Update app.py

app.py

@@ -3,14 +3,13 @@ import torch
 import time
 from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
 from threading import Thread
+import time
+import pytz
+from datetime import datetime
 print("Loading model and tokenizer...")
 model_name = "large-traversaal/Phi-4-Hindi"
 tokenizer = AutoTokenizer.from_pretrained(model_name)
-model = AutoModelForCausalLM.from_pretrained(
-    model_name,
-    torch_dtype=torch.bfloat16,
-    device_map="auto"
-)
+model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.bfloat16, device_map="auto")
 print("Model and tokenizer loaded successfully!")
 def generate_response(message, temperature, max_new_tokens, top_p):
     print(f"Input: {message}")
@@ -30,12 +29,21 @@ def generate_response(message, temperature, max_new_tokens, top_p):
     result = []
     for text in streamer:
         result.append(text)
-        yield "".join(result)
+        current_output = "".join(result)
+        if current_output.startswith(message):
+            yield current_output[len(message):]
+        else:
+            yield current_output
     end_time = time.time()
     time_taken = end_time - start_time
     output_text = "".join(result)
+    if output_text.startswith(message):
+        output_text = output_text[len(message):]
     print(f"Output: {output_text}")
     print(f"Time taken: {time_taken:.2f} seconds")
+    pst_timezone = pytz.timezone('America/Los_Angeles')
+    current_time_pst = datetime.now(pst_timezone).strftime("%Y-%m-%d %H:%M:%S %Z%z")
+    print(f"Current timestamp (PST): {current_time_pst}")
 with gr.Blocks() as demo:
     gr.Markdown("# Phi-4-Hindi Demo")
     with gr.Row():