Update app.py

app.py

@@ -1,30 +1,10 @@
-import streamlit as st
-from huggingface_hub import InferenceClient
-import os
-
-# Constants
-SPACE_URL = "https://z7svds7k42bwhhgm.us-east-1.aws.endpoints.huggingface.cloud"
-HF_API_KEY = os.getenv("HF_API_KEY")
-DUBS_PATH = "🐾"  # Optional: Replace with an avatar path if needed
-
-# Streamlit Configuration
-st.set_page_config(page_title="Chatbot Test", page_icon="🤖", layout="centered")
-
-client = InferenceClient(SPACE_URL, token=HF_API_KEY)
-
-
-
-    
-
-# Function to Fetch Response (Non-Streaming)
-def fetch_response(prompt_text):
+def fetch_response(prompt):
     """
     Fetch full text response from the HF Inference Endpoint using the InferenceClient.
-    Returns the generated text as a whole.
+    Returns tokens in a streaming fashion.
     """
-
-
-        gen_kwargs = {
+    partial_text = ""
+    gen_kwargs = {
         "max_new_tokens": 512,
         "top_k": 30,
         "top_p": 0.9,
@@ -32,37 +12,14 @@ def fetch_response(prompt_text):
         "repetition_penalty": 1.02,
         "stop_sequences": ["<|endoftext|>"]
     }
-        stream = client.text_generation(prompt, stream=True, details=True, **gen_kwargs)
-
-        for response in stream:
-            if response.token.special:
-                continue
-            # Break if we encounter a stop sequence
-            if response.token.text in gen_kwargs["stop_sequences"]:
-                break
-            # Append the new token's text to the partial text
-            partial_text += response.token.text
-            yield response.token.text  # Yield only the new chunk
-
-
-
-
-
-# Streamlit Chat Interface
-st.title("Chatbot Testing Interface")
-
-# User Input Section
-prompt = st.chat_input("Enter your message...")
-
-if prompt:
-    # Display the user's message
-    st.chat_message("user").write(prompt)
 
-    # Build the chat history (use prompt directly for stateless behavior)
-    chat_history = f"<|user|>{prompt}<|end|> \n <|assistant|> "
+    stream = client.text_generation(prompt, stream=True, details=True, **gen_kwargs)
 
-    # Generate the response
-    with st.spinner("Dubs is thinking... Woof Woof! 🐾"):
-        with st.chat_message("assistant", avatar=DUBS_PATH):
-            full_response = fetch_response(chat_history)
-            st.write_stream(full_response)
+    for response in stream:
+        if response.token.special:
+            continue
+        # Stop if we encounter stop_sequences
+        if response.token.text in gen_kwargs["stop_sequences"]:
+            break
+        partial_text += response.token.text
+        yield response.token.text