Update app.py

app.py

@@ -15,7 +15,6 @@ import cv2
 
 from transformers import (
     Qwen2_5_VLForConditionalGeneration,
-    Qwen2VLForConditionalGeneration,
     AutoProcessor,
     TextIteratorStreamer,
 )
@@ -46,8 +45,8 @@ model_x = Qwen2_5_VLForConditionalGeneration.from_pretrained(
     torch_dtype=torch.float16
 ).to(device).eval()
 
-# Load InkScope Relaxed
-MODEL_ID_Z = "prithivMLmods/Inkscope-Captions-2B-0526"
+# Load Relaxed
+MODEL_ID_Z = "Ertugrul/Qwen2.5-VL-7B-Captioner-Relaxed"
 processor_z = AutoProcessor.from_pretrained(MODEL_ID_Z, trust_remote_code=True)
 model_z = Qwen2_5_VLForConditionalGeneration.from_pretrained(
     MODEL_ID_Z,
@@ -92,7 +91,7 @@ def generate_image(model_name: str, text: str, image: Image.Image,
     elif model_name == "docscopeOCR-7B-050425-exp":
         processor = processor_x
         model = model_x
-    elif model_name == "Captions-Mini":
+    elif model_name == "Relaxed-Captions":
         processor = processor_z
         model = model_z
     else:
@@ -146,7 +145,7 @@ def generate_video(model_name: str, text: str, video_path: str,
     elif model_name == "docscopeOCR-7B-050425-exp":
         processor = processor_x
         model = model_x
-    elif model_name == "Captions-Mini":
+    elif model_name == "Relaxed-Captions":
         processor = processor_z
         model = model_z
     else:
@@ -247,7 +246,7 @@ with gr.Blocks(css=css, theme="bethecloud/storj_theme") as demo:
         with gr.Column():
             output = gr.Textbox(label="Output", interactive=False)
             model_choice = gr.Radio(
-                choices=["Cosmos-Reason1-7B", "docscopeOCR-7B-050425-exp", "Captions-Mini"],
+                choices=["Cosmos-Reason1-7B", "docscopeOCR-7B-050425-exp", "Relaxed-Captions"],
                 label="Select Model",
                 value="Cosmos-Reason1-7B"
             )