Update app_v2.py

app_v2.py

@@ -6,7 +6,7 @@ from diffusers.hooks import apply_group_offloading
 from diffusers import FluxControlNetModel, FluxControlNetPipeline, AutoencoderKL
 from diffusers import BitsAndBytesConfig as DiffusersBitsAndBytesConfig
 from transformers import T5EncoderModel
-from transformers import LlavaForConditionalGeneration, TextIteratorStreamer, AutoProcessor
+from transformers import LlavaForConditionalGeneration, TextIteratorStreamer, AutoProcessor, AutoTokenizer
 from transformers import BitsAndBytesConfig as TransformersBitsAndBytesConfig
 from liger_kernel.transformers import apply_liger_kernel_to_llama
 from PIL import Image
@@ -95,6 +95,7 @@ def generate_image(prompt, scale, steps, control_image, controlnet_conditioning_
     h = h - h % 32
     control_image = control_image.resize((int(w * scale), int(h * scale)), resample=2)  # Resample.BILINEAR
     print("Size to: " + str(control_image.size[0]) + ", " + str(control_image.size[1]))
+    print("Cond Prompt: " + str(prompt))
     with torch.inference_mode():
         image = pipe(
             generator=generator,
@@ -178,7 +179,7 @@ with gr.Blocks(title="FLUX Turbo Upscaler", fill_height=True) as iface:
             caption_button = gr.Button("Generate Caption", variant="secondary")
         with gr.Column(scale=1):
             seed = gr.Slider(0, MAX_SEED, value=42, label="Seed", step=1)
-            steps = gr.Slider(2, 16, value=8, label="Steps")
+            steps = gr.Slider(2, 16, value=8, label="Steps", step=1)
             controlnet_conditioning_scale = gr.Slider(0, 1, value=0.6, label="ControlNet Scale")
             guidance_scale = gr.Slider(1, 30, value=3.5, label="Guidance Scale")
             guidance_end = gr.Slider(0, 1, value=1.0, label="Guidance End")