Update app.py

app.py

@@ -6,6 +6,7 @@ import random
 import torch
 from diffusers import StableDiffusion3Pipeline, SD3Transformer2DModel, FlowMatchEulerDiscreteScheduler
 import spaces
+from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
 
 device = "cuda"
 dtype = torch.float16
@@ -13,6 +14,27 @@ dtype = torch.float16
 repo = "stabilityai/stable-diffusion-3-medium"
 pipe = StableDiffusion3Pipeline.from_pretrained(repo, torch_dtype=torch.float16, revision="refs/pr/26",token=os.environ["TOKEN"]).to(device)
 
+model = AutoModelForCausalLM.from_pretrained(
+    model_id,
+    device_map=device, 
+    torch_dtype=torch.bfloat16, 
+    trust_remote_code=True, 
+)
+tokenizer = AutoTokenizer.from_pretrained(model_id)
+
+pipe = pipeline(
+    "text-generation",
+    model=model,
+    tokenizer=tokenizer,
+)
+
+generation_args = {
+    "max_new_tokens": 300,
+    "return_full_text": False,
+    "temperature": 0.7,
+    "do_sample": True,
+}
+
 MAX_SEED = np.iinfo(np.int32).max
 MAX_IMAGE_SIZE = 1344
 
@@ -23,9 +45,19 @@ def infer(prompt, negative_prompt, seed, randomize_seed, width, height, guidance
         seed = random.randint(0, MAX_SEED)
         
     generator = torch.Generator().manual_seed(seed)
-    
+
+    messages = [
+        {"role": "user", "content": "クールなアニメ風の少女"},
+        {"role": "assistant", "content": "An anime style illustration of a cool-looking teenage girl with an edgy, confident expression. She has piercing eyes, a slight smirk, and colorful hair that flows in the wind. She wears a trendy punk-inspired outfit with a leather jacket, ripped jeans, and combat boots. The background has an urban nighttime feel with city lights and graffiti to match her rebellious vibe. The colors are vibrant with high contrast to give an impactful look. The overall style captures her undeniable coolness and fearless attitude."},
+        {"role": "user", "content": "美味しそうな肉"},
+        {"role": "assistant", "content": "A gourmet scene in a high-end restaurant kitchen where a chef is presenting a plate of cooked beef testicles, garnished elegantly with herbs and spices. The chef, a middle-aged Caucasian man wearing a white chef's hat and coat, is inspecting the dish with a satisfied expression. The kitchen background is bustling with other chefs and kitchen staff, and the atmosphere is warm and inviting with hanging pots and pans, and a glowing, busy stove in the background. The focus is on the chef's proud presentation of this unusual but delicately prepared dish."},
+        {"role": "user", "content": prompt},
+    ]
+    output = pipe(messages, **generation_args)
+    upsampled_prompt=output[0]['generated_text']
+
     image = pipe(
-        prompt = prompt, 
+        prompt = upsampled_prompt, 
         negative_prompt = negative_prompt,
         guidance_scale = guidance_scale, 
         num_inference_steps = num_inference_steps,