Update app.py

app.py

@@ -34,7 +34,7 @@ if "follow_up" not in st.session_state:
     st.session_state.follow_up = ""
 
 # ✅ Initialize Hugging Face Model (Explicitly Set to CPU/GPU)
-def get_llm_hf_inference(model_id="meta-llama/Llama-2-7b-chat-hf", max_new_tokens=512, temperature=0.7):
+def get_llm_hf_inference(model_id="meta-llama/Llama-2-7b-chat-hf", max_new_tokens=800, temperature=0.8):
     #mistralai/Mistral-7B-Instruct-v0.3
     return HuggingFaceEndpoint(
         repo_id=model_id,
@@ -91,7 +91,7 @@ def generate_follow_up(user_text):
         "Ensure it's concise and structured exactly as requested without extra commentary."
     )
 
-    hf = get_llm_hf_inference(max_new_tokens=30, temperature=0.6)  # 🔥 Lower temp for consistency
+    hf = get_llm_hf_inference(max_new_tokens=30, temperature=0.8)  # 🔥 Lower temp for consistency
     output = hf.invoke(input=prompt_text).strip()
 
     # ✅ Extract the relevant part using regex to remove unwanted symbols or truncations
@@ -104,7 +104,7 @@ def generate_follow_up(user_text):
     return cleaned_output
 
 # ✅ Main Response Function
-def get_response(system_message, chat_history, user_text, max_new_tokens=512):
+def get_response(system_message, chat_history, user_text, max_new_tokens=800):
     action = predict_action(user_text)
 
     # ✅ Handle NASA-Specific Queries