time variables edited

app.py

@@ -17,8 +17,10 @@ def generate_response(user_message): #generate_response_token_by_token
 
 
     start_time = time.time()
+    monitor_start_time = time.time()
     alltokens = ""
     token_buffer = ''
+    tokencount = 0
     try:
         while True:
             # Read one character at a time
@@ -30,20 +32,22 @@ def generate_response(user_message): #generate_response_token_by_token
                 if char == ' ' or char == '\n':  # Token delimiters
                     elapsed_time = time.time() - start_time  # Calculate elapsed time
                     alltokens += token_buffer
-                    yield f"{alltokens} \n\n [Inference time: {elapsed_time:.2f} seconds]"
+                    tokencount += 1
+                    yield f"{alltokens} \n\n [Inference time: {elapsed_time:.2f} seconds | Tokens: { tokencount }]"
                     token_buffer = ''  # Reset token buffer
             # Log resource usage every minute
-            if time.time() - start_time > 60:
+            if time.time() - monitor_start_time > 60:
                 cpu_usage = process_monitor.cpu_percent()
                 memory_usage = process_monitor.memory_info().rss  # in bytes 
                 print(f"Subprocess CPU Usage: {cpu_usage}%, Memory Usage: {memory_usage / 1024 ** 2} MB")
-                start_time = time.time()  # Reset the timer
+                monitor_start_time = time.time()  # Reset the timer
     
         # Yield the last token if there is any
         if token_buffer:
             elapsed_time = time.time() - start_time  # Calculate elapsed time
             alltokens += token_buffer
-            yield f"{alltokens} \n\n [Inference time: {elapsed_time:.2f} seconds]"
+            yield f"{alltokens} \n\n [Inference time: {elapsed_time:.2f} seconds | Average Tokens per second: { tokencount / elapsed_time}]"
+
     finally:
         try:
             # Wait for the process to complete, with a timeout