Create main.py

main.py

@@ -0,0 +1,51 @@
+import os
+from dotenv import load_dotenv
+from fastapi import FastAPI, Request, HTTPException
+from langchain.document_loaders import WebBaseLoader
+from langchain.chains.summarize import load_summarize_chain
+from langchain import HuggingFaceHub
+from huggingface_hub import InferenceClient
+
+load_dotenv()
+hf_token = os.environ.get('HUGGINGFACEHUB_API_TOKEN')
+#starchat_repo_id = os.environ.get('starchat_repo_id')
+#repo_id=os.environ.get('repo_id')
+#port = os.getenv('port')
+
+llm = HuggingFaceHub(repo_id=starchat_repo_id,  # for StarChat
+                     huggingfacehub_api_token=hf_token,
+                     model_kwargs={"min_length": 512,  # for StarChat
+                                   "max_new_tokens": 1024, "do_sample": True,  # for StarChat
+                                   "temperature": 0.01,
+                                   "top_k": 50,
+                                   "top_p": 0.95, "eos_token_id": 49155})
+chain = load_summarize_chain(llm, chain_type="stuff")
+
+app = FastAPI()
+
+@app.post('/')
+async def home_api(request: Request):
+    data = await request.json()
+    user_query = data['user_question']
+    print(user_query)
+    return {"Message": "FastAPI Home API Deploy Success on HF"}
+
+@app.post('/api/chat')
+async def chat(request: Request):
+    data = await request.json()
+    user_query = data['user_question']
+    print(user_query)
+    if user_query.strip() != "":
+        try:
+            loader = WebBaseLoader(user_query)
+            print(user_query)
+            docs = loader.load()
+            result = chain.run(docs)
+            print("AI Summarization: " + result)
+            return {'response': result}
+        except Exception as e:
+            err_msg = "Wrong URL or URL not parsable."
+            print(err_msg)
+            raise HTTPException(status_code=400, detail=err_msg)
+    else:
+        raise HTTPException(status_code=400, detail="Invalid user question")