Update app.py

app.py

@@ -7,132 +7,134 @@ import datetime
 import json
 
 _ = load_dotenv(find_dotenv())
-st.set_page_config(page_icon="", layout="wide", page_title="...")
+st.set_page_config(page_icon="💬", layout="wide", page_title="...")
 
 def icon(emoji: str):
-  """Shows an emoji as a Notion-style page icon."""
-  st.write(
-    f'<span style="font-size: 78px; line-height: 1">{emoji}</span>',
-    unsafe_allow_html=True,
-  )
-
+    """Shows an emoji as a Notion-style page icon."""
+    st.write(
+        f'<span style="font-size: 78px; line-height: 1">{emoji}</span>',
+        unsafe_allow_html=True,
+    )
 
 icon("⚡")
 
 st.subheader("Chatbot", divider="rainbow", anchor=False)
 
 client = Groq(
-  api_key=os.environ['GROQ_API_KEY'],
+    api_key=os.environ['GROQ_API_KEY'],
 )
 
-# Read saved prompts from file
-with open("saved_prompts.txt", "r") as f:
-  saved_prompts = f.read().split("<|>")
-
-prompt_names = [p.split(" ", 1)[0] for p in saved_prompts]
-prompt_map = {name: prompt for name, prompt in zip(prompt_names, saved_prompts)}
-
 # Initialize chat history and selected model
 if "messages" not in st.session_state:
-  st.session_state.messages = []
+    st.session_state.messages = []
 
 if "selected_model" not in st.session_state:
-  st.session_state.selected_model = None
+    st.session_state.selected_model = None
+
+# prompts
+prompts = {
+    "none": "",
+    "python interpreter": "emulate the output of this program like you are the python interpreter, only answer with the result of this emulation. Ask the user for each missing input, sequentially and only once per message, in the same way a python interpreter would. Do not fill in for my inputs. Take my inputs from the message directly after you ask for input."
+}
 
 # Define model details
 models = {
-  "mixtral-8x7b-32768": {
-    "name": "Mixtral-8x7b-Instruct-v0.1",
-    "tokens": 32768,
-    "developer": "Mistral",
-  },
-  "gemma-7b-it": {"name": "Gemma-7b-it", "tokens": 8192, "developer": "Google"},
-  "llama2-70b-4096": {"name": "LLaMA2-70b-chat", "tokens": 4096, "developer": "Meta"},
-  "llama3-70b-8192": {"name": "LLaMA3-70b-8192", "tokens": 8192, "developer": "Meta"},
-  "llama3-8b-8192": {"name": "LLaMA3-8b-8192", "tokens": 8192, "developer": "Meta"},
+    "mixtral-8x7b-32768": {
+        "name": "Mixtral-8x7b-Instruct-v0.1",
+        "tokens": 32768,
+        "developer": "Mistral",
+    },
+    "gemma-7b-it": {"name": "Gemma-7b-it", "tokens": 8192, "developer": "Google"},
+    "llama2-70b-4096": {"name": "LLaMA2-70b-chat", "tokens": 4096, "developer": "Meta"},
+    "llama3-70b-8192": {"name": "LLaMA3-70b-8192", "tokens": 8192, "developer": "Meta"},
+    "llama3-8b-8192": {"name": "LLaMA3-8b-8192", "tokens": 8192, "developer": "Meta"},
 }
 
 # Layout for model selection and max_tokens slider
-col1, col2 = st.columns(2)
+col1, col2, col3 = st.columns(3)
 
 with col1:
-  def update_prompt(selected_prompt):  # Callback function for dropdown
-    global prompt
-    prompt = prompt_map[selected_prompt]
-
-  prompt_option = st.selectbox(
-    "Choose a prompt:",
-    options=list(models.keys()),
-    format_func=lambda x: models[x]["name"],
-    index=0,  # Default to the first model in the list
-    on_change=update_prompt,  # Call update_prompt on selection change
-  )
-
-# Chat input without value argument
-if prompt := st.chat_input("Enter your prompt here..."):
-  st.session_state.messages.append({"role": "user", "content": prompt})
-
-  with st.chat_message("user", avatar="❓"):
-    st.markdown(prompt)
+    model_option = st.selectbox(
+        "Choose a model:",
+        options=list(models.keys()),
+        format_func=lambda x: models[x]["name"],
+        index=0,  # Default to the first model in the list
+    )
 
 # Detect model change and clear chat history if model has changed
 if st.session_state.selected_model != model_option:
-  st.session_state.messages = []
-  st.session_state.selected_model = model_option
+    st.session_state.messages = []
+    st.session_state.selected_model = model_option
 
 max_tokens_range = models[model_option]["tokens"]
 
 with col2:
-  # Adjust max_tokens slider dynamically based on the selected model
-  max_tokens = st.slider(
-    "Max Tokens:",
-    min_value=512,  # Minimum value to allow some flexibility
-    max_value=max_tokens_range,
-    # Default value or max allowed if less
-    value=min(32768, max_tokens_range),
-    step=512,
-    help=f"Adjust the maximum number of tokens (words) for the model's response. Max for selected model: {max_tokens_range}",
-  )
+    # Adjust max_tokens slider dynamically based on the selected model
+    max_tokens = st.slider(
+        "Max Tokens:",
+        min_value=512,  # Minimum value to allow some flexibility
+        max_value=max_tokens_range,
+        # Default value or max allowed if less
+        value=min(32768, max_tokens_range),
+        step=512,
+        help=f"Adjust the maximum number of tokens (words) for the model's response. Max for selected model: {max_tokens_range}",
+    )
+    
+with col3:
+   prompt_selection = st.selectbox(
+       "Choose a prompt:",
+       options=list(prompts.keys()),
+       format_func=lambda x: prompts[x]["name"],
+       index=0,
+   )
 
 # Display chat messages from history on app rerun
 for message in st.session_state.messages:
-  avatar = "" if message["role"] == "assistant" else "❓"
-  with st.chat_message(message["role"], avatar=avatar):
-    st.markdown(message["content"])
+    avatar = "🧠" if message["role"] == "assistant" else "❓"
+    with st.chat_message(message["role"], avatar=avatar):
+        st.markdown(message["content"])
 
 def generate_chat_responses(chat_completion) -> Generator[str, None, None]:
-  """Yield chat response content from the Groq API response."""
-  for chunk in chat_completion:
-    if chunk.choices[0].delta.content:
-      yield chunk.choices[0].delta.content
-
-# Fetch response from Groq API
-try:
-  chat_completion = client.chat.completions.create(
-    model=model_option,
-    messages=[
-      {"role": m["role"], "content": m["content"]}
-      for m in st.session_state.messages
-    ],
-    max_tokens=max_tokens,
-    stream=True,
-  )
-
-  # Use the generator function with st.write_stream
-  with st.chat_message("assistant", avatar=""):
-    chat_responses_generator = generate_chat_responses(chat_completion)
-    full_response = st.write_stream(chat_responses_generator)
-except Exception as e:
-  st.error(e, icon="")
-
-# Append the full response to session_state.messages
-if isinstance(full_response, str):
-  st.session_state.messages.append(
-    {"role": "assistant", "content": full_response}
-  )
-else:
-  # Handle the case where full_response is not a string
-  combined_response = "\n".join(str(item) for item in full_response)
-  st.session_state.messages.append(
-    {"role": "assistant", "content": combined_response}
-  )
+    """Yield chat response content from the Groq API response."""
+    for chunk in chat_completion:
+        if chunk.choices[0].delta.content:
+            yield chunk.choices[0].delta.content
+
+selected_prompt_text = prompts.get(prompt_selection)
+
+if prompt := st.chat_input("Enter your prompt here...") or prompt := selected_prompt_text:
+    st.session_state.messages.append({"role": "user", "content": prompt})
+
+    with st.chat_message("user", avatar="❓"):  
+        st.markdown(prompt)
+
+    # Fetch response from Groq API
+    try:
+        chat_completion = client.chat.completions.create(
+            model=model_option,
+            messages=[
+                {"role": m["role"], "content": m["content"]}
+                for m in st.session_state.messages
+            ],
+            max_tokens=max_tokens,
+            stream=True,
+        )
+
+        # Use the generator function with st.write_stream
+        with st.chat_message("assistant", avatar="🧠"):
+            chat_responses_generator = generate_chat_responses(chat_completion)
+            full_response = st.write_stream(chat_responses_generator)
+    except Exception as e:
+        st.error(e, icon="🚨")
+
+    # Append the full response to session_state.messages
+    if isinstance(full_response, str):
+        st.session_state.messages.append(
+            {"role": "assistant", "content": full_response}
+        )
+    else:
+        # Handle the case where full_response is not a string
+        combined_response = "\n".join(str(item) for item in full_response)
+        st.session_state.messages.append(
+            {"role": "assistant", "content": combined_response}
+        )