Update app.py

app.py

@@ -17,6 +17,7 @@ from transformers import (
     Qwen2_5_VLForConditionalGeneration,
     Qwen2VLForConditionalGeneration,
     Gemma3ForConditionalGeneration,
+    LlavaForConditionalGeneration,
     AutoProcessor,
     TextIteratorStreamer,
 )
@@ -48,9 +49,9 @@ model_x = Qwen2VLForConditionalGeneration.from_pretrained(
 ).to(device).eval()
 
 # Load Relaxed
-MODEL_ID_Z = "Ertugrul/Qwen2.5-VL-7B-Captioner-Relaxed"
+MODEL_ID_Z = "fancyfeast/llama-joycaption-beta-one-hf-llava"
 processor_z = AutoProcessor.from_pretrained(MODEL_ID_Z, trust_remote_code=True)
-model_z = Qwen2_5_VLForConditionalGeneration.from_pretrained(
+model_z = LlavaForConditionalGeneration.from_pretrained(
     MODEL_ID_Z,
     trust_remote_code=True,
     torch_dtype=torch.float16
@@ -93,7 +94,7 @@ def generate_image(model_name: str, text: str, image: Image.Image,
     elif model_name == "coreOCR-7B-050325-preview":
         processor = processor_x
         model = model_x
-    elif model_name == "Captioner-7B":
+    elif model_name == "llama-joycaption":
         processor = processor_z
         model = model_z
     else:
@@ -146,7 +147,7 @@ def generate_video(model_name: str, text: str, video_path: str,
     elif model_name == "coreOCR-7B-050325-preview":
         processor = processor_x
         model = model_x
-    elif model_name == "Captioner-7B":
+    elif model_name == "llama-joycaption":
         processor = processor_z
         model = model_z
     else:
@@ -246,7 +247,7 @@ with gr.Blocks(css=css, theme="bethecloud/storj_theme") as demo:
         with gr.Column():
             output = gr.Textbox(label="Output", interactive=False, lines=2, scale=2)
             model_choice = gr.Radio(
-                choices=["gemma-3-4b-it", "coreOCR-7B-050325-preview", "Captioner-7B"],
+                choices=["gemma-3-4b-it", "coreOCR-7B-050325-preview", "llama-joycaption"],
                 label="Select Model",
                 value="gemma-3-4b-it"
             )