full documentation and refactoring.

app.py

@@ -3,45 +3,72 @@ import os
 from huggingface_hub import InferenceClient
 import cohere
 
+# Model & API setup
+COHERE_MODEL = "command-r-plus"
+HF_MODEL = "mistralai/Mistral-7B-Instruct-v0.3"
+
+# Fetch API keys from environment variables
 HF_API_KEY = os.getenv("HF_API_KEY")
-COHERE_API_KEY = os.getenv("COHERE_API_KEY")  # Get Cohere API key
+COHERE_API_KEY = os.getenv("COHERE_API_KEY")
 
-models = ["HuggingFaceH4/zephyr-7b-beta", "meta-llama/Llama-3.2-3B-Instruct", "mistralai/Mistral-7B-Instruct-v0.3"]
-client_hf = InferenceClient(model=models[2], token=HF_API_KEY)  # HF Client
-client_cohere = cohere.Client(COHERE_API_KEY)  # Cohere Client
+# Initialize clients for Hugging Face and Cohere APIs
+client_hf = InferenceClient(model=HF_MODEL, token=HF_API_KEY)
+client_cohere = cohere.Client(COHERE_API_KEY)
 
 def respond(
-    message,
-    history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
-    use_cohere,  # Checkbox value
+    message: str, 
+    history: list[tuple[str, str]], 
+    system_message: str, 
+    max_tokens: int, 
+    temperature: float, 
+    top_p: float, 
+    use_cohere: bool
 ):
+    """Handles chatbot responses based on user input and chat history.
+
+    This function integrates with either the Cohere API or Hugging Face API to generate AI-based responses.
+
+    Args:
+        message (str): The latest user message.
+        history (list[tuple[str, str]]): A list of previous exchanges where:
+            - Each tuple contains (user_message, assistant_response).
+            - Example: [("Hello", "Hi there!"), ("How are you?", "I'm good!")]
+        system_message (str): A system-level instruction for the chatbot (e.g., personality, style).
+        max_tokens (int): Maximum number of new tokens the model can generate.
+        temperature (float): Controls randomness (higher = more varied responses).
+        top_p (float): Probability threshold for token selection (higher = more diverse responses).
+        use_cohere (bool): If True, uses Cohere API; otherwise, uses Hugging Face API.
+
+    Yields:
+        str: The chatbot's response (streamed for Hugging Face, full response for Cohere).
+    """
+    
+    # Constructing the message history for context
     messages = [{"role": "system", "content": system_message}]
     
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+    for user_msg, assistant_msg in history:
+        if user_msg:
+            messages.append({"role": "user", "content": user_msg})
+        if assistant_msg:
+            messages.append({"role": "assistant", "content": assistant_msg})
 
-    messages.append({"role": "user", "content": message})
+    messages.append({"role": "user", "content": message})  # Append current user message
     
     response = ""
 
-    if use_cohere:  # If Cohere is selected
+    if use_cohere:  
+        # Using Cohere API (no streaming support)
         cohere_response = client_cohere.chat(
             message=message,
-            model="command-r",  # Or "command" depending on your plan
+            model=COHERE_MODEL,
             temperature=temperature,
             max_tokens=max_tokens
         )
         response = cohere_response.text
-        yield response  # Yield full response (Cohere doesn't stream)
+        yield response  # Yield full response immediately
     
-    else:  # If HF is selected
+    else:  
+        # Using Hugging Face API (streaming responses)
         for message in client_hf.chat_completion(
             messages,
             max_tokens=max_tokens,
@@ -49,21 +76,22 @@ def respond(
             temperature=temperature,
             top_p=top_p,
         ):
-            token = message.choices[0].delta.content
+            token = message.choices[0].delta.content  # Extract generated token
             response += token
-            yield response
+            yield response  # Yield response incrementally
 
-# Gradio UI
+# Gradio UI with user-configurable inputs
 demo = gr.ChatInterface(
     respond,
     additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
-        gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p"),
-        gr.Checkbox(label="Use Cohere API"),  # Checkbox to switch API
+        gr.Textbox(value="You are a friendly Chatbot.", label="System prompt"),  # System instruction
+        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),  # Token limit
+        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),  # Randomness control
+        gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p"),  # Probability mass
+        gr.Checkbox(label="Use capable Cohere model instead."),  # API selection toggle
     ],
 )
 
+# Start Gradio interface
 if __name__ == "__main__":
     demo.launch()