Update app.py

app.py

@@ -21,7 +21,6 @@ from transformers import (
     TextIteratorStreamer,
 )
 
-from transformers import Blip2Processor, Blip2ForConditionalGeneration
 from transformers.image_utils import load_image
 
 # Constants for text generation
@@ -49,10 +48,10 @@ model_z = Qwen2_5_VLForConditionalGeneration.from_pretrained(
     torch_dtype=torch.float16
 ).to(device).eval()
 
-# Load blip2-opt-2.7b
-MODEL_ID_K = "Salesforce/blip2-opt-2.7b"
-processor_k = Blip2Processor.from_pretrained(MODEL_ID_K, trust_remote_code=True)
-model_k = Blip2ForConditionalGeneration.from_pretrained(
+# Load coreOCR-7B-050325-preview
+MODEL_ID_K = "prithivMLmods/coreOCR-7B-050325-preview"
+processor_k = AutoProcessor.from_pretrained(MODEL_ID_K, trust_remote_code=True)
+model_k = Qwen2_5_VLForConditionalGeneration.from_pretrained(
     MODEL_ID_K,
     trust_remote_code=True,
     torch_dtype=torch.float16
@@ -95,7 +94,7 @@ def generate_image(model_name: str, text: str, image: Image.Image,
     elif model_name == "SpaceThinker-3B":
         processor = processor_z
         model = model_z
-    elif model_name == "blip2-opt-2.7b":
+    elif model_name == "coreOCR-7B-050325-preview":
         processor = processor_k
         model = model_k
     else:
@@ -148,7 +147,7 @@ def generate_video(model_name: str, text: str, video_path: str,
     elif model_name == "SpaceThinker-3B":
         processor = processor_z
         model = model_z
-    elif model_name == "blip2-opt-2.7b":
+    elif model_name == "coreOCR-7B-050325-preview":
         processor = processor_k
         model = model_k
     else:
@@ -251,7 +250,7 @@ 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=["SkyCaptioner-V1", "SpaceThinker-3B", "blip2-opt-2.7b"],
+                choices=["SkyCaptioner-V1", "SpaceThinker-3B", "coreOCR-7B-050325-preview"],
                 label="Select Model",
                 value="SkyCaptioner-V1"
             )