Added ability to train control loras. Other important bug fixes thrown in

toolkit/models/control_lora_adapter.py

@@ -0,0 +1,239 @@
+import inspect
+import weakref
+import torch
+from typing import TYPE_CHECKING
+from toolkit.lora_special import LoRASpecialNetwork
+from diffusers import FluxTransformer2DModel
+# weakref
+
+
+if TYPE_CHECKING:
+    from toolkit.stable_diffusion_model import StableDiffusion
+    from toolkit.config_modules import AdapterConfig, TrainConfig, ModelConfig
+    from toolkit.custom_adapter import CustomAdapter
+    
+
+# after each step we concat the control image with the latents 
+# latent_model_input = torch.cat([latents, control_image], dim=2)
+# the x_embedder has a full rank lora to handle the additional channels
+# this replaces the x_embedder with a full rank lora. on flux this is 
+# x_embedder(diffusers) or img_in(bfl)
+
+# Flux
+# img_in.lora_A.weight	[128, 128]	
+# img_in.lora_B.bias	[3 072]	
+# img_in.lora_B.weight	[3 072, 128]	
+    
+
+class ImgEmbedder(torch.nn.Module):
+    def __init__(
+        self,
+        adapter: 'ControlLoraAdapter',
+        orig_layer: torch.nn.Module,
+        in_channels=128,
+        out_channels=3072,
+        bias=True
+    ):
+        super().__init__()
+        self.adapter_ref: weakref.ref = weakref.ref(adapter)
+        self.orig_layer_ref: weakref.ref = weakref.ref(orig_layer)
+        self.lora_A = torch.nn.Linear(in_channels, in_channels, bias=False) # lora down
+        self.lora_B = torch.nn.Linear(in_channels, out_channels, bias=bias) # lora up
+        
+    @classmethod
+    def from_model(
+        cls, 
+        model: FluxTransformer2DModel, 
+        adapter: 'ControlLoraAdapter', 
+        num_channel_multiplier=2
+    ):
+        if model.__class__.__name__ == 'FluxTransformer2DModel':
+            x_embedder: torch.nn.Linear = model.x_embedder
+            img_embedder = cls(
+                adapter, 
+                orig_layer=x_embedder,
+                in_channels=x_embedder.in_features * num_channel_multiplier, # adding additional control img channels
+                out_channels=x_embedder.out_features, 
+                bias=x_embedder.bias is not None
+            )
+            
+            # hijack the forward method
+            x_embedder._orig_ctrl_lora_forward = x_embedder.forward
+            x_embedder.forward = img_embedder.forward
+            dtype = x_embedder.weight.dtype
+            device = x_embedder.weight.device
+            
+            # since we are adding control channels, we want those channels to be zero starting out
+            # so they have no effect. It will match lora_B weight and bias, and we concat 0s for the input of the new channels
+            # lora_a needs to be identity so that lora_b output matches lora_a output on init
+            img_embedder.lora_A.weight.data = torch.eye(x_embedder.in_features * num_channel_multiplier).to(dtype=torch.float32, device=device)
+            weight_b = x_embedder.weight.data.clone().to(dtype=torch.float32, device=device)
+            # concat 0s for the new channels
+            weight_b = torch.cat([weight_b, torch.zeros(weight_b.shape[0], weight_b.shape[1] * (num_channel_multiplier - 1)).to(device)], dim=1)
+            img_embedder.lora_B.weight.data = weight_b.clone().to(dtype=torch.float32)
+            img_embedder.lora_B.bias.data = x_embedder.bias.data.clone().to(dtype=torch.float32)
+            
+            # update the config of the transformer
+            model.config.in_channels = model.config.in_channels * num_channel_multiplier
+            model.config["in_channels"] = model.config.in_channels
+            
+            return img_embedder
+        else:
+            raise ValueError("Model not supported") 
+        
+    @property
+    def is_active(self):
+        return self.adapter_ref().is_active
+        
+    
+    def forward(self, x):
+        if not self.is_active:
+            # make sure lora is not active
+            self.adapter_ref().control_lora.is_active = False
+            return self.orig_layer_ref()._orig_ctrl_lora_forward(x)
+        
+        # make sure lora is active
+        self.adapter_ref().control_lora.is_active = True
+        
+        orig_device = x.device
+        orig_dtype = x.dtype
+        x = x.to(self.lora_A.weight.device, dtype=self.lora_A.weight.dtype)
+        
+        x = self.lora_A(x)
+        x = self.lora_B(x)
+        x = x.to(orig_device, dtype=orig_dtype)
+        return x
+    
+
+
+class ControlLoraAdapter(torch.nn.Module):
+    def __init__(
+        self,
+        adapter: 'CustomAdapter',
+        sd: 'StableDiffusion',
+        config: 'AdapterConfig',
+        train_config: 'TrainConfig'
+    ):
+        super().__init__()
+        self.adapter_ref: weakref.ref = weakref.ref(adapter)
+        self.sd_ref = weakref.ref(sd)
+        self.model_config: ModelConfig = sd.model_config
+        self.network_config = config.lora_config
+        self.train_config = train_config
+        if self.network_config is None:
+            raise ValueError("LoRA config is missing")
+        
+        network_kwargs = {} if self.network_config.network_kwargs is None else self.network_config.network_kwargs
+        if hasattr(sd, 'target_lora_modules'):
+            network_kwargs['target_lin_modules'] = self.sd.target_lora_modules
+            
+        if 'ignore_if_contains' not in network_kwargs:
+            network_kwargs['ignore_if_contains'] = []
+        
+        # always ignore x_embedder
+        network_kwargs['ignore_if_contains'].append('x_embedder')
+            
+        self.device_torch = sd.device_torch
+        self.control_lora = LoRASpecialNetwork(
+            text_encoder=sd.text_encoder,
+            unet=sd.unet,
+            lora_dim=self.network_config.linear,
+            multiplier=1.0,
+            alpha=self.network_config.linear_alpha,
+            train_unet=self.train_config.train_unet,
+            train_text_encoder=self.train_config.train_text_encoder,
+            conv_lora_dim=self.network_config.conv,
+            conv_alpha=self.network_config.conv_alpha,
+            is_sdxl=self.model_config.is_xl or self.model_config.is_ssd,
+            is_v2=self.model_config.is_v2,
+            is_v3=self.model_config.is_v3,
+            is_pixart=self.model_config.is_pixart,
+            is_auraflow=self.model_config.is_auraflow,
+            is_flux=self.model_config.is_flux,
+            is_lumina2=self.model_config.is_lumina2,
+            is_ssd=self.model_config.is_ssd,
+            is_vega=self.model_config.is_vega,
+            dropout=self.network_config.dropout,
+            use_text_encoder_1=self.model_config.use_text_encoder_1,
+            use_text_encoder_2=self.model_config.use_text_encoder_2,
+            use_bias=False,
+            is_lorm=False,
+            network_config=self.network_config,
+            network_type=self.network_config.type,
+            transformer_only=self.network_config.transformer_only,
+            is_transformer=sd.is_transformer,
+            base_model=sd,
+            **network_kwargs
+        )
+        self.control_lora.force_to(self.device_torch, dtype=torch.float32)
+        self.control_lora._update_torch_multiplier()
+        self.control_lora.apply_to(
+            sd.text_encoder,
+            sd.unet,
+            self.train_config.train_text_encoder,
+            self.train_config.train_unet
+        )
+        self.control_lora.can_merge_in = False
+        self.control_lora.prepare_grad_etc(sd.text_encoder, sd.unet)
+        if self.train_config.gradient_checkpointing:
+            self.control_lora.enable_gradient_checkpointing()
+            
+        self.x_embedder = ImgEmbedder.from_model(sd.unet, self)
+        self.x_embedder.to(self.device_torch)
+
+    def get_params(self):
+        # LyCORIS doesnt have default_lr
+        config = {
+            'text_encoder_lr': self.train_config.lr,
+            'unet_lr': self.train_config.lr,
+        }
+        sig = inspect.signature(self.control_lora.prepare_optimizer_params)
+        if 'default_lr' in sig.parameters:
+            config['default_lr'] = self.train_config.lr
+        if 'learning_rate' in sig.parameters:
+            config['learning_rate'] = self.train_config.lr
+        params_net = self.control_lora.prepare_optimizer_params(
+            **config
+        )
+        
+        # we want only tensors here
+        params = []
+        for p in params_net:
+            if isinstance(p, dict):
+                params += p["params"]
+            elif isinstance(p, torch.Tensor):
+                params.append(p)
+            elif isinstance(p, list):
+                params += p
+        
+        params += list(self.x_embedder.parameters())
+        
+        # we need to be able to yield from the list like yield from params
+
+        return params
+    
+    def load_weights(self, state_dict, strict=True):
+        lora_sd = {}
+        img_embedder_sd = {}
+        for key, value in state_dict.items():
+            if "x_embedder" in key:
+                new_key = key.replace("transformer.x_embedder.", "")
+                img_embedder_sd[new_key] = value
+            else:
+                lora_sd[key] = value
+        
+        # todo process state dict before loading
+        self.control_lora.load_weights(lora_sd)
+        self.x_embedder.load_state_dict(img_embedder_sd, strict=strict)
+        
+    def get_state_dict(self):
+        lora_sd = self.control_lora.get_state_dict(dtype=torch.float32)
+        # todo make sure we match loras elseware. 
+        img_embedder_sd = self.x_embedder.state_dict()
+        for key, value in img_embedder_sd.items():
+            lora_sd[f"transformer.x_embedder.{key}"] = value
+        return lora_sd
+    
+    @property
+    def is_active(self):
+        return self.adapter_ref().is_active