Update app.py

app.py

@@ -1,22 +1,15 @@
-import gradio as gr
-import os
+import streamlit as st
 import requests
 import json
 from huggingface_hub import HfApi
 import huggingface_hub
+import os
 
 # Constants for Zephyr-7b-Beta model
 zephyr_7b_beta = "https://api-inference.huggingface.co/models/HuggingFaceH4/zephyr-7b-beta/"
 HF_TOKEN = os.getenv("HF_TOKEN")
 HEADERS = {"Authorization": f"Bearer {HF_TOKEN}"}
 
-# Constants for Gradio UI
-constants = """
-SYSTEM_PROMPT = "{}"
-TITLE = "{}"
-EXAMPLE_INPUT = "{}"
-"""
-
 # Function to build input prompt for the model
 def build_input_prompt(message, chatbot, system_prompt):
     input_prompt = "\n" + system_prompt + "</s>\n\n"
@@ -31,7 +24,7 @@ def post_request_beta(payload):
     response.raise_for_status()
     return response.json()
 
-# Function to get model predictions
+# Function to get the response from the model
 def predict_beta(message, chatbot=[], system_prompt=""):
     input_prompt = build_input_prompt(message, chatbot, system_prompt)
     data = {"inputs": input_prompt}
@@ -39,100 +32,33 @@ def predict_beta(message, chatbot=[], system_prompt=""):
     try:
         response_data = post_request_beta(data)
         json_obj = response_data[0]
-        
+
         if 'generated_text' in json_obj and len(json_obj['generated_text']) > 0:
             bot_message = json_obj['generated_text']
             return bot_message
         elif 'error' in json_obj:
-            raise gr.Error(json_obj['error'] + ' Please refresh and try again with smaller input prompt')
+            raise st.error(json_obj['error'] + ' Please refresh and try again with a smaller input prompt')
         else:
             warning_msg = f"Unexpected response: {json_obj}"
-            raise gr.Error(warning_msg)
+            raise st.error(warning_msg)
     except requests.HTTPError as e:
         error_msg = f"Request failed with status code {e.response.status_code}"
-        raise gr.Error(error_msg)
+        raise st.error(error_msg)
     except json.JSONDecodeError as e:
         error_msg = f"Failed to decode response as JSON: {str(e)}"
-        raise gr.Error(error_msg)
-
-# Function to extract title, system prompt, and example input from model response
-def extract_title_prompt_example(text, title, system_prompt, example_input):
-    try:
-        text_start = text.rfind("", ) + len("")
-        text = text[text_start:]
-    except ValueError:
-        pass
-    try:
-        title_start = text.lower().rfind("title:") + len("title:")    
-        prompt_start = text.lower().rfind("system prompt:")
-        title = text[title_start:prompt_start].strip()
-    except ValueError:
-        pass
-    try:
-        prompt_start = text.lower().rfind("system prompt:") + len("system prompt:")
-        example_start = text.lower().rfind("example input:")
-        system_prompt = text[prompt_start:example_start].strip()
-    except ValueError:
-        pass
-    try:
-        example_start = text.lower().rfind("example input:") + len("example input:")
-        example_input = text[example_start:].strip()
-        example_input = example_input[:example_input.index("\n")]
-    except ValueError:
-        pass
-    return text, title, system_prompt, example_input
-
-# Function to make an Open GPT
-def make_open_gpt(message, history, current_title, current_system_prompt, current_example_input):
-    response = predict_beta(message, history, current_system_prompt)
-    response, title, system_prompt, example_input = extract_title_prompt_example(response, current_title, current_system_prompt, current_example_input)
-    return "", history + [(message, response)], title, system_prompt, example_input, [(None, welcome_preview_message.format(title, example_input))], example_input, gr.Column(visible=True), gr.Group(visible=True)
-
-# Function to set title and example input for preview
-def set_title_example(title, example):
-    return [(None, welcome_preview_message.format(title, example))], example, gr.Column(visible=True), gr.Group(visible=True)
+        raise st.error(error_msg)
 
-# Function to publish the GPT to Hugging Face Spaces
-def publish(textbox_system_prompt, textbox_title, textbox_example, textbox_token):
-    source_file = 'app_template.py'
-    destination_file = 'app.py'
-    constants_formatted = constants.format(textbox_system_prompt, textbox_title, textbox_example)
-    with open(source_file, 'r') as file:
-        original_content = file.read()
-    with open(destination_file, 'w') as file:
-        file.write(constants_formatted + original_content)
-    title = strip_invalid_filename_characters(textbox_title, max_bytes=30)
-    api = HfApi(token=textbox_token)
-    new_space = api.create_repo(
-        repo_id=f"open-gpt-{title}",
-        repo_type="space",
-        exist_ok=True,
-        private=False,
-        space_sdk="gradio",
-        token=textbox_token,
-    )
-    api.upload_file(
-        repo_id=new_space.repo_id,
-        path_or_fileobj='app.py',
-        path_in_repo='app.py',
-        token=textbox_token,
-        repo_type="space",
-    )
-    api.upload_file(
-        repo_id=new_space.repo_id,
-        path_or_fileobj='README_template.md',
-        path_in_repo='README.md',
-        token=textbox_token,
-        repo_type="space",
-    )
-    huggingface_hub.add_space_secret(
-        new_space.repo_id, "HF_TOKEN", textbox_token, token=textbox_token
-    )
+# Streamlit UI setup
+st.title("GPT Baker with Streamlit")
 
-    return gr.Markdown(f"Published to https://huggingface.co/spaces/{new_space.repo_id} ✅", visible=True), gr.Button("Publish", interactive=True)
+# ... (Add your Streamlit UI components)
 
-# Gradio UI setup
-with gr.Blocks(css=css) as demo:
-    # ... (The rest of your Gradio UI setup)
+# Example Streamlit components:
+user_input = st.text_input("User Input", "Your default example input")
+submit_button = st.button("Submit")
 
-    demo.launch(share=True)
+if submit_button:
+    # Get the model response
+    model_response = predict_beta(user_input)
+    st.write("Model Response:")
+    st.write(model_response)