Update app.py

app.py

@@ -34,9 +34,7 @@ def load_model_with_retries(model_name, max_retries=3, retry_delay=5):
                 model_name, 
                 use_fast=False,  # Sometimes the fast tokenizer causes issues
                 local_files_only=False,
-                token=os.environ.get("HF_TOKEN", None),  # Use token if available
-                trust_remote_code=True,
-                timeout=600  # 10 minutes timeout
+                token=os.environ.get("HF_TOKEN", None)  # Use token if available
             )
             
             model = AutoModelForCausalLM.from_pretrained(
@@ -44,7 +42,6 @@ def load_model_with_retries(model_name, max_retries=3, retry_delay=5):
                 local_files_only=False, 
                 token=os.environ.get("HF_TOKEN", None),
                 trust_remote_code=True,
-                timeout=600,  # 10 minutes timeout
                 low_cpu_mem_usage=True,  # Help with memory issues
                 torch_dtype="auto"  # Use appropriate dtype
             )
@@ -170,16 +167,23 @@ with gr.Blocks(title="Burmese-GPT-v3 Text Generation") as demo:
     )
     
     # Add examples if available
-    gr.Examples(
-        examples=[
+    with gr.Accordion("Examples", open=False):
+        gr.Markdown("Click on any example to try it:")
+        example_prompts = [
             ["ဟုတ်ကဲ့ ကျွန်တော် ဗမာစကား ပြောတတ်ပါတယ်။", 150, 0.7],
             ["မြန်မာနိုင်ငံမှာ", 200, 0.8],
-        ],
-        inputs=[prompt, max_length, temperature],
-        outputs=output,
-        fn=generate_text,
-        cache_examples=True,
-    )
+        ]
+        for idx, example in enumerate(example_prompts):
+            example_btn = gr.Button(f"Example {idx+1}: {example[0][:20]}...")
+            example_btn.click(
+                lambda e=example: (e[0], e[1], e[2]), 
+                inputs=[], 
+                outputs=[prompt, max_length, temperature]
+            ).then(
+                fn=generate_text,
+                inputs=[prompt, max_length, temperature],
+                outputs=output
+            )
     
     # Add troubleshooting section
     gr.Markdown("### Troubleshooting")
@@ -189,11 +193,9 @@ with gr.Blocks(title="Burmese-GPT-v3 Text Generation") as demo:
     - Make sure you have a stable internet connection
     """)
 
-# Launch the app with caching and concurrency settings
+# Launch the app with appropriate settings
 demo.launch(
-    cache_examples=True,
     show_error=True,
     server_name="0.0.0.0",  # Listen on all network interfaces
-    share=False,  # Set to True for temporary public link
-    max_threads=16  # Increase if needed
+    share=False  # Set to True for temporary public link
 )