Update app.py

app.py

@@ -10,6 +10,7 @@ patient = pipeline("text-generation", model="HuggingFaceTB/SmolLM2-1.7B-Instruct
 patient_system_prompt = "You are a patient describing your symptoms to a physician."
 physician_system_prompt = "You are a physician responding to a patient's symptoms."
 
+
 def generate_conversation(topic, turns):
     conversation = []
     total_tokens = 0
@@ -20,8 +21,8 @@ def generate_conversation(topic, turns):
     patient_input = f"Patient: I'm here to talk about {topic}."
     print(f"Patient Initial Input: {patient_input}")  # Debugging
     patient_response = patient(
-        patient_input,  # Only the patient's message, without the system prompt
-        max_length=50,  # Reduce max_length for faster responses
+        patient_input,
+        max_new_tokens=50,  # Allow the model to generate up to 50 new tokens
         num_return_sequences=1,
         truncation=True,  # Explicitly enable truncation
         do_sample=True,  # Enable sampling
@@ -35,8 +36,8 @@ def generate_conversation(topic, turns):
         # Physician's turn
         print(f"Physician Turn {turn} Prompt: {patient_response}")  # Debugging
         physician_response = physician(
-            f"Physician: {patient_response}",  # Only the physician's response to the patient's message
-            max_length=50,  # Reduce max_length for faster responses
+            f"Physician: {patient_response}",
+            max_new_tokens=50,  # Allow the model to generate up to 50 new tokens
             num_return_sequences=1,
             truncation=True,  # Explicitly enable truncation
             do_sample=True,  # Enable sampling
@@ -49,8 +50,8 @@ def generate_conversation(topic, turns):
         # Patient's turn
         print(f"Patient Turn {turn} Prompt: {physician_response}")  # Debugging
         patient_response = patient(
-            f"Patient: {physician_response}",  # Only the patient's response to the physician's message
-            max_length=50,  # Reduce max_length for faster responses
+            f"Patient: {physician_response}",
+            max_new_tokens=50,  # Allow the model to generate up to 50 new tokens
             num_return_sequences=1,
             truncation=True,  # Explicitly enable truncation
             do_sample=True,  # Enable sampling