks, channel_mult = unet_config["num_res_blocks"], unet_config["channel_mult"]
transformer_depth, transformer_depth_output = unet_config["transformer_depth"][:], unet_config["transformer_depth_output"][:]
num_blocks, transformers_mid = len(channel_mult), unet_config.get("transformer_depth_middle", None)
diffusers_unet_map = {}
for x in range(num_blocks):
n = 1 + (num_res_blocks[x] + 1) * x
for i in range(num_res_blocks[x]):
for b in UNET_MAP_RESNET:
diffusers_unet_map[f"down_blocks.{x}.resnets.{i}.{UNET_MAP_RESNET[b]}"] = f"input_blocks.{n}.0.{b}"
num_transformers = transformer_depth.pop(0)
if num_transformers > 0:
for b in UNET_MAP_ATTENTIONS:
diffusers_unet_map[f"down_blocks.{x}.attentions.{i}.{b}"] = f"input_blocks.{n}.1.{b}"
for t in range(num_transformers):
for b in TRANSFORMER_BLOCKS:
diffusers_unet_map[f"down_blocks.{x}.attentions.{i}.transformer_blocks.{t}.{b}"] = f"input_blocks.{n}.1.transformer_blocks.{t}.{b}"
n += 1
for k in ["weight", "bias"]:
diffusers_unet_map[f"down_blocks.{x}.downsamplers.0.conv.{k}"] = f"input_blocks.{n}.0.op.{k}"
for b in UNET_MAP_ATTENTIONS:
diffusers_unet_map[f"mid_block.attentions.0.{b}"] = f"middle_block.1.{b}"
for t in range(transformers_mid):
for b in TRANSFORMER_BLOCKS:
diffusers_unet_map[f"mid_block.attentions.0.transformer_blocks.{t}.{b}"] = f"middle_block.1.transformer_blocks.{t}.{b}"
for i, n in enumerate([0, 2]):
for b in UNET_MAP_RESNET:
diffusers_unet_map[f"mid_block.resnets.{i}.{UNET_MAP_RESNET[b]}"] = f"middle_block.{n}.{b}"
num_res_blocks = list(reversed(num_res_blocks))
for x in range(num_blocks):
n = (num_res_blocks[x] + 1) * x
for i in range(num_res_blocks[x] + 1):
c = 0
for b in UNET_MAP_RESNET:
diffusers_unet_map[f"up_blocks.{x}.resnets.{i}.{UNET_MAP_RESNET[b]}"] = f"output_blocks.{n}.0.{b}"
c += 1
num_transformers = transformer_depth_output.pop()
if num_transformers > 0:
c += 1
for b in UNET_MAP_ATTENTIONS:
diffusers_unet_map[f"up_blocks.{x}.attentions.{i}.{b}"] = f"output_blocks.{n}.1.{b}"
for t in range(num_transformers):
for b in TRANSFORMER_BLOCKS:
diffusers_unet_map[f"up_blocks.{x}.attentions.{i}.transformer_blocks.{t}.{b}"] = f"output_blocks.{n}.1.transformer_blocks.{t}.{b}"
if i == num_res_blocks[x]:
for k in ["weight", "bias"]:
diffusers_unet_map[f"up_blocks.{x}.upsamplers.0.conv.{k}"] = f"output_blocks.{n}.{c}.conv.{k}"
n += 1
for k in UNET_MAP_BASIC:
diffusers_unet_map[k[1]] = k[0]
return diffusers_unet_map
def apply_control1(h: torch.Tensor, control: any, name: str) -> torch.Tensor:
return h
class UNetModel1(nn.Module):
def __init__(self, image_size: int, in_channels: int, model_channels: int, out_channels: int, num_res_blocks: list,
dropout: float = 0, channel_mult: tuple = (1, 2, 4, 8), conv_resample: bool = True, dims: int = 2,
num_classes: int = None, use_checkpoint: bool = False, dtype: torch.dtype = torch.float32,
num_heads: int = -1, num_head_channels: int = -1, num_heads_upsample: int = -1,
use_scale_shift_norm: bool = False, resblock_updown: bool = False, use_new_attention_order: bool = False,
use_spatial_transformer: bool = False, transformer_depth: int = 1, context_dim: int = None,
n_embed: int = None, legacy: bool = True, disable_self_attentions: list = None,
num_attention_blocks: list = None, disable_middle_self_attn: bool = False,
use_linear_in_transformer: bool = False, adm_in_channels: int = None,
transformer_depth_middle: int = None, transformer_depth_output: list = None,
use_temporal_resblock: bool = False, use_temporal_attention: bool = False,
time_context_dim: int = None, extra_ff_mix_layer: bool = False, use_spatial_context: bool = False,
merge_strategy: any = None, merge_factor: float = 0.0, video_kernel_size: int = None,
disable_temporal_crossattention: bool = False, max_ddpm_temb_period: int = 10000,
device: torch.device = None, operations: any = oai_ops):
super().__init__()
if context_dim is not None:
self.context_dim = context_dim
if num_heads_upsample == -1:
num_heads_upsample = num_heads
if num_head_channels == -1:
assert num_heads != -1
self.in_channels, self.model_channels, self.out_channels = in_channels, model_channels, out_channels
self.num_res_blocks, self.dropout, self.channel_mult = num_res_blocks, dropout, channel_mult
self.conv_resample, self.num_classes, self.use_checkpoint = conv_resample, num_classes, use_checkpoint
self.dtype, self.num_heads, self.num_head_channels = dtype, num_heads, num_head_channels
self.num_heads_upsample, self.use_temporal_resblocks = num_heads_upsample, use_temporal_resblock
self.predict_codebook_ids, self.default_num_video_frames = n_embed is not None, None
transformer_depth, transformer_depth_output = transformer_depth[:], transformer_depth_output[:]
time_embed_dim = model_channels * 4
self.time_embed = nn.Sequential(
operations.Linear(model_channels, time_embed_dim, dtype=self.dtype, device=device), nn.SiLU(),
operations.Linear(time_embed_dim, time_embed_dim, dtype=self.dtype, device=device))
if adm_in_channels is not None:
self.label_emb = nn.Sequential(
nn.Sequential(
operations.Linear(adm_in_channels, time_embed_dim, dtype=self.dtype, device=device),
nn.SiLU(),
operations.Linear(time_embed_dim, time_embed_dim, dtype=self.dtype, device=device)))
self.input_blocks = nn.ModuleList([sampling.TimestepEmbedSequential1(
operations.conv_nd(dims, in_channels, model_channels, 3, padding=1, dtype=self.dtype, device=device))])
self._feature_size, input_block_chans, ch, ds = model_channels, [model_channels], model_channels, 1
self.double_blocks = nn.ModuleList()
def make_attn(ch, depth, context_dim, disable_self_attn=False):
dim_head = ch // num_heads if num_head_channels == -1 else num_head_channels
heads = num_heads if num_head_channels == -1 else ch // num_head_channels
return transformer.SpatialTransformer(ch, heads, dim_head, depth=depth, context_dim=context_dim,
disable_self_attn=disable_self_attn, use_linear=use_linear_in_transformer,
use_checkpoint=use_checkpoint, dtype=self.dtype, device=device, operations=operations)
def make_res(ch, out_ch=None, down=False, up=False):
return ResBlock.ResBlock1(channels=ch, emb_channels=time_embed_dim, dropout=dropout,
out_channels=out_ch, use_checkpoint=use_checkpoint, dims=dims,
use_scale_shift_norm=use_scale_shift_norm, down=down, up=up,
dtype=self.dtype, device=device, operations=operations)
for level, mult in enumerate(channel_mult):
for nr in range(self.num_res_blocks[level]):
layers = [make_res(ch, mult * model_channels)]
ch = mult * model_channels
num_trans = transformer_depth.pop(0)
if num_trans > 0 and (not util.exists(num_attention_blocks) or nr < num_attention_blocks[level]):
layers.append(make_attn(ch, num_trans, context_dim))
self.input_blocks.append(sampling.TimestepEmbedSequential1(*layers))
self._feature_size += ch
input_block_chans.append(ch)
if level != len(channel_mult) - 1:
out_ch = ch
self.input_blocks.append(sampling.TimestepEmbedSequential1(
make_res(ch, out_ch, down=True) if resblock_updown else
ResBlock.Downsample1(ch, conv_resample, dims=dims, out_channels=out_ch, dtype=self.dtype, device=device, operations=operations)))
ch, input_block_chans, ds = out_ch, input_block_chans + [out_ch], ds * 2
self._feature_size += ch
dim_head = ch // num_heads if num_head_channels == -1 else num_head_channels
mid_block = [make_res(ch)]
self.middle_block = None
if transformer_depth_middle >= -1:
if transformer_depth_middle >= 0:
mid_block += [make_attn(ch, transformer_depth_middle, context_dim, disable_middle_self_attn), make_res(ch)]
self.middle_block = sampling.TimestepEmbedSequential1(*mid_block)
self._feature_size += ch
self.output_blocks = nn.ModuleList([])
for level, mult in list(enumerate(channel_mult))[::-1]:
for i in range(self.num_res_blocks[level] + 1):
ich = input_block_chans.pop()
layers = [make_res(ch + ich, model_channels * mult)]
ch = model_channels * mult
num_trans = transformer_depth_output.pop()
if num_trans > 0 and (not util.exists(num_attention_blocks) or i < num_attention_blocks[level]):
layers.append(make_attn(ch, num_trans, context_dim))
if level and i == self.num_res_blocks[level]:
layers.append(make_res(ch, ch, up=True) if resblock_updown else
ResBlock.Upsample1(ch, conv_resample, dims=dims, out_channels=ch, dtype=self.dtype, device=device, operations=operations))
ds //= 2
self.output_blocks.append(sampling.TimestepEmbedSequential1(*layers))
self._feature_size += ch
self.out = nn.Sequential(operations.GroupNorm(32, ch, dtype=self.dtype, device=device), nn.SiLU(),
util.zero_module(operations.conv_nd(dims, model_channels, out_channels, 3, padding=1, dtype=self.dtype, device=device)))
def forward(self, x: torch.Tensor, timesteps: Optional[torch.Tensor] = None, context: Optional[torch.Tensor] = None,
y: Optional[torch.Tensor] = None, control: Optional[torch.Tensor] = None,
transformer_options: Dict[str, Any] = {}, **kwargs: Any) -> torch.Tensor:
transformer_options["original_shape"], transformer_options["transformer_index"] = list(x.shape), 0
num_video_frames = kwargs.get("num_video_frames", self.default_num_video_frames)
image_only_indicator, time_context = kwargs.get("image_only_indicator"), kwargs.get("time_context")
if self.num_classes is None:
y = None
elif y is None:
raise ValueError("y is required for models with num_classes")
emb = self.time_embed(sampling_util.timestep_embedding(timesteps, self.model_channels).to(device=x.device, dtype=x.dtype))
if y is not None:
emb = emb + self.label_emb(y.to(device=x.device, dtype=x.dtype))
hs, h = [], x
for id, module in enumerate(self.input_blocks):
transformer_options["block"] = ("input", id)
h = ResBlock.forward_timestep_embed1(module, h, emb, context, transformer_options,
time_context=time_context, num_video_frames=num_video_frames, image_only_indicator=image_only_indicator)
h = apply_control1(h, control, "input")
hs.append(h)
transformer_options["block"] = ("middle", 0)
if self.middle_block is not None:
h = ResBlock.forward_timestep_embed1(self.middle_block, h, emb, context, transformer_options,
time_context=time_context, num_video_frames=num_video_frames, image_only_indicator=image_only_indicator)
h = apply_control1(h, control, "middle")
for id, module in enumerate(self.output_blocks):
transformer_options["block"] = ("output", id)
hsp = apply_control1(hs.pop(), control, "output")
h = torch.cat([h, hsp], dim=1)
del hsp
h = ResBlock.forward_timestep_embed1(module, h, emb, context, transformer_options,
hs[-1].shape if hs else None, time_context=time_context, num_video_frames=num_video_frames, image_only_indicator=image_only_indicator)
return self.out(h.type(x.dtype))
def detect_unet_config(state_dict: Dict[str, torch.Tensor], key_prefix: str) -> Dict[str, Any]:
state_dict_keys = list(state_dict.keys())
# MMDIT model
if f"{key_prefix}joint_blocks.0.context_block.attn.qkv.weight" in state_dict_keys:
cfg = {"in_channels": state_dict[f"{key_prefix}x_embedder.proj.weight"].shape[1],
"patch_size": state_dict[f"{key_prefix}x_embedder.proj.weight"].shape[2],
"depth": state_dict[f"{key_prefix}x_embedder.proj.weight"].shape[0] // 64, "input_size": None}
if f"{key_prefix}final_layer.linear.weight" in state_dict:
cfg["out_channels"] = state_dict[f"{key_prefix}final_layer.linear.weight"].shape[0] // (cfg["patch_size"] ** 2)
if f"{key_prefix}y_embedder.mlp.0.weight" in state_dict_keys:
cfg["adm_in_channels"] = state_dict[f"{key_prefix}y_embedder.mlp.0.weight"].shape[1]
if f"{key_prefix}context_embedder.weight" in state_dict_keys:
w = state_dict[f"{key_prefix}context_embedder.weight"]
cfg["context_embedder_config"] = {"target": "torch.nn.Linear", "params": {"in_features": w.shape[1], "out_features": w.shape[0]}}
if f"{key_prefix}pos_embed" in state_dict_keys:
cfg["num_patches"] = state_dict[f"{key_prefix}pos_embed"].shape[1]
cfg["pos_embed_max_size"] = round(math.sqrt(cfg["num_patches"]))
if f"{key_prefix}joint_blocks.0.context_block.attn.ln_q.weight" in state_dict_keys:
cfg["qk_norm"] = "rms"
cfg["pos_embed_scaling_factor"] = None
if f"{key_prefix}context_processor.layers.0.attn.qkv.weight" in state_dict_keys:
cfg["context_processor_layers"] = transformer.count_blocks(state_dict_keys, f"{key_prefix}context_processor.layers." + "{}.")
return cfg
# Stable Cascade
if f"{key_prefix}clf.1.weight" in state_dict_keys:
cfg = {}
if f"{key_prefix}clip_txt_mapper.weight" in state_dict_keys:
cfg["stable_cascade_stage"] = "c"
w = state_dict[f"{key_prefix}clip_txt_mapper.weight"]
if w.shape[0] == 1536:
cfg.update({"c_cond": 1536, "c_hidden": [1536, 1536], "nhead": [24, 24], "blocks": [[4, 12], [12, 4]]})
elif w.shape[0] == 2048:
cfg["c_cond"] = 2048
elif f"{key_prefix}clip_mapper.weight" in state_dict_keys:
cfg["stable_cascade_stage"] = "b"
w = state_dict[f"{key_prefix}down_blocks.1.0.channelwise.0.weight"]
if w.shape[-1] == 640:
cfg.update({"c_hidden": [320, 640, 1280, 1280], "nhead": [-1, -1, 20, 20], "blocks": [[2, 6, 28, 6], [6, 28, 6, 2]], "block_repeat": [[1, 1, 1, 1], [3, 3, 2, 2]]})
elif w.shape[-1] == 576:
cfg.update({"c_hidden": [320, 576, 1152, 1152], "nhead": [-1, 9, 18, 18], "blocks": [[2, 4, 14, 4], [4, 14, 4, 2]], "block_repeat": [[1, 1, 1, 1], [2, 2, 2, 2]]})
return cfg
# Stable Audio DIT
if f"{key_prefix}transformer.rotary_pos_emb.inv_freq" in state_dict_keys:
return {"audio_model": "dit1.0"}
# Aura Flow DIT
if f"{key_prefix}double_layers.0.attn.w1q.weight" in state_dict_keys:
double_layers = transformer.count_blocks(state_dict_keys, f"{key_prefix}double_layers." + "{}.")
single_layers = transformer.count_blocks(state_dict_keys, f"{key_prefix}single_layers." + "{}.")
return {"max_seq": state_dict[f"{key_prefix}positional_encoding"].shape[1],
"cond_seq_dim": state_dict[f"{key_prefix}cond_seq_linear.weight"].shape[1],
"n_double_layers": double_layers, "n_layers": double_layers + single_layers}
# Hunyuan DiT
if f"{key_prefix}mlp_t5.0.weight" in state_dict_keys:
cfg = {"image_model": "hydit", "depth": transformer.count_blocks(state_dict_keys, f"{key_prefix}blocks." + "{}."),
"hidden_size": state_dict[f"{key_prefix}x_embedder.proj.weight"].shape[0]}
if cfg["hidden_size"] == 1408 and cfg["depth"] == 40:
cfg["mlp_ratio"] = 4.3637
if state_dict[f"{key_prefix}extra_embedder.0.weight"].shape[1] == 3968:
cfg.update({"size_cond": True, "use_style_cond": True, "image_model": "hydit1"})
return cfg
# Flux
if f"{key_prefix}double_blocks.0.img_attn.norm.key_norm.scale" in state_dict_keys:
return {"image_model": "flux", "in_channels": 16, "vec_in_dim": 768, "context_in_dim": 4096,
"hidden_size": 3072, "mlp_ratio": 4.0, "num_heads": 24,
"depth": transformer.count_blocks(state_dict_keys, f"{key_prefix}double_blocks." + "{}."),
"depth_single_blocks": transformer.count_blocks(state_dict_keys, f"{key_prefix}single_blocks." + "{}."),
"axes_dim": [16, 56, 56], "theta": 10000, "qkv_bias": True,
"guidance_embed": f"{key_prefix}guidance_in.in_layer.weight" in state_dict_keys}
if f"{key_prefix}input_blocks.0.0.weight" not in state_dict_keys:
return None
# Standard UNet
cfg = {"use_checkpoint": False, "image_size": 32, "use_spatial_transformer": True, "legacy": False}
if f"{key_prefix}label_emb.0.0.weight" in state_dict_keys:
cfg["num_classes"], cfg["adm_in_channels"] = "sequential", state_dict[f"{key_prefix}label_emb.0.0.weight"].shape[1]
else:
cfg["adm_in_channels"] = None
model_channels = state_dict[f"{key_prefix}input_blocks.0.0.weight"].shape[0]
in_channels = state_dict[f"{key_prefix}input_blocks.0.0.weight"].shape[1]
out_channels = state_dict.get(f"{key_prefix}out.2.weight", torch.zeros(4)).shape[0] or 4
num_res_blocks, channel_mult, transformer_depth, transformer_depth_output = [], [], [], []
context_dim, use_linear_in_transformer = None, False
current_res, last_res_blocks, last_channel_mult = 1, 0, 0
input_block_count = transformer.count_blocks(state_dict_keys, f"{key_prefix}input_blocks" + ".{}.")
for count in range(input_block_count):
prefix = f"{key_prefix}input_blocks.{count}."
prefix_output = f"{key_prefix}output_blocks.{input_block_count - count - 1}."
block_keys = sorted([k for k in state_dict_keys if k.startswith(prefix)])
if not block_keys:
break
block_keys_output = sorted([k for k in state_dict_keys if k.startswith(prefix_output)])
if f"{prefix}0.op.weight" in block_keys:
num_res_blocks.append(last_res_blocks)
channel_mult.append(last_channel_mult)
current_res *= 2
last_res_blocks, last_channel_mult = 0, 0
out = transformer.calculate_transformer_depth(prefix_output, state_dict_keys, state_dict)
transformer_depth_output.append(out[0] if out else 0)
else:
if f"{prefix}0.in_layers.0.weight" in block_keys:
last_res_blocks += 1
last_channel_mult = state_dict[f"{prefix}0.out_layers.3.weight"].shape[0] // model_channels
out = transformer.calculate_transformer_depth(prefix, state_dict_keys, state_dict)
if out:
transformer_depth.append(out[0])
if context_dim is None:
context_dim, use_linear_in_transformer = out[1], out[2]
else:
transformer_depth.append(0)
if f"{prefix_output}0.in_layers.0.weight" in block_keys_output:
out = transformer.calculate_transformer_depth(prefix_output, state_dict_keys, state_dict)
transformer_depth_output.append(out[0] if out else 0)
num_res_blocks.append(last_res_blocks)
channel_mult.append(last_channel_mult)
if f"{key_prefix}middle_block.1.proj_in.weight" in state_dict_keys:
transformer_depth_middle = transformer.count_blocks(state_dict_keys, f"{key_prefix}middle_block.1.transformer_blocks." + "{}")
elif f"{key_prefix}middle_block.0.in_layers.0.weight" in state_dict_keys:
transformer_depth_middle = -1
else:
transformer_depth_middle = -2
cfg.update({"in_channels": in_channels, "out_channels": out_channels, "model_channels": model_channels,
"num_res_blocks": num_res_blocks, "transformer_depth": transformer_depth,
"transformer_depth_output": transformer_depth_output, "channel_mult": channel_mult,
"transformer_depth_middle": transformer_depth_middle, "use_linear_in_transformer": use_linear_in_transformer,
"context_dim": context_dim, "use_temporal_resblock": False, "use_temporal_attention": False})
return cfg
def model_config_from_unet_config(unet_config: Dict[str, Any], state_dict: Optional[Dict[str, torch.Tensor]] = None) -> Any:
from src.SD15 import SD15
for model_config in SD15.models:
if model_config.matches(unet_config, state_dict):
return model_config(unet_config)
logging.error(f"no match {unet_config}")
return None
def model_config_from_unet(state_dict: Dict[str, torch.Tensor], unet_key_prefix: str, use_base_if_no_match: bool = False) -> Any:
unet_config = detect_unet_config(state_dict, unet_key_prefix)
return model_config_from_unet_config(unet_config, state_dict) if unet_config else None
def unet_dtype1(device: Optional[torch.device] = None, model_params: int = 0,
supported_dtypes: List[torch.dtype] = [torch.float16, torch.bfloat16, torch.float32]) -> torch.dtype:
return torch.float16
|