Merge pull request #16030 from AUTOMATIC1111/sd3

Stable Diffusion 3 support

README.md

@@ -150,7 +150,7 @@ For the purposes of getting Google and other search engines to crawl the wiki, h
 ## Credits
 Licenses for borrowed code can be found in `Settings -> Licenses` screen, and also in `html/licenses.html` file.
 
-- Stable Diffusion - https://github.com/Stability-AI/stablediffusion, https://github.com/CompVis/taming-transformers
+- Stable Diffusion - https://github.com/Stability-AI/stablediffusion, https://github.com/CompVis/taming-transformers, https://github.com/mcmonkey4eva/sd3-ref
 - k-diffusion - https://github.com/crowsonkb/k-diffusion.git
 - Spandrel - https://github.com/chaiNNer-org/spandrel implementing
   - GFPGAN - https://github.com/TencentARC/GFPGAN.git

configs/sd3-inference.yaml

@@ -0,0 +1,5 @@
+model:
+  target: modules.models.sd3.sd3_model.SD3Inferencer
+  params:
+    shift: 3
+    state_dict: null

extensions-builtin/Lora/networks.py

@@ -130,7 +130,9 @@ def assign_network_names_to_compvis_modules(sd_model):
                 network_layer_mapping[network_name] = module
                 module.network_layer_name = network_name
     else:
-        for name, module in shared.sd_model.cond_stage_model.wrapped.named_modules():
+        cond_stage_model = getattr(shared.sd_model.cond_stage_model, 'wrapped', shared.sd_model.cond_stage_model)
+
+        for name, module in cond_stage_model.named_modules():
             network_name = name.replace(".", "_")
             network_layer_mapping[network_name] = module
             module.network_layer_name = network_name

modules/deepbooru.py

@@ -57,7 +57,7 @@ class DeepDanbooru:
         a = np.expand_dims(np.array(pic, dtype=np.float32), 0) / 255
 
         with torch.no_grad(), devices.autocast():
-            x = torch.from_numpy(a).to(devices.device)
+            x = torch.from_numpy(a).to(devices.device, devices.dtype)
             y = self.model(x)[0].detach().cpu().numpy()
 
         probability_dict = {}

modules/lowvram.py

@@ -1,9 +1,12 @@
+from collections import namedtuple
+
 import torch
 from modules import devices, shared
 
 module_in_gpu = None
 cpu = torch.device("cpu")
 
+ModuleWithParent = namedtuple('ModuleWithParent', ['module', 'parent'], defaults=['None'])
 
 def send_everything_to_cpu():
     global module_in_gpu
@@ -75,13 +78,14 @@ def setup_for_low_vram(sd_model, use_medvram):
         (sd_model, 'depth_model'),
         (sd_model, 'embedder'),
         (sd_model, 'model'),
-        (sd_model, 'embedder'),
     ]
 
     is_sdxl = hasattr(sd_model, 'conditioner')
     is_sd2 = not is_sdxl and hasattr(sd_model.cond_stage_model, 'model')
 
-    if is_sdxl:
+    if hasattr(sd_model, 'medvram_fields'):
+        to_remain_in_cpu = sd_model.medvram_fields()
+    elif is_sdxl:
         to_remain_in_cpu.append((sd_model, 'conditioner'))
     elif is_sd2:
         to_remain_in_cpu.append((sd_model.cond_stage_model, 'model'))
@@ -103,7 +107,21 @@ def setup_for_low_vram(sd_model, use_medvram):
         setattr(obj, field, module)
 
     # register hooks for those the first three models
-    if is_sdxl:
+    if hasattr(sd_model, "cond_stage_model") and hasattr(sd_model.cond_stage_model, "medvram_modules"):
+        for module in sd_model.cond_stage_model.medvram_modules():
+            if isinstance(module, ModuleWithParent):
+                parent = module.parent
+                module = module.module
+            else:
+                parent = None
+
+            if module:
+                module.register_forward_pre_hook(send_me_to_gpu)
+
+                if parent:
+                    parents[module] = parent
+
+    elif is_sdxl:
         sd_model.conditioner.register_forward_pre_hook(send_me_to_gpu)
     elif is_sd2:
         sd_model.cond_stage_model.model.register_forward_pre_hook(send_me_to_gpu)
@@ -117,9 +135,9 @@ def setup_for_low_vram(sd_model, use_medvram):
     sd_model.first_stage_model.register_forward_pre_hook(send_me_to_gpu)
     sd_model.first_stage_model.encode = first_stage_model_encode_wrap
     sd_model.first_stage_model.decode = first_stage_model_decode_wrap
-    if sd_model.depth_model:
+    if getattr(sd_model, 'depth_model', None) is not None:
         sd_model.depth_model.register_forward_pre_hook(send_me_to_gpu)
-    if sd_model.embedder:
+    if getattr(sd_model, 'embedder', None) is not None:
         sd_model.embedder.register_forward_pre_hook(send_me_to_gpu)
 
     if use_medvram:

modules/models/sd3/mmdit.py

@@ -0,0 +1,619 @@
+### This file contains impls for MM-DiT, the core model component of SD3
+
+import math
+from typing import Dict, Optional
+import numpy as np
+import torch
+import torch.nn as nn
+from einops import rearrange, repeat
+from modules.models.sd3.other_impls import attention, Mlp
+
+
+class PatchEmbed(nn.Module):
+    """ 2D Image to Patch Embedding"""
+    def __init__(
+            self,
+            img_size: Optional[int] = 224,
+            patch_size: int = 16,
+            in_chans: int = 3,
+            embed_dim: int = 768,
+            flatten: bool = True,
+            bias: bool = True,
+            strict_img_size: bool = True,
+            dynamic_img_pad: bool = False,
+            dtype=None,
+            device=None,
+    ):
+        super().__init__()
+        self.patch_size = (patch_size, patch_size)
+        if img_size is not None:
+            self.img_size = (img_size, img_size)
+            self.grid_size = tuple([s // p for s, p in zip(self.img_size, self.patch_size)])
+            self.num_patches = self.grid_size[0] * self.grid_size[1]
+        else:
+            self.img_size = None
+            self.grid_size = None
+            self.num_patches = None
+
+        # flatten spatial dim and transpose to channels last, kept for bwd compat
+        self.flatten = flatten
+        self.strict_img_size = strict_img_size
+        self.dynamic_img_pad = dynamic_img_pad
+
+        self.proj = nn.Conv2d(in_chans, embed_dim, kernel_size=patch_size, stride=patch_size, bias=bias, dtype=dtype, device=device)
+
+    def forward(self, x):
+        B, C, H, W = x.shape
+        x = self.proj(x)
+        if self.flatten:
+            x = x.flatten(2).transpose(1, 2)  # NCHW -> NLC
+        return x
+
+
+def modulate(x, shift, scale):
+    if shift is None:
+        shift = torch.zeros_like(scale)
+    return x * (1 + scale.unsqueeze(1)) + shift.unsqueeze(1)
+
+
+#################################################################################
+#                   Sine/Cosine Positional Embedding Functions                  #
+#################################################################################
+
+
+def get_2d_sincos_pos_embed(embed_dim, grid_size, cls_token=False, extra_tokens=0, scaling_factor=None, offset=None):
+    """
+    grid_size: int of the grid height and width
+    return:
+    pos_embed: [grid_size*grid_size, embed_dim] or [1+grid_size*grid_size, embed_dim] (w/ or w/o cls_token)
+    """
+    grid_h = np.arange(grid_size, dtype=np.float32)
+    grid_w = np.arange(grid_size, dtype=np.float32)
+    grid = np.meshgrid(grid_w, grid_h)  # here w goes first
+    grid = np.stack(grid, axis=0)
+    if scaling_factor is not None:
+        grid = grid / scaling_factor
+    if offset is not None:
+        grid = grid - offset
+    grid = grid.reshape([2, 1, grid_size, grid_size])
+    pos_embed = get_2d_sincos_pos_embed_from_grid(embed_dim, grid)
+    if cls_token and extra_tokens > 0:
+        pos_embed = np.concatenate([np.zeros([extra_tokens, embed_dim]), pos_embed], axis=0)
+    return pos_embed
+
+
+def get_2d_sincos_pos_embed_from_grid(embed_dim, grid):
+    assert embed_dim % 2 == 0
+    # use half of dimensions to encode grid_h
+    emb_h = get_1d_sincos_pos_embed_from_grid(embed_dim // 2, grid[0])  # (H*W, D/2)
+    emb_w = get_1d_sincos_pos_embed_from_grid(embed_dim // 2, grid[1])  # (H*W, D/2)
+    emb = np.concatenate([emb_h, emb_w], axis=1)  # (H*W, D)
+    return emb
+
+
+def get_1d_sincos_pos_embed_from_grid(embed_dim, pos):
+    """
+    embed_dim: output dimension for each position
+    pos: a list of positions to be encoded: size (M,)
+    out: (M, D)
+    """
+    assert embed_dim % 2 == 0
+    omega = np.arange(embed_dim // 2, dtype=np.float64)
+    omega /= embed_dim / 2.0
+    omega = 1.0 / 10000**omega  # (D/2,)
+    pos = pos.reshape(-1)  # (M,)
+    out = np.einsum("m,d->md", pos, omega)  # (M, D/2), outer product
+    emb_sin = np.sin(out)  # (M, D/2)
+    emb_cos = np.cos(out)  # (M, D/2)
+    return np.concatenate([emb_sin, emb_cos], axis=1)  # (M, D)
+
+
+#################################################################################
+#               Embedding Layers for Timesteps and Class Labels                 #
+#################################################################################
+
+
+class TimestepEmbedder(nn.Module):
+    """Embeds scalar timesteps into vector representations."""
+
+    def __init__(self, hidden_size, frequency_embedding_size=256, dtype=None, device=None):
+        super().__init__()
+        self.mlp = nn.Sequential(
+            nn.Linear(frequency_embedding_size, hidden_size, bias=True, dtype=dtype, device=device),
+            nn.SiLU(),
+            nn.Linear(hidden_size, hidden_size, bias=True, dtype=dtype, device=device),
+        )
+        self.frequency_embedding_size = frequency_embedding_size
+
+    @staticmethod
+    def timestep_embedding(t, dim, max_period=10000):
+        """
+        Create sinusoidal timestep embeddings.
+        :param t: a 1-D Tensor of N indices, one per batch element.
+                          These may be fractional.
+        :param dim: the dimension of the output.
+        :param max_period: controls the minimum frequency of the embeddings.
+        :return: an (N, D) Tensor of positional embeddings.
+        """
+        half = dim // 2
+        freqs = torch.exp(
+            -math.log(max_period)
+            * torch.arange(start=0, end=half, dtype=torch.float32)
+            / half
+        ).to(device=t.device)
+        args = t[:, None].float() * freqs[None]
+        embedding = torch.cat([torch.cos(args), torch.sin(args)], dim=-1)
+        if dim % 2:
+            embedding = torch.cat([embedding, torch.zeros_like(embedding[:, :1])], dim=-1)
+        if torch.is_floating_point(t):
+            embedding = embedding.to(dtype=t.dtype)
+        return embedding
+
+    def forward(self, t, dtype, **kwargs):
+        t_freq = self.timestep_embedding(t, self.frequency_embedding_size).to(dtype)
+        t_emb = self.mlp(t_freq)
+        return t_emb
+
+
+class VectorEmbedder(nn.Module):
+    """Embeds a flat vector of dimension input_dim"""
+
+    def __init__(self, input_dim: int, hidden_size: int, dtype=None, device=None):
+        super().__init__()
+        self.mlp = nn.Sequential(
+            nn.Linear(input_dim, hidden_size, bias=True, dtype=dtype, device=device),
+            nn.SiLU(),
+            nn.Linear(hidden_size, hidden_size, bias=True, dtype=dtype, device=device),
+        )
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        return self.mlp(x)
+
+
+#################################################################################
+#                                 Core DiT Model                                #
+#################################################################################
+
+
+def split_qkv(qkv, head_dim):
+    qkv = qkv.reshape(qkv.shape[0], qkv.shape[1], 3, -1, head_dim).movedim(2, 0)
+    return qkv[0], qkv[1], qkv[2]
+
+def optimized_attention(qkv, num_heads):
+    return attention(qkv[0], qkv[1], qkv[2], num_heads)
+
+class SelfAttention(nn.Module):
+    ATTENTION_MODES = ("xformers", "torch", "torch-hb", "math", "debug")
+
+    def __init__(
+        self,
+        dim: int,
+        num_heads: int = 8,
+        qkv_bias: bool = False,
+        qk_scale: Optional[float] = None,
+        attn_mode: str = "xformers",
+        pre_only: bool = False,
+        qk_norm: Optional[str] = None,
+        rmsnorm: bool = False,
+        dtype=None,
+        device=None,
+    ):
+        super().__init__()
+        self.num_heads = num_heads
+        self.head_dim = dim // num_heads
+
+        self.qkv = nn.Linear(dim, dim * 3, bias=qkv_bias, dtype=dtype, device=device)
+        if not pre_only:
+            self.proj = nn.Linear(dim, dim, dtype=dtype, device=device)
+        assert attn_mode in self.ATTENTION_MODES
+        self.attn_mode = attn_mode
+        self.pre_only = pre_only
+
+        if qk_norm == "rms":
+            self.ln_q = RMSNorm(self.head_dim, elementwise_affine=True, eps=1.0e-6, dtype=dtype, device=device)
+            self.ln_k = RMSNorm(self.head_dim, elementwise_affine=True, eps=1.0e-6, dtype=dtype, device=device)
+        elif qk_norm == "ln":
+            self.ln_q = nn.LayerNorm(self.head_dim, elementwise_affine=True, eps=1.0e-6, dtype=dtype, device=device)
+            self.ln_k = nn.LayerNorm(self.head_dim, elementwise_affine=True, eps=1.0e-6, dtype=dtype, device=device)
+        elif qk_norm is None:
+            self.ln_q = nn.Identity()
+            self.ln_k = nn.Identity()
+        else:
+            raise ValueError(qk_norm)
+
+    def pre_attention(self, x: torch.Tensor):
+        B, L, C = x.shape
+        qkv = self.qkv(x)
+        q, k, v = split_qkv(qkv, self.head_dim)
+        q = self.ln_q(q).reshape(q.shape[0], q.shape[1], -1)
+        k = self.ln_k(k).reshape(q.shape[0], q.shape[1], -1)
+        return (q, k, v)
+
+    def post_attention(self, x: torch.Tensor) -> torch.Tensor:
+        assert not self.pre_only
+        x = self.proj(x)
+        return x
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        (q, k, v) = self.pre_attention(x)
+        x = attention(q, k, v, self.num_heads)
+        x = self.post_attention(x)
+        return x
+
+
+class RMSNorm(torch.nn.Module):
+    def __init__(
+        self, dim: int, elementwise_affine: bool = False, eps: float = 1e-6, device=None, dtype=None
+    ):
+        """
+        Initialize the RMSNorm normalization layer.
+        Args:
+            dim (int): The dimension of the input tensor.
+            eps (float, optional): A small value added to the denominator for numerical stability. Default is 1e-6.
+        Attributes:
+            eps (float): A small value added to the denominator for numerical stability.
+            weight (nn.Parameter): Learnable scaling parameter.
+        """
+        super().__init__()
+        self.eps = eps
+        self.learnable_scale = elementwise_affine
+        if self.learnable_scale:
+            self.weight = nn.Parameter(torch.empty(dim, device=device, dtype=dtype))
+        else:
+            self.register_parameter("weight", None)
+
+    def _norm(self, x):
+        """
+        Apply the RMSNorm normalization to the input tensor.
+        Args:
+            x (torch.Tensor): The input tensor.
+        Returns:
+            torch.Tensor: The normalized tensor.
+        """
+        return x * torch.rsqrt(x.pow(2).mean(-1, keepdim=True) + self.eps)
+
+    def forward(self, x):
+        """
+        Forward pass through the RMSNorm layer.
+        Args:
+            x (torch.Tensor): The input tensor.
+        Returns:
+            torch.Tensor: The output tensor after applying RMSNorm.
+        """
+        x = self._norm(x)
+        if self.learnable_scale:
+            return x * self.weight.to(device=x.device, dtype=x.dtype)
+        else:
+            return x
+
+
+class SwiGLUFeedForward(nn.Module):
+    def __init__(
+        self,
+        dim: int,
+        hidden_dim: int,
+        multiple_of: int,
+        ffn_dim_multiplier: Optional[float] = None,
+    ):
+        """
+        Initialize the FeedForward module.
+
+        Args:
+            dim (int): Input dimension.
+            hidden_dim (int): Hidden dimension of the feedforward layer.
+            multiple_of (int): Value to ensure hidden dimension is a multiple of this value.
+            ffn_dim_multiplier (float, optional): Custom multiplier for hidden dimension. Defaults to None.
+
+        Attributes:
+            w1 (ColumnParallelLinear): Linear transformation for the first layer.
+            w2 (RowParallelLinear): Linear transformation for the second layer.
+            w3 (ColumnParallelLinear): Linear transformation for the third layer.
+
+        """
+        super().__init__()
+        hidden_dim = int(2 * hidden_dim / 3)
+        # custom dim factor multiplier
+        if ffn_dim_multiplier is not None:
+            hidden_dim = int(ffn_dim_multiplier * hidden_dim)
+        hidden_dim = multiple_of * ((hidden_dim + multiple_of - 1) // multiple_of)
+
+        self.w1 = nn.Linear(dim, hidden_dim, bias=False)
+        self.w2 = nn.Linear(hidden_dim, dim, bias=False)
+        self.w3 = nn.Linear(dim, hidden_dim, bias=False)
+
+    def forward(self, x):
+        return self.w2(nn.functional.silu(self.w1(x)) * self.w3(x))
+
+
+class DismantledBlock(nn.Module):
+    """A DiT block with gated adaptive layer norm (adaLN) conditioning."""
+
+    ATTENTION_MODES = ("xformers", "torch", "torch-hb", "math", "debug")
+
+    def __init__(
+        self,
+        hidden_size: int,
+        num_heads: int,
+        mlp_ratio: float = 4.0,
+        attn_mode: str = "xformers",
+        qkv_bias: bool = False,
+        pre_only: bool = False,
+        rmsnorm: bool = False,
+        scale_mod_only: bool = False,
+        swiglu: bool = False,
+        qk_norm: Optional[str] = None,
+        dtype=None,
+        device=None,
+        **block_kwargs,
+    ):
+        super().__init__()
+        assert attn_mode in self.ATTENTION_MODES
+        if not rmsnorm:
+            self.norm1 = nn.LayerNorm(hidden_size, elementwise_affine=False, eps=1e-6, dtype=dtype, device=device)
+        else:
+            self.norm1 = RMSNorm(hidden_size, elementwise_affine=False, eps=1e-6)
+        self.attn = SelfAttention(dim=hidden_size, num_heads=num_heads, qkv_bias=qkv_bias, attn_mode=attn_mode, pre_only=pre_only, qk_norm=qk_norm, rmsnorm=rmsnorm, dtype=dtype, device=device)
+        if not pre_only:
+            if not rmsnorm:
+                self.norm2 = nn.LayerNorm(hidden_size, elementwise_affine=False, eps=1e-6, dtype=dtype, device=device)
+            else:
+                self.norm2 = RMSNorm(hidden_size, elementwise_affine=False, eps=1e-6)
+        mlp_hidden_dim = int(hidden_size * mlp_ratio)
+        if not pre_only:
+            if not swiglu:
+                self.mlp = Mlp(in_features=hidden_size, hidden_features=mlp_hidden_dim, act_layer=nn.GELU(approximate="tanh"), dtype=dtype, device=device)
+            else:
+                self.mlp = SwiGLUFeedForward(dim=hidden_size, hidden_dim=mlp_hidden_dim, multiple_of=256)
+        self.scale_mod_only = scale_mod_only
+        if not scale_mod_only:
+            n_mods = 6 if not pre_only else 2
+        else:
+            n_mods = 4 if not pre_only else 1
+        self.adaLN_modulation = nn.Sequential(nn.SiLU(), nn.Linear(hidden_size, n_mods * hidden_size, bias=True, dtype=dtype, device=device))
+        self.pre_only = pre_only
+
+    def pre_attention(self, x: torch.Tensor, c: torch.Tensor):
+        assert x is not None, "pre_attention called with None input"
+        if not self.pre_only:
+            if not self.scale_mod_only:
+                shift_msa, scale_msa, gate_msa, shift_mlp, scale_mlp, gate_mlp = self.adaLN_modulation(c).chunk(6, dim=1)
+            else:
+                shift_msa = None
+                shift_mlp = None
+                scale_msa, gate_msa, scale_mlp, gate_mlp = self.adaLN_modulation(c).chunk(4, dim=1)
+            qkv = self.attn.pre_attention(modulate(self.norm1(x), shift_msa, scale_msa))
+            return qkv, (x, gate_msa, shift_mlp, scale_mlp, gate_mlp)
+        else:
+            if not self.scale_mod_only:
+                shift_msa, scale_msa = self.adaLN_modulation(c).chunk(2, dim=1)
+            else:
+                shift_msa = None
+                scale_msa = self.adaLN_modulation(c)
+            qkv = self.attn.pre_attention(modulate(self.norm1(x), shift_msa, scale_msa))
+            return qkv, None
+
+    def post_attention(self, attn, x, gate_msa, shift_mlp, scale_mlp, gate_mlp):
+        assert not self.pre_only
+        x = x + gate_msa.unsqueeze(1) * self.attn.post_attention(attn)
+        x = x + gate_mlp.unsqueeze(1) * self.mlp(modulate(self.norm2(x), shift_mlp, scale_mlp))
+        return x
+
+    def forward(self, x: torch.Tensor, c: torch.Tensor) -> torch.Tensor:
+        assert not self.pre_only
+        (q, k, v), intermediates = self.pre_attention(x, c)
+        attn = attention(q, k, v, self.attn.num_heads)
+        return self.post_attention(attn, *intermediates)
+
+
+def block_mixing(context, x, context_block, x_block, c):
+    assert context is not None, "block_mixing called with None context"
+    context_qkv, context_intermediates = context_block.pre_attention(context, c)
+
+    x_qkv, x_intermediates = x_block.pre_attention(x, c)
+
+    o = []
+    for t in range(3):
+        o.append(torch.cat((context_qkv[t], x_qkv[t]), dim=1))
+    q, k, v = tuple(o)
+
+    attn = attention(q, k, v, x_block.attn.num_heads)
+    context_attn, x_attn = (attn[:, : context_qkv[0].shape[1]], attn[:, context_qkv[0].shape[1] :])
+
+    if not context_block.pre_only:
+        context = context_block.post_attention(context_attn, *context_intermediates)
+    else:
+        context = None
+    x = x_block.post_attention(x_attn, *x_intermediates)
+    return context, x
+
+
+class JointBlock(nn.Module):
+    """just a small wrapper to serve as a fsdp unit"""
+
+    def __init__(self, *args, **kwargs):
+        super().__init__()
+        pre_only = kwargs.pop("pre_only")
+        qk_norm = kwargs.pop("qk_norm", None)
+        self.context_block = DismantledBlock(*args, pre_only=pre_only, qk_norm=qk_norm, **kwargs)
+        self.x_block = DismantledBlock(*args, pre_only=False, qk_norm=qk_norm, **kwargs)
+
+    def forward(self, *args, **kwargs):
+        return block_mixing(*args, context_block=self.context_block, x_block=self.x_block, **kwargs)
+
+
+class FinalLayer(nn.Module):
+    """
+    The final layer of DiT.
+    """
+
+    def __init__(self, hidden_size: int, patch_size: int, out_channels: int, total_out_channels: Optional[int] = None, dtype=None, device=None):
+        super().__init__()
+        self.norm_final = nn.LayerNorm(hidden_size, elementwise_affine=False, eps=1e-6, dtype=dtype, device=device)
+        self.linear = (
+            nn.Linear(hidden_size, patch_size * patch_size * out_channels, bias=True, dtype=dtype, device=device)
+            if (total_out_channels is None)
+            else nn.Linear(hidden_size, total_out_channels, bias=True, dtype=dtype, device=device)
+        )
+        self.adaLN_modulation = nn.Sequential(nn.SiLU(), nn.Linear(hidden_size, 2 * hidden_size, bias=True, dtype=dtype, device=device))
+
+    def forward(self, x: torch.Tensor, c: torch.Tensor) -> torch.Tensor:
+        shift, scale = self.adaLN_modulation(c).chunk(2, dim=1)
+        x = modulate(self.norm_final(x), shift, scale)
+        x = self.linear(x)
+        return x
+
+
+class MMDiT(nn.Module):
+    """Diffusion model with a Transformer backbone."""
+
+    def __init__(
+        self,
+        input_size: int = 32,
+        patch_size: int = 2,
+        in_channels: int = 4,
+        depth: int = 28,
+        mlp_ratio: float = 4.0,
+        learn_sigma: bool = False,
+        adm_in_channels: Optional[int] = None,
+        context_embedder_config: Optional[Dict] = None,
+        register_length: int = 0,
+        attn_mode: str = "torch",
+        rmsnorm: bool = False,
+        scale_mod_only: bool = False,
+        swiglu: bool = False,
+        out_channels: Optional[int] = None,
+        pos_embed_scaling_factor: Optional[float] = None,
+        pos_embed_offset: Optional[float] = None,
+        pos_embed_max_size: Optional[int] = None,
+        num_patches = None,
+        qk_norm: Optional[str] = None,
+        qkv_bias: bool = True,
+        dtype = None,
+        device = None,
+    ):
+        super().__init__()
+        self.dtype = dtype
+        self.learn_sigma = learn_sigma
+        self.in_channels = in_channels
+        default_out_channels = in_channels * 2 if learn_sigma else in_channels
+        self.out_channels = out_channels if out_channels is not None else default_out_channels
+        self.patch_size = patch_size
+        self.pos_embed_scaling_factor = pos_embed_scaling_factor
+        self.pos_embed_offset = pos_embed_offset
+        self.pos_embed_max_size = pos_embed_max_size
+
+        # apply magic --> this defines a head_size of 64
+        hidden_size = 64 * depth
+        num_heads = depth
+
+        self.num_heads = num_heads
+
+        self.x_embedder = PatchEmbed(input_size, patch_size, in_channels, hidden_size, bias=True, strict_img_size=self.pos_embed_max_size is None, dtype=dtype, device=device)
+        self.t_embedder = TimestepEmbedder(hidden_size, dtype=dtype, device=device)
+
+        if adm_in_channels is not None:
+            assert isinstance(adm_in_channels, int)
+            self.y_embedder = VectorEmbedder(adm_in_channels, hidden_size, dtype=dtype, device=device)
+
+        self.context_embedder = nn.Identity()
+        if context_embedder_config is not None:
+            if context_embedder_config["target"] == "torch.nn.Linear":
+                self.context_embedder = nn.Linear(**context_embedder_config["params"], dtype=dtype, device=device)
+
+        self.register_length = register_length
+        if self.register_length > 0:
+            self.register = nn.Parameter(torch.randn(1, register_length, hidden_size, dtype=dtype, device=device))
+
+        # num_patches = self.x_embedder.num_patches
+        # Will use fixed sin-cos embedding:
+        # just use a buffer already
+        if num_patches is not None:
+            self.register_buffer(
+                "pos_embed",
+                torch.zeros(1, num_patches, hidden_size, dtype=dtype, device=device),
+            )
+        else:
+            self.pos_embed = None
+
+        self.joint_blocks = nn.ModuleList(
+            [
+                JointBlock(hidden_size, num_heads, mlp_ratio=mlp_ratio, qkv_bias=qkv_bias, attn_mode=attn_mode, pre_only=i == depth - 1, rmsnorm=rmsnorm, scale_mod_only=scale_mod_only, swiglu=swiglu, qk_norm=qk_norm, dtype=dtype, device=device)
+                for i in range(depth)
+            ]
+        )
+
+        self.final_layer = FinalLayer(hidden_size, patch_size, self.out_channels, dtype=dtype, device=device)
+
+    def cropped_pos_embed(self, hw):
+        assert self.pos_embed_max_size is not None
+        p = self.x_embedder.patch_size[0]
+        h, w = hw
+        # patched size
+        h = h // p
+        w = w // p
+        assert h <= self.pos_embed_max_size, (h, self.pos_embed_max_size)
+        assert w <= self.pos_embed_max_size, (w, self.pos_embed_max_size)
+        top = (self.pos_embed_max_size - h) // 2
+        left = (self.pos_embed_max_size - w) // 2
+        spatial_pos_embed = rearrange(
+            self.pos_embed,
+            "1 (h w) c -> 1 h w c",
+            h=self.pos_embed_max_size,
+            w=self.pos_embed_max_size,
+        )
+        spatial_pos_embed = spatial_pos_embed[:, top : top + h, left : left + w, :]
+        spatial_pos_embed = rearrange(spatial_pos_embed, "1 h w c -> 1 (h w) c")
+        return spatial_pos_embed
+
+    def unpatchify(self, x, hw=None):
+        """
+        x: (N, T, patch_size**2 * C)
+        imgs: (N, H, W, C)
+        """
+        c = self.out_channels
+        p = self.x_embedder.patch_size[0]
+        if hw is None:
+            h = w = int(x.shape[1] ** 0.5)
+        else:
+            h, w = hw
+            h = h // p
+            w = w // p
+        assert h * w == x.shape[1]
+
+        x = x.reshape(shape=(x.shape[0], h, w, p, p, c))
+        x = torch.einsum("nhwpqc->nchpwq", x)
+        imgs = x.reshape(shape=(x.shape[0], c, h * p, w * p))
+        return imgs
+
+    def forward_core_with_concat(self, x: torch.Tensor, c_mod: torch.Tensor, context: Optional[torch.Tensor] = None) -> torch.Tensor:
+        if self.register_length > 0:
+            context = torch.cat((repeat(self.register, "1 ... -> b ...", b=x.shape[0]), context if context is not None else torch.Tensor([]).type_as(x)), 1)
+
+        # context is B, L', D
+        # x is B, L, D
+        for block in self.joint_blocks:
+            context, x = block(context, x, c=c_mod)
+
+        x = self.final_layer(x, c_mod)  # (N, T, patch_size ** 2 * out_channels)
+        return x
+
+    def forward(self, x: torch.Tensor, t: torch.Tensor, y: Optional[torch.Tensor] = None, context: Optional[torch.Tensor] = None) -> torch.Tensor:
+        """
+        Forward pass of DiT.
+        x: (N, C, H, W) tensor of spatial inputs (images or latent representations of images)
+        t: (N,) tensor of diffusion timesteps
+        y: (N,) tensor of class labels
+        """
+        hw = x.shape[-2:]
+        x = self.x_embedder(x) + self.cropped_pos_embed(hw)
+        c = self.t_embedder(t, dtype=x.dtype)  # (N, D)
+        if y is not None:
+            y = self.y_embedder(y)  # (N, D)
+            c = c + y  # (N, D)
+
+        context = self.context_embedder(context)
+
+        x = self.forward_core_with_concat(x, c, context)
+
+        x = self.unpatchify(x, hw=hw)  # (N, out_channels, H, W)
+        return x

modules/models/sd3/other_impls.py

@@ -0,0 +1,508 @@
+### This file contains impls for underlying related models (CLIP, T5, etc)
+
+import torch
+import math
+from torch import nn
+from transformers import CLIPTokenizer, T5TokenizerFast
+
+
+#################################################################################################
+### Core/Utility
+#################################################################################################
+
+
+class AutocastLinear(nn.Linear):
+    """Same as usual linear layer, but casts its weights to whatever the parameter type is.
+
+    This is different from torch.autocast in a way that float16 layer processing float32 input
+    will return float16 with autocast on, and float32 with this. T5 seems to be fucked
+    if you do it in full float16 (returning almost all zeros in the final output).
+    """
+
+    def forward(self, x):
+        return torch.nn.functional.linear(x, self.weight.to(x.dtype), self.bias.to(x.dtype) if self.bias is not None else None)
+
+
+def attention(q, k, v, heads, mask=None):
+    """Convenience wrapper around a basic attention operation"""
+    b, _, dim_head = q.shape
+    dim_head //= heads
+    q, k, v = [t.view(b, -1, heads, dim_head).transpose(1, 2) for t in (q, k, v)]
+    out = torch.nn.functional.scaled_dot_product_attention(q, k, v, attn_mask=mask, dropout_p=0.0, is_causal=False)
+    return out.transpose(1, 2).reshape(b, -1, heads * dim_head)
+
+
+class Mlp(nn.Module):
+    """ MLP as used in Vision Transformer, MLP-Mixer and related networks"""
+    def __init__(self, in_features, hidden_features=None, out_features=None, act_layer=nn.GELU, bias=True, dtype=None, device=None):
+        super().__init__()
+        out_features = out_features or in_features
+        hidden_features = hidden_features or in_features
+
+        self.fc1 = nn.Linear(in_features, hidden_features, bias=bias, dtype=dtype, device=device)
+        self.act = act_layer
+        self.fc2 = nn.Linear(hidden_features, out_features, bias=bias, dtype=dtype, device=device)
+
+    def forward(self, x):
+        x = self.fc1(x)
+        x = self.act(x)
+        x = self.fc2(x)
+        return x
+
+
+#################################################################################################
+### CLIP
+#################################################################################################
+
+
+class CLIPAttention(torch.nn.Module):
+    def __init__(self, embed_dim, heads, dtype, device):
+        super().__init__()
+        self.heads = heads
+        self.q_proj = nn.Linear(embed_dim, embed_dim, bias=True, dtype=dtype, device=device)
+        self.k_proj = nn.Linear(embed_dim, embed_dim, bias=True, dtype=dtype, device=device)
+        self.v_proj = nn.Linear(embed_dim, embed_dim, bias=True, dtype=dtype, device=device)
+        self.out_proj = nn.Linear(embed_dim, embed_dim, bias=True, dtype=dtype, device=device)
+
+    def forward(self, x, mask=None):
+        q = self.q_proj(x)
+        k = self.k_proj(x)
+        v = self.v_proj(x)
+        out = attention(q, k, v, self.heads, mask)
+        return self.out_proj(out)
+
+
+ACTIVATIONS = {
+    "quick_gelu": lambda a: a * torch.sigmoid(1.702 * a),
+    "gelu": torch.nn.functional.gelu,
+}
+
+class CLIPLayer(torch.nn.Module):
+    def __init__(self, embed_dim, heads, intermediate_size, intermediate_activation, dtype, device):
+        super().__init__()
+        self.layer_norm1 = nn.LayerNorm(embed_dim, dtype=dtype, device=device)
+        self.self_attn = CLIPAttention(embed_dim, heads, dtype, device)
+        self.layer_norm2 = nn.LayerNorm(embed_dim, dtype=dtype, device=device)
+        #self.mlp = CLIPMLP(embed_dim, intermediate_size, intermediate_activation, dtype, device)
+        self.mlp = Mlp(embed_dim, intermediate_size, embed_dim, act_layer=ACTIVATIONS[intermediate_activation], dtype=dtype, device=device)
+
+    def forward(self, x, mask=None):
+        x += self.self_attn(self.layer_norm1(x), mask)
+        x += self.mlp(self.layer_norm2(x))
+        return x
+
+
+class CLIPEncoder(torch.nn.Module):
+    def __init__(self, num_layers, embed_dim, heads, intermediate_size, intermediate_activation, dtype, device):
+        super().__init__()
+        self.layers = torch.nn.ModuleList([CLIPLayer(embed_dim, heads, intermediate_size, intermediate_activation, dtype, device) for i in range(num_layers)])
+
+    def forward(self, x, mask=None, intermediate_output=None):
+        if intermediate_output is not None:
+            if intermediate_output < 0:
+                intermediate_output = len(self.layers) + intermediate_output
+        intermediate = None
+        for i, layer in enumerate(self.layers):
+            x = layer(x, mask)
+            if i == intermediate_output:
+                intermediate = x.clone()
+        return x, intermediate
+
+
+class CLIPEmbeddings(torch.nn.Module):
+    def __init__(self, embed_dim, vocab_size=49408, num_positions=77, dtype=None, device=None):
+        super().__init__()
+        self.token_embedding = torch.nn.Embedding(vocab_size, embed_dim, dtype=dtype, device=device)
+        self.position_embedding = torch.nn.Embedding(num_positions, embed_dim, dtype=dtype, device=device)
+
+    def forward(self, input_tokens):
+        return self.token_embedding(input_tokens) + self.position_embedding.weight
+
+
+class CLIPTextModel_(torch.nn.Module):
+    def __init__(self, config_dict, dtype, device):
+        num_layers = config_dict["num_hidden_layers"]
+        embed_dim = config_dict["hidden_size"]
+        heads = config_dict["num_attention_heads"]
+        intermediate_size = config_dict["intermediate_size"]
+        intermediate_activation = config_dict["hidden_act"]
+        super().__init__()
+        self.embeddings = CLIPEmbeddings(embed_dim, dtype=torch.float32, device=device)
+        self.encoder = CLIPEncoder(num_layers, embed_dim, heads, intermediate_size, intermediate_activation, dtype, device)
+        self.final_layer_norm = nn.LayerNorm(embed_dim, dtype=dtype, device=device)
+
+    def forward(self, input_tokens, intermediate_output=None, final_layer_norm_intermediate=True):
+        x = self.embeddings(input_tokens)
+        causal_mask = torch.empty(x.shape[1], x.shape[1], dtype=x.dtype, device=x.device).fill_(float("-inf")).triu_(1)
+        x, i = self.encoder(x, mask=causal_mask, intermediate_output=intermediate_output)
+        x = self.final_layer_norm(x)
+        if i is not None and final_layer_norm_intermediate:
+            i = self.final_layer_norm(i)
+        pooled_output = x[torch.arange(x.shape[0], device=x.device), input_tokens.to(dtype=torch.int, device=x.device).argmax(dim=-1),]
+        return x, i, pooled_output
+
+
+class CLIPTextModel(torch.nn.Module):
+    def __init__(self, config_dict, dtype, device):
+        super().__init__()
+        self.num_layers = config_dict["num_hidden_layers"]
+        self.text_model = CLIPTextModel_(config_dict, dtype, device)
+        embed_dim = config_dict["hidden_size"]
+        self.text_projection = nn.Linear(embed_dim, embed_dim, bias=False, dtype=dtype, device=device)
+        self.text_projection.weight.copy_(torch.eye(embed_dim))
+        self.dtype = dtype
+
+    def get_input_embeddings(self):
+        return self.text_model.embeddings.token_embedding
+
+    def set_input_embeddings(self, embeddings):
+        self.text_model.embeddings.token_embedding = embeddings
+
+    def forward(self, *args, **kwargs):
+        x = self.text_model(*args, **kwargs)
+        out = self.text_projection(x[2])
+        return (x[0], x[1], out, x[2])
+
+
+class SDTokenizer:
+    def __init__(self, max_length=77, pad_with_end=True, tokenizer=None, has_start_token=True, pad_to_max_length=True, min_length=None):
+        self.tokenizer = tokenizer
+        self.max_length = max_length
+        self.min_length = min_length
+        empty = self.tokenizer('')["input_ids"]
+        if has_start_token:
+            self.tokens_start = 1
+            self.start_token = empty[0]
+            self.end_token = empty[1]
+        else:
+            self.tokens_start = 0
+            self.start_token = None
+            self.end_token = empty[0]
+        self.pad_with_end = pad_with_end
+        self.pad_to_max_length = pad_to_max_length
+        vocab = self.tokenizer.get_vocab()
+        self.inv_vocab = {v: k for k, v in vocab.items()}
+        self.max_word_length = 8
+
+
+    def tokenize_with_weights(self, text:str):
+        """Tokenize the text, with weight values - presume 1.0 for all and ignore other features here. The details aren't relevant for a reference impl, and weights themselves has weak effect on SD3."""
+        if self.pad_with_end:
+            pad_token = self.end_token
+        else:
+            pad_token = 0
+        batch = []
+        if self.start_token is not None:
+            batch.append((self.start_token, 1.0))
+        to_tokenize = text.replace("\n", " ").split(' ')
+        to_tokenize = [x for x in to_tokenize if x != ""]
+        for word in to_tokenize:
+            batch.extend([(t, 1) for t in self.tokenizer(word)["input_ids"][self.tokens_start:-1]])
+        batch.append((self.end_token, 1.0))
+        if self.pad_to_max_length:
+            batch.extend([(pad_token, 1.0)] * (self.max_length - len(batch)))
+        if self.min_length is not None and len(batch) < self.min_length:
+            batch.extend([(pad_token, 1.0)] * (self.min_length - len(batch)))
+        return [batch]
+
+
+class SDXLClipGTokenizer(SDTokenizer):
+    def __init__(self, tokenizer):
+        super().__init__(pad_with_end=False, tokenizer=tokenizer)
+
+
+class SD3Tokenizer:
+    def __init__(self):
+        clip_tokenizer = CLIPTokenizer.from_pretrained("openai/clip-vit-large-patch14")
+        self.clip_l = SDTokenizer(tokenizer=clip_tokenizer)
+        self.clip_g = SDXLClipGTokenizer(clip_tokenizer)
+        self.t5xxl = T5XXLTokenizer()
+
+    def tokenize_with_weights(self, text:str):
+        out = {}
+        out["g"] = self.clip_g.tokenize_with_weights(text)
+        out["l"] = self.clip_l.tokenize_with_weights(text)
+        out["t5xxl"] = self.t5xxl.tokenize_with_weights(text)
+        return out
+
+
+class ClipTokenWeightEncoder:
+    def encode_token_weights(self, token_weight_pairs):
+        tokens = [a[0] for a in token_weight_pairs[0]]
+        out, pooled = self([tokens])
+        if pooled is not None:
+            first_pooled = pooled[0:1].cpu()
+        else:
+            first_pooled = pooled
+        output = [out[0:1]]
+        return torch.cat(output, dim=-2).cpu(), first_pooled
+
+
+class SDClipModel(torch.nn.Module, ClipTokenWeightEncoder):
+    """Uses the CLIP transformer encoder for text (from huggingface)"""
+    LAYERS = ["last", "pooled", "hidden"]
+    def __init__(self, device="cpu", max_length=77, layer="last", layer_idx=None, textmodel_json_config=None, dtype=None, model_class=CLIPTextModel,
+                 special_tokens=None, layer_norm_hidden_state=True, return_projected_pooled=True):
+        super().__init__()
+        assert layer in self.LAYERS
+        self.transformer = model_class(textmodel_json_config, dtype, device)
+        self.num_layers = self.transformer.num_layers
+        self.max_length = max_length
+        self.transformer = self.transformer.eval()
+        for param in self.parameters():
+            param.requires_grad = False
+        self.layer = layer
+        self.layer_idx = None
+        self.special_tokens = special_tokens if special_tokens is not None else {"start": 49406, "end": 49407, "pad": 49407}
+        self.logit_scale = torch.nn.Parameter(torch.tensor(4.6055))
+        self.layer_norm_hidden_state = layer_norm_hidden_state
+        self.return_projected_pooled = return_projected_pooled
+        if layer == "hidden":
+            assert layer_idx is not None
+            assert abs(layer_idx) < self.num_layers
+            self.set_clip_options({"layer": layer_idx})
+        self.options_default = (self.layer, self.layer_idx, self.return_projected_pooled)
+
+    def set_clip_options(self, options):
+        layer_idx = options.get("layer", self.layer_idx)
+        self.return_projected_pooled = options.get("projected_pooled", self.return_projected_pooled)
+        if layer_idx is None or abs(layer_idx) > self.num_layers:
+            self.layer = "last"
+        else:
+            self.layer = "hidden"
+            self.layer_idx = layer_idx
+
+    def forward(self, tokens):
+        backup_embeds = self.transformer.get_input_embeddings()
+        tokens = torch.asarray(tokens, dtype=torch.int64, device=backup_embeds.weight.device)
+        outputs = self.transformer(tokens, intermediate_output=self.layer_idx, final_layer_norm_intermediate=self.layer_norm_hidden_state)
+        self.transformer.set_input_embeddings(backup_embeds)
+        if self.layer == "last":
+            z = outputs[0]
+        else:
+            z = outputs[1]
+        pooled_output = None
+        if len(outputs) >= 3:
+            if not self.return_projected_pooled and len(outputs) >= 4 and outputs[3] is not None:
+                pooled_output = outputs[3].float()
+            elif outputs[2] is not None:
+                pooled_output = outputs[2].float()
+        return z.float(), pooled_output
+
+
+class SDXLClipG(SDClipModel):
+    """Wraps the CLIP-G model into the SD-CLIP-Model interface"""
+    def __init__(self, config, device="cpu", layer="penultimate", layer_idx=None, dtype=None):
+        if layer == "penultimate":
+            layer="hidden"
+            layer_idx=-2
+        super().__init__(device=device, layer=layer, layer_idx=layer_idx, textmodel_json_config=config, dtype=dtype, special_tokens={"start": 49406, "end": 49407, "pad": 0}, layer_norm_hidden_state=False)
+
+
+class T5XXLModel(SDClipModel):
+    """Wraps the T5-XXL model into the SD-CLIP-Model interface for convenience"""
+    def __init__(self, config, device="cpu", layer="last", layer_idx=None, dtype=None):
+        super().__init__(device=device, layer=layer, layer_idx=layer_idx, textmodel_json_config=config, dtype=dtype, special_tokens={"end": 1, "pad": 0}, model_class=T5)
+
+
+#################################################################################################
+### T5 implementation, for the T5-XXL text encoder portion, largely pulled from upstream impl
+#################################################################################################
+
+class T5XXLTokenizer(SDTokenizer):
+    """Wraps the T5 Tokenizer from HF into the SDTokenizer interface"""
+    def __init__(self):
+        super().__init__(pad_with_end=False, tokenizer=T5TokenizerFast.from_pretrained("google/t5-v1_1-xxl"), has_start_token=False, pad_to_max_length=False, max_length=99999999, min_length=77)
+
+
+class T5LayerNorm(torch.nn.Module):
+    def __init__(self, hidden_size, eps=1e-6, dtype=None, device=None):
+        super().__init__()
+        self.weight = torch.nn.Parameter(torch.ones(hidden_size, dtype=dtype, device=device))
+        self.variance_epsilon = eps
+
+    def forward(self, x):
+        variance = x.pow(2).mean(-1, keepdim=True)
+        x = x * torch.rsqrt(variance + self.variance_epsilon)
+        return self.weight.to(device=x.device, dtype=x.dtype) * x
+
+
+class T5DenseGatedActDense(torch.nn.Module):
+    def __init__(self, model_dim, ff_dim, dtype, device):
+        super().__init__()
+        self.wi_0 = AutocastLinear(model_dim, ff_dim, bias=False, dtype=dtype, device=device)
+        self.wi_1 = AutocastLinear(model_dim, ff_dim, bias=False, dtype=dtype, device=device)
+        self.wo = AutocastLinear(ff_dim, model_dim, bias=False, dtype=dtype, device=device)
+
+    def forward(self, x):
+        hidden_gelu = torch.nn.functional.gelu(self.wi_0(x), approximate="tanh")
+        hidden_linear = self.wi_1(x)
+        x = hidden_gelu * hidden_linear
+        x = self.wo(x)
+        return x
+
+
+class T5LayerFF(torch.nn.Module):
+    def __init__(self, model_dim, ff_dim, dtype, device):
+        super().__init__()
+        self.DenseReluDense = T5DenseGatedActDense(model_dim, ff_dim, dtype, device)
+        self.layer_norm = T5LayerNorm(model_dim, dtype=dtype, device=device)
+
+    def forward(self, x):
+        forwarded_states = self.layer_norm(x)
+        forwarded_states = self.DenseReluDense(forwarded_states)
+        x += forwarded_states
+        return x
+
+
+class T5Attention(torch.nn.Module):
+    def __init__(self, model_dim, inner_dim, num_heads, relative_attention_bias, dtype, device):
+        super().__init__()
+        # Mesh TensorFlow initialization to avoid scaling before softmax
+        self.q = AutocastLinear(model_dim, inner_dim, bias=False, dtype=dtype, device=device)
+        self.k = AutocastLinear(model_dim, inner_dim, bias=False, dtype=dtype, device=device)
+        self.v = AutocastLinear(model_dim, inner_dim, bias=False, dtype=dtype, device=device)
+        self.o = AutocastLinear(inner_dim, model_dim, bias=False, dtype=dtype, device=device)
+        self.num_heads = num_heads
+        self.relative_attention_bias = None
+        if relative_attention_bias:
+            self.relative_attention_num_buckets = 32
+            self.relative_attention_max_distance = 128
+            self.relative_attention_bias = torch.nn.Embedding(self.relative_attention_num_buckets, self.num_heads, device=device)
+
+    @staticmethod
+    def _relative_position_bucket(relative_position, bidirectional=True, num_buckets=32, max_distance=128):
+        """
+        Adapted from Mesh Tensorflow:
+        https://github.com/tensorflow/mesh/blob/0cb87fe07da627bf0b7e60475d59f95ed6b5be3d/mesh_tensorflow/transformer/transformer_layers.py#L593
+
+        Translate relative position to a bucket number for relative attention. The relative position is defined as
+        memory_position - query_position, i.e. the distance in tokens from the attending position to the attended-to
+        position. If bidirectional=False, then positive relative positions are invalid. We use smaller buckets for
+        small absolute relative_position and larger buckets for larger absolute relative_positions. All relative
+        positions >=max_distance map to the same bucket. All relative positions <=-max_distance map to the same bucket.
+        This should allow for more graceful generalization to longer sequences than the model has been trained on
+
+        Args:
+            relative_position: an int32 Tensor
+            bidirectional: a boolean - whether the attention is bidirectional
+            num_buckets: an integer
+            max_distance: an integer
+
+        Returns:
+            a Tensor with the same shape as relative_position, containing int32 values in the range [0, num_buckets)
+        """
+        relative_buckets = 0
+        if bidirectional:
+            num_buckets //= 2
+            relative_buckets += (relative_position > 0).to(torch.long) * num_buckets
+            relative_position = torch.abs(relative_position)
+        else:
+            relative_position = -torch.min(relative_position, torch.zeros_like(relative_position))
+        # now relative_position is in the range [0, inf)
+        # half of the buckets are for exact increments in positions
+        max_exact = num_buckets // 2
+        is_small = relative_position < max_exact
+        # The other half of the buckets are for logarithmically bigger bins in positions up to max_distance
+        relative_position_if_large = max_exact + (
+            torch.log(relative_position.float() / max_exact)
+            / math.log(max_distance / max_exact)
+            * (num_buckets - max_exact)
+        ).to(torch.long)
+        relative_position_if_large = torch.min(relative_position_if_large, torch.full_like(relative_position_if_large, num_buckets - 1))
+        relative_buckets += torch.where(is_small, relative_position, relative_position_if_large)
+        return relative_buckets
+
+    def compute_bias(self, query_length, key_length, device):
+        """Compute binned relative position bias"""
+        context_position = torch.arange(query_length, dtype=torch.long, device=device)[:, None]
+        memory_position = torch.arange(key_length, dtype=torch.long, device=device)[None, :]
+        relative_position = memory_position - context_position  # shape (query_length, key_length)
+        relative_position_bucket = self._relative_position_bucket(
+            relative_position,  # shape (query_length, key_length)
+            bidirectional=True,
+            num_buckets=self.relative_attention_num_buckets,
+            max_distance=self.relative_attention_max_distance,
+        )
+        values = self.relative_attention_bias(relative_position_bucket)  # shape (query_length, key_length, num_heads)
+        values = values.permute([2, 0, 1]).unsqueeze(0)  # shape (1, num_heads, query_length, key_length)
+        return values
+
+    def forward(self, x, past_bias=None):
+        q = self.q(x)
+        k = self.k(x)
+        v = self.v(x)
+
+        if self.relative_attention_bias is not None:
+            past_bias = self.compute_bias(x.shape[1], x.shape[1], x.device)
+        if past_bias is not None:
+            mask = past_bias
+        else:
+            mask = None
+
+        out = attention(q, k * ((k.shape[-1] / self.num_heads) ** 0.5), v, self.num_heads, mask.to(x.dtype) if mask is not None else None)
+
+        return self.o(out), past_bias
+
+
+class T5LayerSelfAttention(torch.nn.Module):
+    def __init__(self, model_dim, inner_dim, ff_dim, num_heads, relative_attention_bias, dtype, device):
+        super().__init__()
+        self.SelfAttention = T5Attention(model_dim, inner_dim, num_heads, relative_attention_bias, dtype, device)
+        self.layer_norm = T5LayerNorm(model_dim, dtype=dtype, device=device)
+
+    def forward(self, x, past_bias=None):
+        output, past_bias = self.SelfAttention(self.layer_norm(x), past_bias=past_bias)
+        x += output
+        return x, past_bias
+
+
+class T5Block(torch.nn.Module):
+    def __init__(self, model_dim, inner_dim, ff_dim, num_heads, relative_attention_bias, dtype, device):
+        super().__init__()
+        self.layer = torch.nn.ModuleList()
+        self.layer.append(T5LayerSelfAttention(model_dim, inner_dim, ff_dim, num_heads, relative_attention_bias, dtype, device))
+        self.layer.append(T5LayerFF(model_dim, ff_dim, dtype, device))
+
+    def forward(self, x, past_bias=None):
+        x, past_bias = self.layer[0](x, past_bias)
+        x = self.layer[-1](x)
+        return x, past_bias
+
+
+class T5Stack(torch.nn.Module):
+    def __init__(self, num_layers, model_dim, inner_dim, ff_dim, num_heads, vocab_size, dtype, device):
+        super().__init__()
+        self.embed_tokens = torch.nn.Embedding(vocab_size, model_dim, device=device)
+        self.block = torch.nn.ModuleList([T5Block(model_dim, inner_dim, ff_dim, num_heads, relative_attention_bias=(i == 0), dtype=dtype, device=device) for i in range(num_layers)])
+        self.final_layer_norm = T5LayerNorm(model_dim, dtype=dtype, device=device)
+
+    def forward(self, input_ids, intermediate_output=None, final_layer_norm_intermediate=True):
+        intermediate = None
+        x = self.embed_tokens(input_ids).to(torch.float32)  # needs float32 or else T5 returns all zeroes
+        past_bias = None
+        for i, layer in enumerate(self.block):
+            x, past_bias = layer(x, past_bias)
+            if i == intermediate_output:
+                intermediate = x.clone()
+        x = self.final_layer_norm(x)
+        if intermediate is not None and final_layer_norm_intermediate:
+            intermediate = self.final_layer_norm(intermediate)
+        return x, intermediate
+
+
+class T5(torch.nn.Module):
+    def __init__(self, config_dict, dtype, device):
+        super().__init__()
+        self.num_layers = config_dict["num_layers"]
+        self.encoder = T5Stack(self.num_layers, config_dict["d_model"], config_dict["d_model"], config_dict["d_ff"], config_dict["num_heads"], config_dict["vocab_size"], dtype, device)
+        self.dtype = dtype
+
+    def get_input_embeddings(self):
+        return self.encoder.embed_tokens
+
+    def set_input_embeddings(self, embeddings):
+        self.encoder.embed_tokens = embeddings
+
+    def forward(self, *args, **kwargs):
+        return self.encoder(*args, **kwargs)

modules/models/sd3/sd3_cond.py

@@ -0,0 +1,218 @@
+import os
+import safetensors
+import torch
+import typing
+
+from transformers import CLIPTokenizer, T5TokenizerFast
+
+from modules import shared, devices, modelloader, sd_hijack_clip, prompt_parser
+from modules.models.sd3.other_impls import SDClipModel, SDXLClipG, T5XXLModel, SD3Tokenizer
+
+
+class SafetensorsMapping(typing.Mapping):
+    def __init__(self, file):
+        self.file = file
+
+    def __len__(self):
+        return len(self.file.keys())
+
+    def __iter__(self):
+        for key in self.file.keys():
+            yield key
+
+    def __getitem__(self, key):
+        return self.file.get_tensor(key)
+
+
+CLIPL_URL = "https://huggingface.co/AUTOMATIC/stable-diffusion-3-medium-text-encoders/resolve/main/clip_l.safetensors"
+CLIPL_CONFIG = {
+    "hidden_act": "quick_gelu",
+    "hidden_size": 768,
+    "intermediate_size": 3072,
+    "num_attention_heads": 12,
+    "num_hidden_layers": 12,
+}
+
+CLIPG_URL = "https://huggingface.co/AUTOMATIC/stable-diffusion-3-medium-text-encoders/resolve/main/clip_g.safetensors"
+CLIPG_CONFIG = {
+    "hidden_act": "gelu",
+    "hidden_size": 1280,
+    "intermediate_size": 5120,
+    "num_attention_heads": 20,
+    "num_hidden_layers": 32,
+}
+
+T5_URL = "https://huggingface.co/AUTOMATIC/stable-diffusion-3-medium-text-encoders/resolve/main/t5xxl_fp16.safetensors"
+T5_CONFIG = {
+    "d_ff": 10240,
+    "d_model": 4096,
+    "num_heads": 64,
+    "num_layers": 24,
+    "vocab_size": 32128,
+}
+
+
+class Sd3ClipLG(sd_hijack_clip.TextConditionalModel):
+    def __init__(self, clip_l, clip_g):
+        super().__init__()
+
+        self.clip_l = clip_l
+        self.clip_g = clip_g
+
+        self.tokenizer = CLIPTokenizer.from_pretrained("openai/clip-vit-large-patch14")
+
+        empty = self.tokenizer('')["input_ids"]
+        self.id_start = empty[0]
+        self.id_end = empty[1]
+        self.id_pad = empty[1]
+
+        self.return_pooled = True
+
+    def tokenize(self, texts):
+        return self.tokenizer(texts, truncation=False, add_special_tokens=False)["input_ids"]
+
+    def encode_with_transformers(self, tokens):
+        tokens_g = tokens.clone()
+
+        for batch_pos in range(tokens_g.shape[0]):
+            index = tokens_g[batch_pos].cpu().tolist().index(self.id_end)
+            tokens_g[batch_pos, index+1:tokens_g.shape[1]] = 0
+
+        l_out, l_pooled = self.clip_l(tokens)
+        g_out, g_pooled = self.clip_g(tokens_g)
+
+        lg_out = torch.cat([l_out, g_out], dim=-1)
+        lg_out = torch.nn.functional.pad(lg_out, (0, 4096 - lg_out.shape[-1]))
+
+        vector_out = torch.cat((l_pooled, g_pooled), dim=-1)
+
+        lg_out.pooled = vector_out
+        return lg_out
+
+    def encode_embedding_init_text(self, init_text, nvpt):
+        return torch.zeros((nvpt, 768+1280), device=devices.device) # XXX
+
+
+class Sd3T5(torch.nn.Module):
+    def __init__(self, t5xxl):
+        super().__init__()
+
+        self.t5xxl = t5xxl
+        self.tokenizer = T5TokenizerFast.from_pretrained("google/t5-v1_1-xxl")
+
+        empty = self.tokenizer('', padding='max_length', max_length=2)["input_ids"]
+        self.id_end = empty[0]
+        self.id_pad = empty[1]
+
+    def tokenize(self, texts):
+        return self.tokenizer(texts, truncation=False, add_special_tokens=False)["input_ids"]
+
+    def tokenize_line(self, line, *, target_token_count=None):
+        if shared.opts.emphasis != "None":
+            parsed = prompt_parser.parse_prompt_attention(line)
+        else:
+            parsed = [[line, 1.0]]
+
+        tokenized = self.tokenize([text for text, _ in parsed])
+
+        tokens = []
+        multipliers = []
+
+        for text_tokens, (text, weight) in zip(tokenized, parsed):
+            if text == 'BREAK' and weight == -1:
+                continue
+
+            tokens += text_tokens
+            multipliers += [weight] * len(text_tokens)
+
+        tokens += [self.id_end]
+        multipliers += [1.0]
+
+        if target_token_count is not None:
+            if len(tokens) < target_token_count:
+                tokens += [self.id_pad] * (target_token_count - len(tokens))
+                multipliers += [1.0] * (target_token_count - len(tokens))
+            else:
+                tokens = tokens[0:target_token_count]
+                multipliers = multipliers[0:target_token_count]
+
+        return tokens, multipliers
+
+    def forward(self, texts, *, token_count):
+        if not self.t5xxl or not shared.opts.sd3_enable_t5:
+            return torch.zeros((len(texts), token_count, 4096), device=devices.device, dtype=devices.dtype)
+
+        tokens_batch = []
+
+        for text in texts:
+            tokens, multipliers = self.tokenize_line(text, target_token_count=token_count)
+            tokens_batch.append(tokens)
+
+        t5_out, t5_pooled = self.t5xxl(tokens_batch)
+
+        return t5_out
+
+    def encode_embedding_init_text(self, init_text, nvpt):
+        return torch.zeros((nvpt, 4096), device=devices.device) # XXX
+
+
+class SD3Cond(torch.nn.Module):
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+
+        self.tokenizer = SD3Tokenizer()
+
+        with torch.no_grad():
+            self.clip_g = SDXLClipG(CLIPG_CONFIG, device="cpu", dtype=devices.dtype)
+            self.clip_l = SDClipModel(layer="hidden", layer_idx=-2, device="cpu", dtype=devices.dtype, layer_norm_hidden_state=False, return_projected_pooled=False, textmodel_json_config=CLIPL_CONFIG)
+
+            if shared.opts.sd3_enable_t5:
+                self.t5xxl = T5XXLModel(T5_CONFIG, device="cpu", dtype=devices.dtype)
+            else:
+                self.t5xxl = None
+
+            self.model_lg = Sd3ClipLG(self.clip_l, self.clip_g)
+            self.model_t5 = Sd3T5(self.t5xxl)
+
+    def forward(self, prompts: list[str]):
+        with devices.without_autocast():
+            lg_out, vector_out = self.model_lg(prompts)
+            t5_out = self.model_t5(prompts, token_count=lg_out.shape[1])
+            lgt_out = torch.cat([lg_out, t5_out], dim=-2)
+
+        return {
+            'crossattn': lgt_out,
+            'vector': vector_out,
+        }
+
+    def before_load_weights(self, state_dict):
+        clip_path = os.path.join(shared.models_path, "CLIP")
+
+        if 'text_encoders.clip_g.transformer.text_model.embeddings.position_embedding.weight' not in state_dict:
+            clip_g_file = modelloader.load_file_from_url(CLIPG_URL, model_dir=clip_path, file_name="clip_g.safetensors")
+            with safetensors.safe_open(clip_g_file, framework="pt") as file:
+                self.clip_g.transformer.load_state_dict(SafetensorsMapping(file))
+
+        if 'text_encoders.clip_l.transformer.text_model.embeddings.position_embedding.weight' not in state_dict:
+            clip_l_file = modelloader.load_file_from_url(CLIPL_URL, model_dir=clip_path, file_name="clip_l.safetensors")
+            with safetensors.safe_open(clip_l_file, framework="pt") as file:
+                self.clip_l.transformer.load_state_dict(SafetensorsMapping(file), strict=False)
+
+        if self.t5xxl and 'text_encoders.t5xxl.transformer.encoder.embed_tokens.weight' not in state_dict:
+            t5_file = modelloader.load_file_from_url(T5_URL, model_dir=clip_path, file_name="t5xxl_fp16.safetensors")
+            with safetensors.safe_open(t5_file, framework="pt") as file:
+                self.t5xxl.transformer.load_state_dict(SafetensorsMapping(file), strict=False)
+
+    def encode_embedding_init_text(self, init_text, nvpt):
+        return torch.tensor([[0]], device=devices.device) # XXX
+
+    def medvram_modules(self):
+        return [self.clip_g, self.clip_l, self.t5xxl]
+
+    def get_token_count(self, text):
+        _, token_count = self.model_lg.process_texts([text])
+
+        return token_count
+
+    def get_target_prompt_token_count(self, token_count):
+        return self.model_lg.get_target_prompt_token_count(token_count)

modules/models/sd3/sd3_impls.py

@@ -0,0 +1,373 @@
+### Impls of the SD3 core diffusion model and VAE
+
+import torch
+import math
+import einops
+from modules.models.sd3.mmdit import MMDiT
+from PIL import Image
+
+
+#################################################################################################
+### MMDiT Model Wrapping
+#################################################################################################
+
+
+class ModelSamplingDiscreteFlow(torch.nn.Module):
+    """Helper for sampler scheduling (ie timestep/sigma calculations) for Discrete Flow models"""
+    def __init__(self, shift=1.0):
+        super().__init__()
+        self.shift = shift
+        timesteps = 1000
+        ts = self.sigma(torch.arange(1, timesteps + 1, 1))
+        self.register_buffer('sigmas', ts)
+
+    @property
+    def sigma_min(self):
+        return self.sigmas[0]
+
+    @property
+    def sigma_max(self):
+        return self.sigmas[-1]
+
+    def timestep(self, sigma):
+        return sigma * 1000
+
+    def sigma(self, timestep: torch.Tensor):
+        timestep = timestep / 1000.0
+        if self.shift == 1.0:
+            return timestep
+        return self.shift * timestep / (1 + (self.shift - 1) * timestep)
+
+    def calculate_denoised(self, sigma, model_output, model_input):
+        sigma = sigma.view(sigma.shape[:1] + (1,) * (model_output.ndim - 1))
+        return model_input - model_output * sigma
+
+    def noise_scaling(self, sigma, noise, latent_image, max_denoise=False):
+        return sigma * noise + (1.0 - sigma) * latent_image
+
+
+class BaseModel(torch.nn.Module):
+    """Wrapper around the core MM-DiT model"""
+    def __init__(self, shift=1.0, device=None, dtype=torch.float32, state_dict=None, prefix=""):
+        super().__init__()
+        # Important configuration values can be quickly determined by checking shapes in the source file
+        # Some of these will vary between models (eg 2B vs 8B primarily differ in their depth, but also other details change)
+        patch_size = state_dict[f"{prefix}x_embedder.proj.weight"].shape[2]
+        depth = state_dict[f"{prefix}x_embedder.proj.weight"].shape[0] // 64
+        num_patches = state_dict[f"{prefix}pos_embed"].shape[1]
+        pos_embed_max_size = round(math.sqrt(num_patches))
+        adm_in_channels = state_dict[f"{prefix}y_embedder.mlp.0.weight"].shape[1]
+        context_shape = state_dict[f"{prefix}context_embedder.weight"].shape
+        context_embedder_config = {
+            "target": "torch.nn.Linear",
+            "params": {
+                "in_features": context_shape[1],
+                "out_features": context_shape[0]
+            }
+        }
+        self.diffusion_model = MMDiT(input_size=None, pos_embed_scaling_factor=None, pos_embed_offset=None, pos_embed_max_size=pos_embed_max_size, patch_size=patch_size, in_channels=16, depth=depth, num_patches=num_patches, adm_in_channels=adm_in_channels, context_embedder_config=context_embedder_config, device=device, dtype=dtype)
+        self.model_sampling = ModelSamplingDiscreteFlow(shift=shift)
+
+    def apply_model(self, x, sigma, c_crossattn=None, y=None):
+        dtype = self.get_dtype()
+        timestep = self.model_sampling.timestep(sigma).float()
+        model_output = self.diffusion_model(x.to(dtype), timestep, context=c_crossattn.to(dtype), y=y.to(dtype)).float()
+        return self.model_sampling.calculate_denoised(sigma, model_output, x)
+
+    def forward(self, *args, **kwargs):
+        return self.apply_model(*args, **kwargs)
+
+    def get_dtype(self):
+        return self.diffusion_model.dtype
+
+
+class CFGDenoiser(torch.nn.Module):
+    """Helper for applying CFG Scaling to diffusion outputs"""
+    def __init__(self, model):
+        super().__init__()
+        self.model = model
+
+    def forward(self, x, timestep, cond, uncond, cond_scale):
+        # Run cond and uncond in a batch together
+        batched = self.model.apply_model(torch.cat([x, x]), torch.cat([timestep, timestep]), c_crossattn=torch.cat([cond["c_crossattn"], uncond["c_crossattn"]]), y=torch.cat([cond["y"], uncond["y"]]))
+        # Then split and apply CFG Scaling
+        pos_out, neg_out = batched.chunk(2)
+        scaled = neg_out + (pos_out - neg_out) * cond_scale
+        return scaled
+
+
+class SD3LatentFormat:
+    """Latents are slightly shifted from center - this class must be called after VAE Decode to correct for the shift"""
+    def __init__(self):
+        self.scale_factor = 1.5305
+        self.shift_factor = 0.0609
+
+    def process_in(self, latent):
+        return (latent - self.shift_factor) * self.scale_factor
+
+    def process_out(self, latent):
+        return (latent / self.scale_factor) + self.shift_factor
+
+    def decode_latent_to_preview(self, x0):
+        """Quick RGB approximate preview of sd3 latents"""
+        factors = torch.tensor([
+            [-0.0645,  0.0177,  0.1052], [ 0.0028,  0.0312,  0.0650],
+            [ 0.1848,  0.0762,  0.0360], [ 0.0944,  0.0360,  0.0889],
+            [ 0.0897,  0.0506, -0.0364], [-0.0020,  0.1203,  0.0284],
+            [ 0.0855,  0.0118,  0.0283], [-0.0539,  0.0658,  0.1047],
+            [-0.0057,  0.0116,  0.0700], [-0.0412,  0.0281, -0.0039],
+            [ 0.1106,  0.1171,  0.1220], [-0.0248,  0.0682, -0.0481],
+            [ 0.0815,  0.0846,  0.1207], [-0.0120, -0.0055, -0.0867],
+            [-0.0749, -0.0634, -0.0456], [-0.1418, -0.1457, -0.1259]
+        ], device="cpu")
+        latent_image = x0[0].permute(1, 2, 0).cpu() @ factors
+
+        latents_ubyte = (((latent_image + 1) / 2)
+                            .clamp(0, 1)  # change scale from -1..1 to 0..1
+                            .mul(0xFF)  # to 0..255
+                            .byte()).cpu()
+
+        return Image.fromarray(latents_ubyte.numpy())
+
+
+#################################################################################################
+### K-Diffusion Sampling
+#################################################################################################
+
+
+def append_dims(x, target_dims):
+    """Appends dimensions to the end of a tensor until it has target_dims dimensions."""
+    dims_to_append = target_dims - x.ndim
+    return x[(...,) + (None,) * dims_to_append]
+
+
+def to_d(x, sigma, denoised):
+    """Converts a denoiser output to a Karras ODE derivative."""
+    return (x - denoised) / append_dims(sigma, x.ndim)
+
+
+@torch.no_grad()
+@torch.autocast("cuda", dtype=torch.float16)
+def sample_euler(model, x, sigmas, extra_args=None):
+    """Implements Algorithm 2 (Euler steps) from Karras et al. (2022)."""
+    extra_args = {} if extra_args is None else extra_args
+    s_in = x.new_ones([x.shape[0]])
+    for i in range(len(sigmas) - 1):
+        sigma_hat = sigmas[i]
+        denoised = model(x, sigma_hat * s_in, **extra_args)
+        d = to_d(x, sigma_hat, denoised)
+        dt = sigmas[i + 1] - sigma_hat
+        # Euler method
+        x = x + d * dt
+    return x
+
+
+#################################################################################################
+### VAE
+#################################################################################################
+
+
+def Normalize(in_channels, num_groups=32, dtype=torch.float32, device=None):
+    return torch.nn.GroupNorm(num_groups=num_groups, num_channels=in_channels, eps=1e-6, affine=True, dtype=dtype, device=device)
+
+
+class ResnetBlock(torch.nn.Module):
+    def __init__(self, *, in_channels, out_channels=None, dtype=torch.float32, device=None):
+        super().__init__()
+        self.in_channels = in_channels
+        out_channels = in_channels if out_channels is None else out_channels
+        self.out_channels = out_channels
+
+        self.norm1 = Normalize(in_channels, dtype=dtype, device=device)
+        self.conv1 = torch.nn.Conv2d(in_channels, out_channels, kernel_size=3, stride=1, padding=1, dtype=dtype, device=device)
+        self.norm2 = Normalize(out_channels, dtype=dtype, device=device)
+        self.conv2 = torch.nn.Conv2d(out_channels, out_channels, kernel_size=3, stride=1, padding=1, dtype=dtype, device=device)
+        if self.in_channels != self.out_channels:
+            self.nin_shortcut = torch.nn.Conv2d(in_channels, out_channels, kernel_size=1, stride=1, padding=0, dtype=dtype, device=device)
+        else:
+            self.nin_shortcut = None
+        self.swish = torch.nn.SiLU(inplace=True)
+
+    def forward(self, x):
+        hidden = x
+        hidden = self.norm1(hidden)
+        hidden = self.swish(hidden)
+        hidden = self.conv1(hidden)
+        hidden = self.norm2(hidden)
+        hidden = self.swish(hidden)
+        hidden = self.conv2(hidden)
+        if self.in_channels != self.out_channels:
+            x = self.nin_shortcut(x)
+        return x + hidden
+
+
+class AttnBlock(torch.nn.Module):
+    def __init__(self, in_channels, dtype=torch.float32, device=None):
+        super().__init__()
+        self.norm = Normalize(in_channels, dtype=dtype, device=device)
+        self.q = torch.nn.Conv2d(in_channels, in_channels, kernel_size=1, stride=1, padding=0, dtype=dtype, device=device)
+        self.k = torch.nn.Conv2d(in_channels, in_channels, kernel_size=1, stride=1, padding=0, dtype=dtype, device=device)
+        self.v = torch.nn.Conv2d(in_channels, in_channels, kernel_size=1, stride=1, padding=0, dtype=dtype, device=device)
+        self.proj_out = torch.nn.Conv2d(in_channels, in_channels, kernel_size=1, stride=1, padding=0, dtype=dtype, device=device)
+
+    def forward(self, x):
+        hidden = self.norm(x)
+        q = self.q(hidden)
+        k = self.k(hidden)
+        v = self.v(hidden)
+        b, c, h, w = q.shape
+        q, k, v = [einops.rearrange(x, "b c h w -> b 1 (h w) c").contiguous() for x in (q, k, v)]
+        hidden = torch.nn.functional.scaled_dot_product_attention(q, k, v)  # scale is dim ** -0.5 per default
+        hidden = einops.rearrange(hidden, "b 1 (h w) c -> b c h w", h=h, w=w, c=c, b=b)
+        hidden = self.proj_out(hidden)
+        return x + hidden
+
+
+class Downsample(torch.nn.Module):
+    def __init__(self, in_channels, dtype=torch.float32, device=None):
+        super().__init__()
+        self.conv = torch.nn.Conv2d(in_channels, in_channels, kernel_size=3, stride=2, padding=0, dtype=dtype, device=device)
+
+    def forward(self, x):
+        pad = (0,1,0,1)
+        x = torch.nn.functional.pad(x, pad, mode="constant", value=0)
+        x = self.conv(x)
+        return x
+
+
+class Upsample(torch.nn.Module):
+    def __init__(self, in_channels, dtype=torch.float32, device=None):
+        super().__init__()
+        self.conv = torch.nn.Conv2d(in_channels, in_channels, kernel_size=3, stride=1, padding=1, dtype=dtype, device=device)
+
+    def forward(self, x):
+        x = torch.nn.functional.interpolate(x, scale_factor=2.0, mode="nearest")
+        x = self.conv(x)
+        return x
+
+
+class VAEEncoder(torch.nn.Module):
+    def __init__(self, ch=128, ch_mult=(1,2,4,4), num_res_blocks=2, in_channels=3, z_channels=16, dtype=torch.float32, device=None):
+        super().__init__()
+        self.num_resolutions = len(ch_mult)
+        self.num_res_blocks = num_res_blocks
+        # downsampling
+        self.conv_in = torch.nn.Conv2d(in_channels, ch, kernel_size=3, stride=1, padding=1, dtype=dtype, device=device)
+        in_ch_mult = (1,) + tuple(ch_mult)
+        self.in_ch_mult = in_ch_mult
+        self.down = torch.nn.ModuleList()
+        for i_level in range(self.num_resolutions):
+            block = torch.nn.ModuleList()
+            attn = torch.nn.ModuleList()
+            block_in = ch*in_ch_mult[i_level]
+            block_out = ch*ch_mult[i_level]
+            for _ in range(num_res_blocks):
+                block.append(ResnetBlock(in_channels=block_in, out_channels=block_out, dtype=dtype, device=device))
+                block_in = block_out
+            down = torch.nn.Module()
+            down.block = block
+            down.attn = attn
+            if i_level != self.num_resolutions - 1:
+                down.downsample = Downsample(block_in, dtype=dtype, device=device)
+            self.down.append(down)
+        # middle
+        self.mid = torch.nn.Module()
+        self.mid.block_1 = ResnetBlock(in_channels=block_in, out_channels=block_in, dtype=dtype, device=device)
+        self.mid.attn_1 = AttnBlock(block_in, dtype=dtype, device=device)
+        self.mid.block_2 = ResnetBlock(in_channels=block_in, out_channels=block_in, dtype=dtype, device=device)
+        # end
+        self.norm_out = Normalize(block_in, dtype=dtype, device=device)
+        self.conv_out = torch.nn.Conv2d(block_in, 2 * z_channels, kernel_size=3, stride=1, padding=1, dtype=dtype, device=device)
+        self.swish = torch.nn.SiLU(inplace=True)
+
+    def forward(self, x):
+        # downsampling
+        hs = [self.conv_in(x)]
+        for i_level in range(self.num_resolutions):
+            for i_block in range(self.num_res_blocks):
+                h = self.down[i_level].block[i_block](hs[-1])
+                hs.append(h)
+            if i_level != self.num_resolutions-1:
+                hs.append(self.down[i_level].downsample(hs[-1]))
+        # middle
+        h = hs[-1]
+        h = self.mid.block_1(h)
+        h = self.mid.attn_1(h)
+        h = self.mid.block_2(h)
+        # end
+        h = self.norm_out(h)
+        h = self.swish(h)
+        h = self.conv_out(h)
+        return h
+
+
+class VAEDecoder(torch.nn.Module):
+    def __init__(self, ch=128, out_ch=3, ch_mult=(1, 2, 4, 4), num_res_blocks=2, resolution=256, z_channels=16, dtype=torch.float32, device=None):
+        super().__init__()
+        self.num_resolutions = len(ch_mult)
+        self.num_res_blocks = num_res_blocks
+        block_in = ch * ch_mult[self.num_resolutions - 1]
+        curr_res = resolution // 2 ** (self.num_resolutions - 1)
+        # z to block_in
+        self.conv_in = torch.nn.Conv2d(z_channels, block_in, kernel_size=3, stride=1, padding=1, dtype=dtype, device=device)
+        # middle
+        self.mid = torch.nn.Module()
+        self.mid.block_1 = ResnetBlock(in_channels=block_in, out_channels=block_in, dtype=dtype, device=device)
+        self.mid.attn_1 = AttnBlock(block_in, dtype=dtype, device=device)
+        self.mid.block_2 = ResnetBlock(in_channels=block_in, out_channels=block_in, dtype=dtype, device=device)
+        # upsampling
+        self.up = torch.nn.ModuleList()
+        for i_level in reversed(range(self.num_resolutions)):
+            block = torch.nn.ModuleList()
+            block_out = ch * ch_mult[i_level]
+            for _ in range(self.num_res_blocks + 1):
+                block.append(ResnetBlock(in_channels=block_in, out_channels=block_out, dtype=dtype, device=device))
+                block_in = block_out
+            up = torch.nn.Module()
+            up.block = block
+            if i_level != 0:
+                up.upsample = Upsample(block_in, dtype=dtype, device=device)
+                curr_res = curr_res * 2
+            self.up.insert(0, up) # prepend to get consistent order
+        # end
+        self.norm_out = Normalize(block_in, dtype=dtype, device=device)
+        self.conv_out = torch.nn.Conv2d(block_in, out_ch, kernel_size=3, stride=1, padding=1, dtype=dtype, device=device)
+        self.swish = torch.nn.SiLU(inplace=True)
+
+    def forward(self, z):
+        # z to block_in
+        hidden = self.conv_in(z)
+        # middle
+        hidden = self.mid.block_1(hidden)
+        hidden = self.mid.attn_1(hidden)
+        hidden = self.mid.block_2(hidden)
+        # upsampling
+        for i_level in reversed(range(self.num_resolutions)):
+            for i_block in range(self.num_res_blocks + 1):
+                hidden = self.up[i_level].block[i_block](hidden)
+            if i_level != 0:
+                hidden = self.up[i_level].upsample(hidden)
+        # end
+        hidden = self.norm_out(hidden)
+        hidden = self.swish(hidden)
+        hidden = self.conv_out(hidden)
+        return hidden
+
+
+class SDVAE(torch.nn.Module):
+    def __init__(self, dtype=torch.float32, device=None):
+        super().__init__()
+        self.encoder = VAEEncoder(dtype=dtype, device=device)
+        self.decoder = VAEDecoder(dtype=dtype, device=device)
+
+    @torch.autocast("cuda", dtype=torch.float16)
+    def decode(self, latent):
+        return self.decoder(latent)
+
+    @torch.autocast("cuda", dtype=torch.float16)
+    def encode(self, image):
+        hidden = self.encoder(image)
+        mean, logvar = torch.chunk(hidden, 2, dim=1)
+        logvar = torch.clamp(logvar, -30.0, 20.0)
+        std = torch.exp(0.5 * logvar)
+        return mean + std * torch.randn_like(mean)

modules/models/sd3/sd3_model.py

@@ -0,0 +1,84 @@
+import contextlib
+
+import torch
+
+import k_diffusion
+from modules.models.sd3.sd3_impls import BaseModel, SDVAE, SD3LatentFormat
+from modules.models.sd3.sd3_cond import SD3Cond
+
+from modules import shared, devices
+
+
+class SD3Denoiser(k_diffusion.external.DiscreteSchedule):
+    def __init__(self, inner_model, sigmas):
+        super().__init__(sigmas, quantize=shared.opts.enable_quantization)
+        self.inner_model = inner_model
+
+    def forward(self, input, sigma, **kwargs):
+        return self.inner_model.apply_model(input, sigma, **kwargs)
+
+
+class SD3Inferencer(torch.nn.Module):
+    def __init__(self, state_dict, shift=3, use_ema=False):
+        super().__init__()
+
+        self.shift = shift
+
+        with torch.no_grad():
+            self.model = BaseModel(shift=shift, state_dict=state_dict, prefix="model.diffusion_model.", device="cpu", dtype=devices.dtype)
+            self.first_stage_model = SDVAE(device="cpu", dtype=devices.dtype_vae)
+            self.first_stage_model.dtype = self.model.diffusion_model.dtype
+
+        self.alphas_cumprod = 1 / (self.model.model_sampling.sigmas ** 2 + 1)
+
+        self.text_encoders = SD3Cond()
+        self.cond_stage_key = 'txt'
+
+        self.parameterization = "eps"
+        self.model.conditioning_key = "crossattn"
+
+        self.latent_format = SD3LatentFormat()
+        self.latent_channels = 16
+
+    @property
+    def cond_stage_model(self):
+        return self.text_encoders
+
+    def before_load_weights(self, state_dict):
+        self.cond_stage_model.before_load_weights(state_dict)
+
+    def ema_scope(self):
+        return contextlib.nullcontext()
+
+    def get_learned_conditioning(self, batch: list[str]):
+        return self.cond_stage_model(batch)
+
+    def apply_model(self, x, t, cond):
+        return self.model(x, t, c_crossattn=cond['crossattn'], y=cond['vector'])
+
+    def decode_first_stage(self, latent):
+        latent = self.latent_format.process_out(latent)
+        return self.first_stage_model.decode(latent)
+
+    def encode_first_stage(self, image):
+        latent = self.first_stage_model.encode(image)
+        return self.latent_format.process_in(latent)
+
+    def get_first_stage_encoding(self, x):
+        return x
+
+    def create_denoiser(self):
+        return SD3Denoiser(self, self.model.model_sampling.sigmas)
+
+    def medvram_fields(self):
+        return [
+            (self, 'first_stage_model'),
+            (self, 'text_encoders'),
+            (self, 'model'),
+        ]
+
+    def add_noise_to_latent(self, x, noise, amount):
+        return x * (1 - amount) + noise * amount
+
+    def fix_dimensions(self, width, height):
+        return width // 16 * 16, height // 16 * 16

modules/processing.py

@@ -884,6 +884,9 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed:
         if p.refiner_checkpoint_info is None:
             raise Exception(f'Could not find checkpoint with name {p.refiner_checkpoint}')
 
+    if hasattr(shared.sd_model, 'fix_dimensions'):
+        p.width, p.height = shared.sd_model.fix_dimensions(p.width, p.height)
+
     p.sd_model_name = shared.sd_model.sd_checkpoint_info.name_for_extra
     p.sd_model_hash = shared.sd_model.sd_model_hash
     p.sd_vae_name = sd_vae.get_loaded_vae_name()
@@ -942,7 +945,8 @@ def process_images_inner(p: StableDiffusionProcessing) -> Processed:
             p.seeds = p.all_seeds[n * p.batch_size:(n + 1) * p.batch_size]
             p.subseeds = p.all_subseeds[n * p.batch_size:(n + 1) * p.batch_size]
 
-            p.rng = rng.ImageRNG((opt_C, p.height // opt_f, p.width // opt_f), p.seeds, subseeds=p.subseeds, subseed_strength=p.subseed_strength, seed_resize_from_h=p.seed_resize_from_h, seed_resize_from_w=p.seed_resize_from_w)
+            latent_channels = getattr(shared.sd_model, 'latent_channels', opt_C)
+            p.rng = rng.ImageRNG((latent_channels, p.height // opt_f, p.width // opt_f), p.seeds, subseeds=p.subseeds, subseed_strength=p.subseed_strength, seed_resize_from_h=p.seed_resize_from_h, seed_resize_from_w=p.seed_resize_from_w)
 
             if p.scripts is not None:
                 p.scripts.before_process_batch(p, batch_number=n, prompts=p.prompts, seeds=p.seeds, subseeds=p.subseeds)
@@ -1736,10 +1740,10 @@ class StableDiffusionProcessingImg2Img(StableDiffusionProcessing):
             latmask = latmask[0]
             if self.mask_round:
                 latmask = np.around(latmask)
-            latmask = np.tile(latmask[None], (4, 1, 1))
+            latmask = np.tile(latmask[None], (self.init_latent.shape[1], 1, 1))
 
-            self.mask = torch.asarray(1.0 - latmask).to(shared.device).type(self.sd_model.dtype)
-            self.nmask = torch.asarray(latmask).to(shared.device).type(self.sd_model.dtype)
+            self.mask = torch.asarray(1.0 - latmask).to(shared.device).type(devices.dtype)
+            self.nmask = torch.asarray(latmask).to(shared.device).type(devices.dtype)
 
             # this needs to be fixed to be done in sample() using actual seeds for batches
             if self.inpainting_fill == 2:

modules/prompt_parser.py

@@ -268,7 +268,7 @@ def get_multicond_learned_conditioning(model, prompts, steps, hires_steps=None,
 
 
 class DictWithShape(dict):
-    def __init__(self, x, shape):
+    def __init__(self, x, shape=None):
         super().__init__()
         self.update(x)
 

modules/sd_hijack.py

@@ -325,7 +325,10 @@ class StableDiffusionModelHijack:
         if self.clip is None:
             return "-", "-"
 
-        _, token_count = self.clip.process_texts([text])
+        if hasattr(self.clip, 'get_token_count'):
+            token_count = self.clip.get_token_count(text)
+        else:
+            _, token_count = self.clip.process_texts([text])
 
         return token_count, self.clip.get_target_prompt_token_count(token_count)
 

modules/sd_hijack_clip.py

@@ -27,24 +27,21 @@ chunk. Those objects are found in PromptChunk.fixes and, are placed into FrozenC
 are applied by sd_hijack.EmbeddingsWithFixes's forward function."""
 
 
-class FrozenCLIPEmbedderWithCustomWordsBase(torch.nn.Module):
-    """A pytorch module that is a wrapper for FrozenCLIPEmbedder module. it enhances FrozenCLIPEmbedder, making it possible to
-    have unlimited prompt length and assign weights to tokens in prompt.
-    """
-
-    def __init__(self, wrapped, hijack):
+class TextConditionalModel(torch.nn.Module):
+    def __init__(self):
         super().__init__()
 
-        self.wrapped = wrapped
-        """Original FrozenCLIPEmbedder module; can also be FrozenOpenCLIPEmbedder or xlmr.BertSeriesModelWithTransformation,
-        depending on model."""
-
-        self.hijack: sd_hijack.StableDiffusionModelHijack = hijack
+        self.hijack = sd_hijack.model_hijack
         self.chunk_length = 75
 
-        self.is_trainable = getattr(wrapped, 'is_trainable', False)
-        self.input_key = getattr(wrapped, 'input_key', 'txt')
-        self.legacy_ucg_val = None
+        self.is_trainable = False
+        self.input_key = 'txt'
+        self.return_pooled = False
+
+        self.comma_token = None
+        self.id_start = None
+        self.id_end = None
+        self.id_pad = None
 
     def empty_chunk(self):
         """creates an empty PromptChunk and returns it"""
@@ -210,10 +207,6 @@ class FrozenCLIPEmbedderWithCustomWordsBase(torch.nn.Module):
         is when you do prompt editing: "a picture of a [cat:dog:0.4] eating ice cream"
         """
 
-        if opts.use_old_emphasis_implementation:
-            import modules.sd_hijack_clip_old
-            return modules.sd_hijack_clip_old.forward_old(self, texts)
-
         batch_chunks, token_count = self.process_texts(texts)
 
         used_embeddings = {}
@@ -252,7 +245,7 @@ class FrozenCLIPEmbedderWithCustomWordsBase(torch.nn.Module):
         if any(x for x in texts if "(" in x or "[" in x) and opts.emphasis != "Original":
             self.hijack.extra_generation_params["Emphasis"] = opts.emphasis
 
-        if getattr(self.wrapped, 'return_pooled', False):
+        if self.return_pooled:
             return torch.hstack(zs), zs[0].pooled
         else:
             return torch.hstack(zs)
@@ -292,6 +285,34 @@ class FrozenCLIPEmbedderWithCustomWordsBase(torch.nn.Module):
         return z
 
 
+class FrozenCLIPEmbedderWithCustomWordsBase(TextConditionalModel):
+    """A pytorch module that is a wrapper for FrozenCLIPEmbedder module. it enhances FrozenCLIPEmbedder, making it possible to
+    have unlimited prompt length and assign weights to tokens in prompt.
+    """
+
+    def __init__(self, wrapped, hijack):
+        super().__init__()
+
+        self.hijack = hijack
+
+        self.wrapped = wrapped
+        """Original FrozenCLIPEmbedder module; can also be FrozenOpenCLIPEmbedder or xlmr.BertSeriesModelWithTransformation,
+        depending on model."""
+
+        self.is_trainable = getattr(wrapped, 'is_trainable', False)
+        self.input_key = getattr(wrapped, 'input_key', 'txt')
+        self.return_pooled = getattr(self.wrapped, 'return_pooled', False)
+
+        self.legacy_ucg_val = None  # for sgm codebase
+
+    def forward(self, texts):
+        if opts.use_old_emphasis_implementation:
+            import modules.sd_hijack_clip_old
+            return modules.sd_hijack_clip_old.forward_old(self, texts)
+
+        return super().forward(texts)
+
+
 class FrozenCLIPEmbedderWithCustomWords(FrozenCLIPEmbedderWithCustomWordsBase):
     def __init__(self, wrapped, hijack):
         super().__init__(wrapped, hijack)

modules/sd_models.py

@@ -1,7 +1,9 @@
 import collections
+import importlib
 import os
 import sys
 import threading
+import enum
 
 import torch
 import re
@@ -10,8 +12,6 @@ from omegaconf import OmegaConf, ListConfig
 from urllib import request
 import ldm.modules.midas as midas
 
-from ldm.util import instantiate_from_config
-
 from modules import paths, shared, modelloader, devices, script_callbacks, sd_vae, sd_disable_initialization, errors, hashes, sd_models_config, sd_unet, sd_models_xl, cache, extra_networks, processing, lowvram, sd_hijack, patches
 from modules.timer import Timer
 from modules.shared import opts
@@ -27,6 +27,14 @@ checkpoint_alisases = checkpoint_aliases  # for compatibility with old name
 checkpoints_loaded = collections.OrderedDict()
 
 
+class ModelType(enum.Enum):
+    SD1 = 1
+    SD2 = 2
+    SDXL = 3
+    SSD = 4
+    SD3 = 5
+
+
 def replace_key(d, key, new_key, value):
     keys = list(d.keys())
 
@@ -368,6 +376,37 @@ def check_fp8(model):
     return enable_fp8
 
 
+def set_model_type(model, state_dict):
+    model.is_sd1 = False
+    model.is_sd2 = False
+    model.is_sdxl = False
+    model.is_ssd = False
+    model.is_sd3 = False
+
+    if "model.diffusion_model.x_embedder.proj.weight" in state_dict:
+        model.is_sd3 = True
+        model.model_type = ModelType.SD3
+    elif hasattr(model, 'conditioner'):
+        model.is_sdxl = True
+
+        if 'model.diffusion_model.middle_block.1.transformer_blocks.0.attn1.to_q.weight' not in state_dict.keys():
+            model.is_ssd = True
+            model.model_type = ModelType.SSD
+        else:
+            model.model_type = ModelType.SDXL
+    elif hasattr(model.cond_stage_model, 'model'):
+        model.is_sd2 = True
+        model.model_type = ModelType.SD2
+    else:
+        model.is_sd1 = True
+        model.model_type = ModelType.SD1
+
+
+def set_model_fields(model):
+    if not hasattr(model, 'latent_channels'):
+        model.latent_channels = 4
+
+
 def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer):
     sd_model_hash = checkpoint_info.calculate_shorthash()
     timer.record("calculate hash")
@@ -382,10 +421,9 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer
     if state_dict is None:
         state_dict = get_checkpoint_state_dict(checkpoint_info, timer)
 
-    model.is_sdxl = hasattr(model, 'conditioner')
-    model.is_sd2 = not model.is_sdxl and hasattr(model.cond_stage_model, 'model')
-    model.is_sd1 = not model.is_sdxl and not model.is_sd2
-    model.is_ssd = model.is_sdxl and 'model.diffusion_model.middle_block.1.transformer_blocks.0.attn1.to_q.weight' not in state_dict.keys()
+    set_model_type(model, state_dict)
+    set_model_fields(model)
+
     if model.is_sdxl:
         sd_models_xl.extend_sdxl(model)
 
@@ -396,9 +434,15 @@ def load_model_weights(model, checkpoint_info: CheckpointInfo, state_dict, timer
         # cache newly loaded model
         checkpoints_loaded[checkpoint_info] = state_dict.copy()
 
+    if hasattr(model, "before_load_weights"):
+        model.before_load_weights(state_dict)
+
     model.load_state_dict(state_dict, strict=False)
     timer.record("apply weights to model")
 
+    if hasattr(model, "after_load_weights"):
+        model.after_load_weights(state_dict)
+
     del state_dict
 
     # Set is_sdxl_inpaint flag.
@@ -552,8 +596,7 @@ def patch_given_betas():
     original_register_schedule = patches.patch(__name__, ldm.models.diffusion.ddpm.DDPM, 'register_schedule', patched_register_schedule)
 
 
-def repair_config(sd_config):
-
+def repair_config(sd_config, state_dict=None):
     if not hasattr(sd_config.model.params, "use_ema"):
         sd_config.model.params.use_ema = False
 
@@ -563,8 +606,9 @@ def repair_config(sd_config):
         elif shared.cmd_opts.upcast_sampling or shared.cmd_opts.precision == "half":
             sd_config.model.params.unet_config.params.use_fp16 = True
 
-    if getattr(sd_config.model.params.first_stage_config.params.ddconfig, "attn_type", None) == "vanilla-xformers" and not shared.xformers_available:
-        sd_config.model.params.first_stage_config.params.ddconfig.attn_type = "vanilla"
+    if hasattr(sd_config.model.params, 'first_stage_config'):
+        if getattr(sd_config.model.params.first_stage_config.params.ddconfig, "attn_type", None) == "vanilla-xformers" and not shared.xformers_available:
+            sd_config.model.params.first_stage_config.params.ddconfig.attn_type = "vanilla"
 
     # For UnCLIP-L, override the hardcoded karlo directory
     if hasattr(sd_config.model.params, "noise_aug_config") and hasattr(sd_config.model.params.noise_aug_config.params, "clip_stats_path"):
@@ -580,6 +624,7 @@ def repair_config(sd_config):
         sd_config.model.params.unet_config.params.use_checkpoint = False
 
 
+
 def rescale_zero_terminal_snr_abar(alphas_cumprod):
     alphas_bar_sqrt = alphas_cumprod.sqrt()
 
@@ -679,11 +724,15 @@ def get_empty_cond(sd_model):
     p = processing.StableDiffusionProcessingTxt2Img()
     extra_networks.activate(p, {})
 
-    if hasattr(sd_model, 'conditioner'):
+    if hasattr(sd_model, 'get_learned_conditioning'):
         d = sd_model.get_learned_conditioning([""])
-        return d['crossattn']
     else:
-        return sd_model.cond_stage_model([""])
+        d = sd_model.cond_stage_model([""])
+
+    if isinstance(d, dict):
+        d = d['crossattn']
+
+    return d
 
 
 def send_model_to_cpu(m):
@@ -715,6 +764,25 @@ def send_model_to_trash(m):
     devices.torch_gc()
 
 
+def instantiate_from_config(config, state_dict=None):
+    constructor = get_obj_from_str(config["target"])
+
+    params = {**config.get("params", {})}
+
+    if state_dict and "state_dict" in params and params["state_dict"] is None:
+        params["state_dict"] = state_dict
+
+    return constructor(**params)
+
+
+def get_obj_from_str(string, reload=False):
+    module, cls = string.rsplit(".", 1)
+    if reload:
+        module_imp = importlib.import_module(module)
+        importlib.reload(module_imp)
+    return getattr(importlib.import_module(module, package=None), cls)
+
+
 def load_model(checkpoint_info=None, already_loaded_state_dict=None):
     from modules import sd_hijack
     checkpoint_info = checkpoint_info or select_checkpoint()
@@ -739,7 +807,7 @@ def load_model(checkpoint_info=None, already_loaded_state_dict=None):
     timer.record("find config")
 
     sd_config = OmegaConf.load(checkpoint_config)
-    repair_config(sd_config)
+    repair_config(sd_config, state_dict)
 
     timer.record("load config")
 
@@ -749,7 +817,7 @@ def load_model(checkpoint_info=None, already_loaded_state_dict=None):
     try:
         with sd_disable_initialization.DisableInitialization(disable_clip=clip_is_included_into_sd or shared.cmd_opts.do_not_download_clip):
             with sd_disable_initialization.InitializeOnMeta():
-                sd_model = instantiate_from_config(sd_config.model)
+                sd_model = instantiate_from_config(sd_config.model, state_dict)
 
     except Exception as e:
         errors.display(e, "creating model quickly", full_traceback=True)
@@ -758,7 +826,7 @@ def load_model(checkpoint_info=None, already_loaded_state_dict=None):
         print('Failed to create model quickly; will retry using slow method.', file=sys.stderr)
 
         with sd_disable_initialization.InitializeOnMeta():
-            sd_model = instantiate_from_config(sd_config.model)
+            sd_model = instantiate_from_config(sd_config.model, state_dict)
 
     sd_model.used_config = checkpoint_config
 
@@ -775,6 +843,7 @@ def load_model(checkpoint_info=None, already_loaded_state_dict=None):
 
     with sd_disable_initialization.LoadStateDictOnMeta(state_dict, device=model_target_device(sd_model), weight_dtype_conversion=weight_dtype_conversion):
         load_model_weights(sd_model, checkpoint_info, state_dict, timer)
+
     timer.record("load weights from state dict")
 
     send_model_to_device(sd_model)

modules/sd_models_config.py

@@ -23,6 +23,8 @@ config_inpainting = os.path.join(sd_configs_path, "v1-inpainting-inference.yaml"
 config_instruct_pix2pix = os.path.join(sd_configs_path, "instruct-pix2pix.yaml")
 config_alt_diffusion = os.path.join(sd_configs_path, "alt-diffusion-inference.yaml")
 config_alt_diffusion_m18 = os.path.join(sd_configs_path, "alt-diffusion-m18-inference.yaml")
+config_sd3 = os.path.join(sd_configs_path, "sd3-inference.yaml")
+
 
 def is_using_v_parameterization_for_sd2(state_dict):
     """
@@ -71,11 +73,15 @@ def guess_model_config_from_state_dict(sd, filename):
     diffusion_model_input = sd.get('model.diffusion_model.input_blocks.0.0.weight', None)
     sd2_variations_weight = sd.get('embedder.model.ln_final.weight', None)
 
+    if "model.diffusion_model.x_embedder.proj.weight" in sd:
+        return config_sd3
+
     if sd.get('conditioner.embedders.1.model.ln_final.weight', None) is not None:
         if diffusion_model_input.shape[1] == 9:
             return config_sdxl_inpainting
         else:
             return config_sdxl
+
     if sd.get('conditioner.embedders.0.model.ln_final.weight', None) is not None:
         return config_sdxl_refiner
     elif sd.get('depth_model.model.pretrained.act_postprocess3.0.project.0.bias', None) is not None:
@@ -99,7 +105,6 @@ def guess_model_config_from_state_dict(sd, filename):
         if diffusion_model_input.shape[1] == 8:
             return config_instruct_pix2pix
 
-
     if sd.get('cond_stage_model.roberta.embeddings.word_embeddings.weight', None) is not None:
         if sd.get('cond_stage_model.transformation.weight').size()[0] == 1024:
             return config_alt_diffusion_m18

modules/sd_models_types.py

@@ -32,3 +32,9 @@ class WebuiSdModel(LatentDiffusion):
 
     is_sd1: bool
     """True if the model's architecture is SD 1.x"""
+
+    is_sd3: bool
+    """True if the model's architecture is SD 3"""
+
+    latent_channels: int
+    """number of layer in latent image representation; will be 16 in SD3 and 4 in other version"""

modules/sd_samplers_common.py

@@ -54,7 +54,7 @@ def samples_to_images_tensor(sample, approximation=None, model=None):
     else:
         if model is None:
             model = shared.sd_model
-        with devices.without_autocast(): # fixes an issue with unstable VAEs that are flaky even in fp32
+        with torch.no_grad(), devices.without_autocast(): # fixes an issue with unstable VAEs that are flaky even in fp32
             x_sample = model.decode_first_stage(sample.to(model.first_stage_model.dtype))
 
     return x_sample
@@ -163,7 +163,7 @@ def apply_refiner(cfg_denoiser, sigma=None):
     else:
         # torch.max(sigma) only to handle rare case where we might have different sigmas in the same batch
         try:
-            timestep = torch.argmin(torch.abs(cfg_denoiser.inner_model.sigmas - torch.max(sigma)))
+            timestep = torch.argmin(torch.abs(cfg_denoiser.inner_model.sigmas.to(sigma.device) - torch.max(sigma)))
         except AttributeError:  # for samplers that don't use sigmas (DDIM) sigma is actually the timestep
             timestep = torch.max(sigma).to(dtype=int)
         completed_ratio = (999 - timestep) / 1000
@@ -246,7 +246,7 @@ class Sampler:
         self.eta_infotext_field = 'Eta'
         self.eta_default = 1.0
 
-        self.conditioning_key = shared.sd_model.model.conditioning_key
+        self.conditioning_key = getattr(shared.sd_model.model, 'conditioning_key', 'crossattn')
 
         self.p = None
         self.model_wrap_cfg = None

modules/sd_samplers_kdiffusion.py

@@ -53,8 +53,13 @@ class CFGDenoiserKDiffusion(sd_samplers_cfg_denoiser.CFGDenoiser):
     @property
     def inner_model(self):
         if self.model_wrap is None:
-            denoiser = k_diffusion.external.CompVisVDenoiser if shared.sd_model.parameterization == "v" else k_diffusion.external.CompVisDenoiser
-            self.model_wrap = denoiser(shared.sd_model, quantize=shared.opts.enable_quantization)
+            denoiser_constructor = getattr(shared.sd_model, 'create_denoiser', None)
+
+            if denoiser_constructor is not None:
+                self.model_wrap = denoiser_constructor()
+            else:
+                denoiser = k_diffusion.external.CompVisVDenoiser if shared.sd_model.parameterization == "v" else k_diffusion.external.CompVisDenoiser
+                self.model_wrap = denoiser(shared.sd_model, quantize=shared.opts.enable_quantization)
 
         return self.model_wrap
 
@@ -120,7 +125,7 @@ class KDiffusionSampler(sd_samplers_common.Sampler):
         if discard_next_to_last_sigma:
             sigmas = torch.cat([sigmas[:-2], sigmas[-1:]])
 
-        return sigmas
+        return sigmas.cpu()
 
     def sample_img2img(self, p, x, noise, conditioning, unconditional_conditioning, steps=None, image_conditioning=None):
         steps, t_enc = sd_samplers_common.setup_img2img_steps(p, steps)
@@ -128,7 +133,10 @@ class KDiffusionSampler(sd_samplers_common.Sampler):
         sigmas = self.get_sigmas(p, steps)
         sigma_sched = sigmas[steps - t_enc - 1:]
 
-        xi = x + noise * sigma_sched[0]
+        if hasattr(shared.sd_model, 'add_noise_to_latent'):
+            xi = shared.sd_model.add_noise_to_latent(x, noise, sigma_sched[0])
+        else:
+            xi = x + noise * sigma_sched[0]
 
         if opts.img2img_extra_noise > 0:
             p.extra_generation_params["Extra noise"] = opts.img2img_extra_noise

modules/sd_vae_approx.py

@@ -8,9 +8,9 @@ sd_vae_approx_models = {}
 
 
 class VAEApprox(nn.Module):
-    def __init__(self):
+    def __init__(self, latent_channels=4):
         super(VAEApprox, self).__init__()
-        self.conv1 = nn.Conv2d(4, 8, (7, 7))
+        self.conv1 = nn.Conv2d(latent_channels, 8, (7, 7))
         self.conv2 = nn.Conv2d(8, 16, (5, 5))
         self.conv3 = nn.Conv2d(16, 32, (3, 3))
         self.conv4 = nn.Conv2d(32, 64, (3, 3))
@@ -40,7 +40,13 @@ def download_model(model_path, model_url):
 
 
 def model():
-    model_name = "vaeapprox-sdxl.pt" if getattr(shared.sd_model, 'is_sdxl', False) else "model.pt"
+    if shared.sd_model.is_sd3:
+        model_name = "vaeapprox-sd3.pt"
+    elif shared.sd_model.is_sdxl:
+        model_name = "vaeapprox-sdxl.pt"
+    else:
+        model_name = "model.pt"
+
     loaded_model = sd_vae_approx_models.get(model_name)
 
     if loaded_model is None:
@@ -52,7 +58,7 @@ def model():
             model_path = os.path.join(paths.models_path, "VAE-approx", model_name)
             download_model(model_path, 'https://github.com/AUTOMATIC1111/stable-diffusion-webui/releases/download/v1.0.0-pre/' + model_name)
 
-        loaded_model = VAEApprox()
+        loaded_model = VAEApprox(latent_channels=shared.sd_model.latent_channels)
         loaded_model.load_state_dict(torch.load(model_path, map_location='cpu' if devices.device.type != 'cuda' else None))
         loaded_model.eval()
         loaded_model.to(devices.device, devices.dtype)
@@ -64,7 +70,18 @@ def model():
 def cheap_approximation(sample):
     # https://discuss.huggingface.co/t/decoding-latents-to-rgb-without-upscaling/23204/2
 
-    if shared.sd_model.is_sdxl:
+    if shared.sd_model.is_sd3:
+        coeffs = [
+            [-0.0645,  0.0177,  0.1052], [ 0.0028,  0.0312,  0.0650],
+            [ 0.1848,  0.0762,  0.0360], [ 0.0944,  0.0360,  0.0889],
+            [ 0.0897,  0.0506, -0.0364], [-0.0020,  0.1203,  0.0284],
+            [ 0.0855,  0.0118,  0.0283], [-0.0539,  0.0658,  0.1047],
+            [-0.0057,  0.0116,  0.0700], [-0.0412,  0.0281, -0.0039],
+            [ 0.1106,  0.1171,  0.1220], [-0.0248,  0.0682, -0.0481],
+            [ 0.0815,  0.0846,  0.1207], [-0.0120, -0.0055, -0.0867],
+            [-0.0749, -0.0634, -0.0456], [-0.1418, -0.1457, -0.1259],
+        ]
+    elif shared.sd_model.is_sdxl:
         coeffs = [
             [ 0.3448,  0.4168,  0.4395],
             [-0.1953, -0.0290,  0.0250],

modules/sd_vae_taesd.py

@@ -34,9 +34,9 @@ class Block(nn.Module):
         return self.fuse(self.conv(x) + self.skip(x))
 
 
-def decoder():
+def decoder(latent_channels=4):
     return nn.Sequential(
-        Clamp(), conv(4, 64), nn.ReLU(),
+        Clamp(), conv(latent_channels, 64), nn.ReLU(),
         Block(64, 64), Block(64, 64), Block(64, 64), nn.Upsample(scale_factor=2), conv(64, 64, bias=False),
         Block(64, 64), Block(64, 64), Block(64, 64), nn.Upsample(scale_factor=2), conv(64, 64, bias=False),
         Block(64, 64), Block(64, 64), Block(64, 64), nn.Upsample(scale_factor=2), conv(64, 64, bias=False),
@@ -44,13 +44,13 @@ def decoder():
     )
 
 
-def encoder():
+def encoder(latent_channels=4):
     return nn.Sequential(
         conv(3, 64), Block(64, 64),
         conv(64, 64, stride=2, bias=False), Block(64, 64), Block(64, 64), Block(64, 64),
         conv(64, 64, stride=2, bias=False), Block(64, 64), Block(64, 64), Block(64, 64),
         conv(64, 64, stride=2, bias=False), Block(64, 64), Block(64, 64), Block(64, 64),
-        conv(64, 4),
+        conv(64, latent_channels),
     )
 
 
@@ -58,10 +58,14 @@ class TAESDDecoder(nn.Module):
     latent_magnitude = 3
     latent_shift = 0.5
 
-    def __init__(self, decoder_path="taesd_decoder.pth"):
+    def __init__(self, decoder_path="taesd_decoder.pth", latent_channels=None):
         """Initialize pretrained TAESD on the given device from the given checkpoints."""
         super().__init__()
-        self.decoder = decoder()
+
+        if latent_channels is None:
+            latent_channels = 16 if "taesd3" in str(decoder_path) else 4
+
+        self.decoder = decoder(latent_channels)
         self.decoder.load_state_dict(
             torch.load(decoder_path, map_location='cpu' if devices.device.type != 'cuda' else None))
 
@@ -70,10 +74,14 @@ class TAESDEncoder(nn.Module):
     latent_magnitude = 3
     latent_shift = 0.5
 
-    def __init__(self, encoder_path="taesd_encoder.pth"):
+    def __init__(self, encoder_path="taesd_encoder.pth", latent_channels=None):
         """Initialize pretrained TAESD on the given device from the given checkpoints."""
         super().__init__()
-        self.encoder = encoder()
+
+        if latent_channels is None:
+            latent_channels = 16 if "taesd3" in str(encoder_path) else 4
+
+        self.encoder = encoder(latent_channels)
         self.encoder.load_state_dict(
             torch.load(encoder_path, map_location='cpu' if devices.device.type != 'cuda' else None))
 
@@ -87,7 +95,13 @@ def download_model(model_path, model_url):
 
 
 def decoder_model():
-    model_name = "taesdxl_decoder.pth" if getattr(shared.sd_model, 'is_sdxl', False) else "taesd_decoder.pth"
+    if shared.sd_model.is_sd3:
+        model_name = "taesd3_decoder.pth"
+    elif shared.sd_model.is_sdxl:
+        model_name = "taesdxl_decoder.pth"
+    else:
+        model_name = "taesd_decoder.pth"
+
     loaded_model = sd_vae_taesd_models.get(model_name)
 
     if loaded_model is None:
@@ -106,7 +120,13 @@ def decoder_model():
 
 
 def encoder_model():
-    model_name = "taesdxl_encoder.pth" if getattr(shared.sd_model, 'is_sdxl', False) else "taesd_encoder.pth"
+    if shared.sd_model.is_sd3:
+        model_name = "taesd3_encoder.pth"
+    elif shared.sd_model.is_sdxl:
+        model_name = "taesdxl_encoder.pth"
+    else:
+        model_name = "taesd_encoder.pth"
+
     loaded_model = sd_vae_taesd_models.get(model_name)
 
     if loaded_model is None:

modules/shared_options.py

@@ -191,6 +191,10 @@ options_templates.update(options_section(('sdxl', "Stable Diffusion XL", "sd"),
     "sdxl_refiner_high_aesthetic_score": OptionInfo(6.0, "SDXL high aesthetic score", gr.Number).info("used for refiner model prompt"),
 }))
 
+options_templates.update(options_section(('sd3', "Stable Diffusion 3", "sd"), {
+    "sd3_enable_t5": OptionInfo(False, "Enable T5").info("load T5 text encoder; increases VRAM use by a lot, potentially improving quality of generation; requires model reload to apply"),
+}))
+
 options_templates.update(options_section(('vae', "VAE", "sd"), {
     "sd_vae_explanation": OptionHTML("""
 <abbr title='Variational autoencoder'>VAE</abbr> is a neural network that transforms a standard <abbr title='red/green/blue'>RGB</abbr>

requirements.txt

@@ -18,6 +18,7 @@ omegaconf
 open-clip-torch
 
 piexif
+protobuf==3.20.0
 psutil
 pytorch_lightning
 requests

requirements_versions.txt

@@ -18,6 +18,7 @@ numpy==1.26.2
 omegaconf==2.2.3
 open-clip-torch==2.20.0
 piexif==1.1.3
+protobuf==3.20.0
 psutil==5.9.5
 pytorch_lightning==1.9.4
 resize-right==0.0.2