Updates with onnx model

app.py

@@ -3,6 +3,7 @@ import time
 import streamlit as st
 from streamlit_chat import message
 from transformers import AutoModelForSeq2SeqLM, AutoTokenizer
+from optimum.onnxruntime import ORTModelForSeq2SeqLM
 import textwrap
 
 from chat import generate_response, generate_tag
@@ -49,12 +50,15 @@ db = create_database()
 @st.cache_resource()
 def load_model():
     print("test")
-    tokenizer = AutoTokenizer.from_pretrained(
-        "MBZUAI/LaMini-Flan-T5-783M"
-    )
-    model = AutoModelForSeq2SeqLM.from_pretrained(
-        "MBZUAI/LaMini-Flan-T5-783M"
-    )
+    # tokenizer = AutoTokenizer.from_pretrained(
+    #     "MBZUAI/LaMini-Flan-T5-783M"
+    # )
+    # model = AutoModelForSeq2SeqLM.from_pretrained(
+    #     "MBZUAI/LaMini-Flan-T5-783M"
+    
+    tokenizer = AutoTokenizer.from_pretrained("Xenova/LaMini-Flan-T5-783M")
+    model = ORTModelForSeq2SeqLM.from_pretrained("Xenova/LaMini-Flan-T5-783M", subfolder = "onnx", decoder_file_name="decoder_with_past_model.onnx")
+
     return tokenizer, model
 
 st.title("BGPT : Bibek's Personal Chatbot")