flux controlnet bug fix (#64)

Author: tenderness-git

diffsynth_engine/models/flux/flux_dit.py

@@ -411,7 +411,7 @@ def forward(
                     interval_control = int(np.ceil(interval_control))
                     hidden_states = hidden_states + controlnet_double_block_output[i // interval_control]
             hidden_states = torch.cat([prompt_emb, hidden_states], dim=1)
-            for block in self.single_blocks:
+            for i, block in enumerate(self.single_blocks):
                 hidden_states = block(hidden_states, conditioning, rope_emb, image_emb)
                 if controlnet_single_block_output is not None:
                     interval_control = len(self.single_blocks) / len(controlnet_double_block_output)

diffsynth_engine/models/flux/flux_ipadapter.py

@@ -5,7 +5,7 @@
 from typing import Dict, List
 from functools import partial
 from diffsynth_engine.models.utils import no_init_weights
-from diffsynth_engine.models.components.siglip import SiglipImageEncoder
+from diffsynth_engine.models.text_encoder.siglip import SiglipImageEncoder
 from diffsynth_engine.models.basic.transformer_helper import RMSNorm
 from diffsynth_engine.models.basic.attention import attention
 from diffsynth_engine.models.base import PreTrainedModel

diffsynth_engine/models/flux/flux_redux.py

@@ -5,7 +5,7 @@
 from diffsynth_engine.utils.download import fetch_model
 from diffsynth_engine.models.base import PreTrainedModel
 from diffsynth_engine.models.utils import no_init_weights
-from diffsynth_engine.models.components.siglip import SiglipImageEncoder
+from diffsynth_engine.models.text_encoder.siglip import SiglipImageEncoder
 
 
 class FluxReduxImageEmbedder(nn.Module):

diffsynth_engine/models/flux/flux_text_encoder.py

@@ -3,7 +3,7 @@
 from typing import Dict
 
 from diffsynth_engine.models.sd import SDTextEncoder
-from diffsynth_engine.models.components.t5 import T5EncoderModel
+from diffsynth_engine.models.text_encoder.t5 import T5EncoderModel
 from diffsynth_engine.models.base import StateDictConverter
 from diffsynth_engine.models.utils import no_init_weights
 from diffsynth_engine.utils.constants import FLUX_TEXT_ENCODER_CONFIG_FILE

diffsynth_engine/models/flux/flux_vae.py

@@ -2,7 +2,7 @@
 import torch
 from typing import Dict
 
-from diffsynth_engine.models.components.vae import VAEDecoder, VAEEncoder, VAEStateDictConverter
+from diffsynth_engine.models.vae import VAEDecoder, VAEEncoder, VAEStateDictConverter
 from diffsynth_engine.models.utils import no_init_weights
 from diffsynth_engine.utils.constants import FLUX_VAE_CONFIG_FILE
 from diffsynth_engine.utils import logging

diffsynth_engine/models/sd/sd_text_encoder.py

@@ -3,7 +3,7 @@
 import torch.nn as nn
 from typing import Dict
 
-from diffsynth_engine.models.components.clip import CLIPEncoderLayer
+from diffsynth_engine.models.text_encoder.clip import CLIPEncoderLayer
 from diffsynth_engine.models.base import PreTrainedModel, StateDictConverter
 from diffsynth_engine.models.utils import no_init_weights
 from diffsynth_engine.utils.constants import SD_TEXT_ENCODER_CONFIG_FILE

diffsynth_engine/models/sd/sd_vae.py

@@ -1,7 +1,7 @@
 import torch
 from typing import Dict
 
-from diffsynth_engine.models.components.vae import VAEDecoder, VAEEncoder
+from diffsynth_engine.models.vae import VAEDecoder, VAEEncoder
 from diffsynth_engine.models.utils import no_init_weights
 
 

diffsynth_engine/models/sd3/sd3_text_encoder.py

@@ -2,7 +2,7 @@
 import torch
 from typing import Dict
 
-from diffsynth_engine.models.components.t5 import T5EncoderModel
+from diffsynth_engine.models.text_encoder.t5 import T5EncoderModel
 from diffsynth_engine.models.base import StateDictConverter
 from diffsynth_engine.models.sd import SDTextEncoder
 from diffsynth_engine.models.sdxl import SDXLTextEncoder2

diffsynth_engine/models/sd3/sd3_vae.py

@@ -1,7 +1,7 @@
 import torch
 from typing import Dict
 
-from diffsynth_engine.models.components.vae import VAEDecoder, VAEEncoder
+from diffsynth_engine.models.vae import VAEDecoder, VAEEncoder
 from diffsynth_engine.models.utils import no_init_weights
 
 

diffsynth_engine/models/sdxl/sdxl_text_encoder.py

@@ -3,7 +3,7 @@
 import torch.nn as nn
 from typing import Dict
 
-from diffsynth_engine.models.components.clip import CLIPEncoderLayer
+from diffsynth_engine.models.text_encoder.clip import CLIPEncoderLayer
 from diffsynth_engine.models.base import PreTrainedModel, StateDictConverter, split_suffix
 from diffsynth_engine.models.utils import no_init_weights
 from diffsynth_engine.utils.constants import SDXL_TEXT_ENCODER_CONFIG_FILE