Create app.py

app.py

@@ -0,0 +1,38 @@
+import gradio as gr
+from gradio_client import Client
+from qdrant_client import QdrantClient
+from load_encoder import encoder 
+from utils import NeuralSearcher
+import spaces
+import os
+
+mytheme = gr.Theme.from_hub("JohnSmith9982/small_and_pretty")
+
+collection_name = os.getenv("QDRANT_COLLECTION")
+QDRANT_API_KEY = os.getenv("QDRANT_API")
+QDRANT_URL = os.getenv("QDRANT_URL")
+qdrant_client = QdrantClient(
+    url=QDRANT_URL,
+    api_key=QDRANT_API_KEY,
+)
+api_client = Client("eswardivi/Phi-3-mini-128k-instruct")
+
+@spaces.GPU(duration=120)
+def reply(message, history):
+    global encoder
+    global api_client
+    global qdrant_client
+    txt2txt = NeuralSearcher(collection_name, qdrant_client, encoder)
+    context = txt2txt.search(message)
+    to_phi = f"Instructions: you are a useful assistant focused on providing valuable content on Climate-related Financial Disclosures, and you should always give coincise and precise answers; Context: {context}; User prompt: {message}"
+    response = api_client.predict(
+        to_phi,	# str  in 'Message' Textbox component
+        0.2,	# float (numeric value between 0 and 1) in 'Temperature' Slider component
+        True,	# bool  in 'Sampling' Checkbox component
+        512,	# float (numeric value between 128 and 4096) in 'Max new tokens' Slider component
+        api_name="/chat"
+    )
+    return response
+
+demo = gr.ChatInterface(fn=reply, title="Climate-related Financial Disclosures Counselor", theme=mytheme)
+demo.launch(server_name="0.0.0.0", share=False)