Create app.py

app.py

@@ -0,0 +1,73 @@
+import spaces
+import os
+import gradio as gr
+import json
+import base64
+from huggingface_hub import InferenceClient, login
+
+# Get the API key from environment variables
+key = "UCODE_SECRET"
+login(os.getenv(key))
+
+# Initialize the InferenceClient with the specified model
+client = InferenceClient("meta-llama/Meta-Llama-3-70B-Instruct")
+
+def decode_base64_to_json(base64_str):
+    try:
+        # Decode the base64 string
+        decoded_bytes = base64.b64decode(base64_str)
+        # Convert bytes to string
+        decoded_str = decoded_bytes.decode('utf-8')
+        # Fix escaped characters
+        decoded_str = decoded_str.replace("\\'", "'").replace('\\"', '"').replace('\\\\', '\\')
+        print(f"===================================================\nDecoded string: {decoded_str}\n===================================================")  # Log the decoded string
+        # Parse the JSON string
+        return json.loads(decoded_str)
+    except Exception as e:
+        raise ValueError(f"Error decoding base64 to JSON: {str(e)}")
+
+
+@spaces.GPU(enable_queue=True)
+def chat_completion(user_input, max_tokens, temperature, top_p):
+    try:
+        # Decode the base64-encoded JSON input
+        input_data = decode_base64_to_json(user_input)
+        
+        # Ensure the input is a list of messages
+        if not isinstance(input_data, list):
+            raise ValueError("Input must be a list of messages.")
+        
+        response = ""
+        
+        # Generate chat completion
+        for message in client.chat_completion(
+            input_data,
+            max_tokens=max_tokens,
+            stream=True,
+            temperature=temperature,
+            top_p=top_p,
+        ):
+            token = message.choices[0].delta.get("content", "")
+            response += token
+        
+        return json.dumps({"status": "success", "output": response})
+    except Exception as e:
+        return json.dumps({"status": "error", "message": str(e)})
+
+# Create Gradio components for user input
+user_input = gr.Textbox(label="User Input as Base64-encoded JSON String", lines=10)
+max_tokens = gr.Slider(minimum=1, maximum=8092, value=150, label="Max Tokens")
+temperature = gr.Slider(minimum=0.0, maximum=1.0, value=0.7, label="Temperature")
+top_p = gr.Slider(minimum=0.0, maximum=1.0, value=0.9, label="Top P")
+
+# Set up Gradio interface
+iface = gr.Interface(
+    fn=chat_completion,
+    inputs=[user_input, max_tokens, temperature, top_p],
+    outputs="text",
+    title="UCode Agent",
+    description="Provide Base64-encoded JSON input with a list of messages and set the max tokens, temperature, and top_p to generate a chat completion."
+)
+
+# Launch the Gradio interface
+iface.launch()