Update app.py

app.py

@@ -333,7 +333,6 @@ def generate_video(model_name: str, text: str, video_path: str,
         time.sleep(0.01)
         yield buffer, buffer
 
-# --- Hàm generate_pdf (MỚI - Từ Script 1 và ĐÃ CHỈNH SỬA) ---
 @spaces.GPU
 def generate_pdf(model_name: str, text: str, state: Dict[str, Any], 
                  max_new_tokens: int = 2048, 
@@ -403,20 +402,20 @@ def generate_pdf(model_name: str, text: str, state: Dict[str, Any],
 
 # --- Định nghĩa Examples (Kết hợp từ 2 script) ---
 image_examples = [
-    ["Explain the content in detail.", "images/D.jpg"],
-    ["Explain the content (ocr).", "images/O.jpg"],
-    ["What is the core meaning of the poem?", "images/S.jpg"],
-    ["Provide a detailed caption for the image.", "images/A.jpg"],
-]
-video_examples = [
-    ["Explain the ad in detail", "videos/1.mp4"],
-    ["Identify the main actions in the video", "videos/2.mp4"],
-]
-# Thêm từ Script 1
-pdf_examples = [
-    ["Extract the content precisely.", "examples/pdfs/doc1.pdf"],
-    ["Analyze and provide a short report.", "examples/pdfs/doc2.pdf"]
+    ["Explain the content in detail.", "images/force.jpg"],
+    ["Explain the content (ocr).", "images/ocr.jpg"],
+    ["Extract the content in the json format", "images/bill.jpg"],
+    ["Choose the right answer .", "images/math.jpg"],
 ]
+# video_examples = [
+#     ["Explain the ad in detail", "videos/1.mp4"],
+#     ["Identify the main actions in the video", "videos/2.mp4"],
+# ]
+# # Thêm từ Script 1
+# pdf_examples = [
+#     ["Extract the content precisely.", "examples/pdfs/doc1.pdf"],
+#     ["Analyze and provide a short report.", "examples/pdfs/doc2.pdf"]
+# ]
 
 css = """
 #main-title h1 {
@@ -427,11 +426,9 @@ css = """
 }
 """
 
-# --- Giao diện Gradio (Từ Script 2, đã thêm Tab PDF) ---
 with gr.Blocks(css=css, theme=steel_blue_theme) as demo:
     
-    # Thêm từ Script 1
-    pdf_state = gr.State(value=get_initial_pdf_state())
+    # pdf_state = gr.State(value=get_initial_pdf_state())
     
     gr.Markdown("# **Qwen3-VL-Outpost**", elem_id="main-title")
     with gr.Row():
@@ -443,27 +440,25 @@ with gr.Blocks(css=css, theme=steel_blue_theme) as demo:
                     image_submit = gr.Button("Submit", variant="primary")
                     gr.Examples(examples=image_examples, inputs=[image_query, image_upload])
                 
-                with gr.TabItem("Video Inference"):
-                    video_query = gr.Textbox(label="Query Input", placeholder="Enter your query here...")
-                    video_upload = gr.Video(label="Upload Video", height=290)
-                    video_submit = gr.Button("Submit", variant="primary")
-                    gr.Examples(examples=video_examples, inputs=[video_query, video_upload])
+                # with gr.TabItem("Video Inference"):
+                #     video_query = gr.Textbox(label="Query Input", placeholder="Enter your query here...")
+                #     video_upload = gr.Video(label="Upload Video", height=290)
+                #     video_submit = gr.Button("Submit", variant="primary")
+                #     gr.Examples(examples=video_examples, inputs=[video_query, video_upload])
                 
-                # --- Tab PDF MỚI (Từ Script 1) ---
-                with gr.TabItem("PDF Inference"):
-                    with gr.Row():
-                        with gr.Column(scale=1):
-                            pdf_query = gr.Textbox(label="Query Input", placeholder="e.g., 'Summarize this document'")
-                            pdf_upload = gr.File(label="Upload PDF", file_types=[".pdf"])
-                            pdf_submit = gr.Button("Submit", variant="primary")
-                        with gr.Column(scale=1):
-                            pdf_preview_img = gr.Image(label="PDF Preview", height=290)
-                            with gr.Row():
-                                prev_page_btn = gr.Button("◀ Previous")
-                                page_info = gr.HTML('<div style="text-align:center;">No file loaded</div>')
-                                next_page_btn = gr.Button("Next ▶")
-                    gr.Examples(examples=pdf_examples, inputs=[pdf_query, pdf_upload])
-                # --- Kết thúc Tab PDF ---
+                # with gr.TabItem("PDF Inference"):
+                #     with gr.Row():
+                #         with gr.Column(scale=1):
+                #             pdf_query = gr.Textbox(label="Query Input", placeholder="e.g., 'Summarize this document'")
+                #             pdf_upload = gr.File(label="Upload PDF", file_types=[".pdf"])
+                #             pdf_submit = gr.Button("Submit", variant="primary")
+                #         with gr.Column(scale=1):
+                #             pdf_preview_img = gr.Image(label="PDF Preview", height=290)
+                #             with gr.Row():
+                #                 prev_page_btn = gr.Button("◀ Previous")
+                #                 page_info = gr.HTML('<div style="text-align:center;">No file loaded</div>')
+                #                 next_page_btn = gr.Button("Next ▶")
+                #     gr.Examples(examples=pdf_examples, inputs=[pdf_query, pdf_upload])
 
             with gr.Accordion("Advanced options", open=False):
                 max_new_tokens = gr.Slider(label="Max new tokens", minimum=1, maximum=MAX_MAX_NEW_TOKENS, step=1, value=DEFAULT_MAX_NEW_TOKENS)
@@ -487,45 +482,44 @@ with gr.Blocks(css=css, theme=steel_blue_theme) as demo:
                 value="Qwen3-VL-4B-Instruct"
             )
 
-    # --- Event Handlers (Đã thêm các sự kiện PDF) ---
     image_submit.click(
         fn=generate_image,
         inputs=[model_choice, image_query, image_upload, max_new_tokens, temperature, top_p, top_k, repetition_penalty],
         outputs=[output, markdown_output]
     )
     
-    video_submit.click(
-        fn=generate_video,
-        inputs=[model_choice, video_query, video_upload, max_new_tokens, temperature, top_p, top_k, repetition_penalty],
-        outputs=[output, markdown_output]
-    )
+    # video_submit.click(
+    #     fn=generate_video,
+    #     inputs=[model_choice, video_query, video_upload, max_new_tokens, temperature, top_p, top_k, repetition_penalty],
+    #     outputs=[output, markdown_output]
+    # )
     
-    # --- Thêm sự kiện cho PDF ---
-    pdf_submit.click(
-        fn=generate_pdf,
-        # Thêm 'model_choice' vào inputs
-        inputs=[model_choice, pdf_query, pdf_state, max_new_tokens, temperature, top_p, top_k, repetition_penalty],
-        outputs=[output, markdown_output]
-    )
+    # # --- Thêm sự kiện cho PDF ---
+    # pdf_submit.click(
+    #     fn=generate_pdf,
+    #     # Thêm 'model_choice' vào inputs
+    #     inputs=[model_choice, pdf_query, pdf_state, max_new_tokens, temperature, top_p, top_k, repetition_penalty],
+    #     outputs=[output, markdown_output]
+    # )
     
-    pdf_upload.change(
-        fn=load_and_preview_pdf, 
-        inputs=[pdf_upload], 
-        outputs=[pdf_preview_img, pdf_state, page_info]
-    )
+    # pdf_upload.change(
+    #     fn=load_and_preview_pdf, 
+    #     inputs=[pdf_upload], 
+    #     outputs=[pdf_preview_img, pdf_state, page_info]
+    # )
     
-    prev_page_btn.click(
-        fn=lambda s: navigate_pdf_page("prev", s), 
-        inputs=[pdf_state], 
-        outputs=[pdf_preview_img, pdf_state, page_info]
-    )
+    # prev_page_btn.click(
+    #     fn=lambda s: navigate_pdf_page("prev", s), 
+    #     inputs=[pdf_state], 
+    #     outputs=[pdf_preview_img, pdf_state, page_info]
+    # )
     
-    next_page_btn.click(
-        fn=lambda s: navigate_pdf_page("next", s), 
-        inputs=[pdf_state], 
-        outputs=[pdf_preview_img, pdf_state, page_info]
-    )
-    # --- Kết thúc thêm sự kiện PDF ---
+    # next_page_btn.click(
+    #     fn=lambda s: navigate_pdf_page("next", s), 
+    #     inputs=[pdf_state], 
+    #     outputs=[pdf_preview_img, pdf_state, page_info]
+    # )
+
 
 if __name__ == "__main__":
     demo.queue(max_size=50).launch(mcp_server=True, ssr_mode=False, show_error=True)