Update app.py

app.py

@@ -85,9 +85,9 @@ pipe = StableDiffusion3Pipeline.from_pretrained(
     #torch_dtype=torch.bfloat16,
     #use_safetensors=False,
 )
-text_encoder=CLIPTextModelWithProjection.from_pretrained("ford442/stable-diffusion-3.5-large-bf16", subfolder='text_encoder', token=True).to(device=device, dtype=torch.bfloat16)
-text_encoder_2=CLIPTextModelWithProjection.from_pretrained("ford442/stable-diffusion-3.5-large-bf16", subfolder='text_encoder_2',token=True).to(device=device, dtype=torch.bfloat16)
-text_encoder_3=T5EncoderModel.from_pretrained("ford442/stable-diffusion-3.5-large-bf16", subfolder='text_encoder_3',token=True).to(device=device, dtype=torch.bfloat16)
+text_encoder=CLIPTextModelWithProjection.from_pretrained("ford442/stable-diffusion-3.5-large-bf16", subfolder='text_encoder', token=True).to(torch.device("cuda:0", dtype=torch.bfloat16)
+text_encoder_2=CLIPTextModelWithProjection.from_pretrained("ford442/stable-diffusion-3.5-large-bf16", subfolder='text_encoder_2',token=True).to(torch.device("cuda:0", dtype=torch.bfloat16)
+text_encoder_3=T5EncoderModel.from_pretrained("ford442/stable-diffusion-3.5-large-bf16", subfolder='text_encoder_3',token=True).to(torch.device("cuda:0", dtype=torch.bfloat16)
     
 pipe.load_lora_weights("ford442/sdxl-vae-bf16", weight_name="LoRA/UltraReal.safetensors")
 
@@ -118,17 +118,11 @@ def infer_30(
     torch.set_float32_matmul_precision("highest")
     seed = random.randint(0, MAX_SEED)
     generator = torch.Generator(device='cuda').manual_seed(seed)
-    input_ids = pipe.tokenizer(prompt, return_tensors="pt").input_ids
-    max_length = 77
-    if input_ids.shape[1] > max_length:
-        input_ids = input_ids[:, :max_length]
-    input_ids = input_ids.to(device)
     print('-- generating image --')
     sd_image = pipe(
-            prompt_embeds = pipe.text_encoder(input_ids)[0], #ensure that the input_ids are on the correct device.
-            #prompt=prompt,
-            #prompt_2=prompt,
-            #prompt_3=prompt,
+            prompt=prompt,
+            prompt_2=prompt,
+            prompt_3=prompt,
             negative_prompt=negative_prompt_1,
             negative_prompt_2=negative_prompt_2,
             negative_prompt_3=negative_prompt_3,
@@ -174,17 +168,11 @@ def infer_60(
     torch.set_float32_matmul_precision("highest")
     seed = random.randint(0, MAX_SEED)
     generator = torch.Generator(device='cuda').manual_seed(seed)
-    input_ids = pipe.tokenizer(prompt, return_tensors="pt").input_ids
-    max_length = 77
-    if input_ids.shape[1] > max_length:
-        input_ids = input_ids[:, :max_length]
-    input_ids = input_ids.to(device)
     print('-- generating image --')
     sd_image = pipe(
-            prompt_embeds = pipe.text_encoder(input_ids)[0], #ensure that the input_ids are on the correct device.
-            #prompt=prompt,
-            #prompt_2=prompt,
-            #prompt_3=prompt,
+            prompt=prompt,
+            prompt_2=prompt,
+            prompt_3=prompt,
             negative_prompt=negative_prompt_1,
             negative_prompt_2=negative_prompt_2,
             negative_prompt_3=negative_prompt_3,
@@ -230,17 +218,11 @@ def infer_90(
     torch.set_float32_matmul_precision("highest")
     seed = random.randint(0, MAX_SEED)
     generator = torch.Generator(device='cuda').manual_seed(seed)
-    input_ids = pipe.tokenizer(prompt, return_tensors="pt").input_ids
-    max_length = 77
-    if input_ids.shape[1] > max_length:
-        input_ids = input_ids[:, :max_length]
-    input_ids = input_ids.to(device)
     print('-- generating image --')
     sd_image = pipe(
-            prompt_embeds = pipe.text_encoder(input_ids)[0], #ensure that the input_ids are on the correct device.
-            #prompt=prompt,
-            #prompt_2=prompt,
-            #prompt_3=prompt,
+            prompt=prompt,
+            prompt_2=prompt,
+            prompt_3=prompt,
             negative_prompt=negative_prompt_1,
             negative_prompt_2=negative_prompt_2,
             negative_prompt_3=negative_prompt_3,