Update app.py

app.py

@@ -8,7 +8,7 @@ from llama_index.core import (
 from llama_index.core.memory import ChatMemoryBuffer
 from llama_index.llms.huggingface import HuggingFaceLLM
 from llama_index.embeddings.langchain import LangchainEmbedding
-from langchain_community.embeddings import HuggingFaceEmbeddings
+from langchain_huggingface import HuggingFaceEmbeddings
 
 # ---------- Constants ----------
 SYSTEM_PROMPT = """
@@ -53,8 +53,6 @@ def get_chat_engine():
             generate_kwargs={"temperature": 0.2, "do_sample": True},
             device_map="auto",
             model_kwargs={
-                "torch_dtype": torch.float16,
-                "load_in_4bit": True,
                 "use_auth_token": os.environ["HF_TOKEN"]
             },
             system_prompt=SYSTEM_PROMPT,