try fix lora vram (#357)

* try fix lora vram

* try fix lora vram

fooocus_version.py

@@ -1 +1 @@
-version = '2.0.0'
+version = '2.0.1'

modules/core.py

@@ -8,14 +8,13 @@ import comfy.model_management
 import comfy.utils
 
 from comfy.sd import load_checkpoint_guess_config
-from nodes import VAEDecode, EmptyLatentImage, CLIPTextEncode
+from nodes import VAEDecode, EmptyLatentImage
 from comfy.sample import prepare_mask, broadcast_cond, load_additional_models, cleanup_additional_models
 from modules.samplers_advanced import KSampler, KSamplerWithRefiner
 from modules.patch import patch_all
 
 
 patch_all()
-opCLIPTextEncode = CLIPTextEncode()
 opEmptyLatentImage = EmptyLatentImage()
 opVAEDecode = VAEDecode()
 
@@ -52,11 +51,6 @@ def load_lora(model, lora_filename, strength_model=1.0, strength_clip=1.0):
     return StableDiffusionModel(unet=unet, clip=clip, vae=model.vae, clip_vision=model.clip_vision)
 
 
-@torch.no_grad()
-def encode_prompt_condition(clip, prompt):
-    return opCLIPTextEncode.encode(clip=clip, text=prompt)[0]
-
-
 @torch.no_grad()
 def generate_empty_latent(width=1024, height=1024, batch_size=1):
     return opEmptyLatentImage.generate(width=width, height=height, batch_size=batch_size)[0]

modules/default_pipeline.py

@@ -106,6 +106,7 @@ refresh_loras([(modules.path.default_lora_name, 0.5), ('None', 0.5), ('None', 0.
 expansion = FooocusExpansion()
 
 
+@torch.no_grad()
 def clip_encode_single(clip, text, verbose=False):
     cached = clip.fcs_cond_cache.get(text, None)
     if cached is not None:
@@ -120,6 +121,7 @@ def clip_encode_single(clip, text, verbose=False):
     return result
 
 
+@torch.no_grad()
 def clip_encode(sd, texts, pool_top_k=1):
     if sd is None:
         return None
@@ -143,6 +145,7 @@ def clip_encode(sd, texts, pool_top_k=1):
     return [[torch.cat(cond_list, dim=1), {"pooled_output": pooled_acc}]]
 
 
+@torch.no_grad()
 def clear_sd_cond_cache(sd):
     if sd is None:
         return None
@@ -152,6 +155,7 @@ def clear_sd_cond_cache(sd):
     return
 
 
+@torch.no_grad()
 def clear_all_caches():
     clear_sd_cond_cache(xl_base_patched)
     clear_sd_cond_cache(xl_refiner)

modules/patch.py

@@ -3,6 +3,7 @@ import comfy.model_base
 import comfy.ldm.modules.diffusionmodules.openaimodel
 import comfy.samplers
 import comfy.k_diffusion.external
+import comfy.model_management
 import modules.anisotropic as anisotropic
 
 from comfy.k_diffusion import utils
@@ -68,6 +69,11 @@ def sdxl_encode_adm_patched(self, **kwargs):
     return torch.cat((clip_pooled.to(flat.device), flat), dim=1)
 
 
+def text_encoder_device_patched():
+    return torch.device("cpu")
+
+
 def patch_all():
+    # comfy.model_management.text_encoder_device = text_encoder_device_patched
     comfy.k_diffusion.external.DiscreteEpsDDPMDenoiser.forward = patched_discrete_eps_ddpm_denoiser_forward
     comfy.model_base.SDXL.encode_adm = sdxl_encode_adm_patched