Refactor: Move sd2_clip.py to text_encoders folder.

comfy/sd.py

@@ -17,8 +17,8 @@ from . import diffusers_convert
 from . import model_detection
 
 from . import sd1_clip
-from . import sd2_clip
 from . import sdxl_clip
+import comfy.text_encoders.sd2_clip
 import comfy.text_encoders.sd3_clip
 import comfy.text_encoders.sa_t5
 import comfy.text_encoders.aura_t5
@@ -414,8 +414,8 @@ def load_clip(ckpt_paths, embedding_directory=None, clip_type=CLIPType.STABLE_DI
                 clip_target.clip = sdxl_clip.SDXLRefinerClipModel
                 clip_target.tokenizer = sdxl_clip.SDXLTokenizer
         elif "text_model.encoder.layers.22.mlp.fc1.weight" in clip_data[0]:
-            clip_target.clip = sd2_clip.SD2ClipModel
-            clip_target.tokenizer = sd2_clip.SD2Tokenizer
+            clip_target.clip = comfy.text_encoders.sd2_clip.SD2ClipModel
+            clip_target.tokenizer = comfy.text_encoders.sd2_clip.SD2Tokenizer
         elif "encoder.block.23.layer.1.DenseReluDense.wi_1.weight" in clip_data[0]:
             weight = clip_data[0]["encoder.block.23.layer.1.DenseReluDense.wi_1.weight"]
             dtype_t5 = weight.dtype

comfy/supported_models.py

@@ -3,8 +3,8 @@ from . import model_base
 from . import utils
 
 from . import sd1_clip
-from . import sd2_clip
 from . import sdxl_clip
+import comfy.text_encoders.sd2_clip
 import comfy.text_encoders.sd3_clip
 import comfy.text_encoders.sa_t5
 import comfy.text_encoders.aura_t5
@@ -101,7 +101,7 @@ class SD20(supported_models_base.BASE):
         return state_dict
 
     def clip_target(self, state_dict={}):
-        return supported_models_base.ClipTarget(sd2_clip.SD2Tokenizer, sd2_clip.SD2ClipModel)
+        return supported_models_base.ClipTarget(comfy.text_encoders.sd2_clip.SD2Tokenizer, comfy.text_encoders.sd2_clip.SD2ClipModel)
 
 class SD21UnclipL(SD20):
     unet_config = {