Update app.py

app.py

@@ -3,7 +3,7 @@ import torch
 from sentence_transformers import SentenceTransformer, util
 import gradio as gr
 import json
-from transformers import AutoTokenizer, AutoModelForQuestionAnswering, AutoModelForSequenceClassification
+from transformers import AutoTokenizer, AutoModelForCausalLM, AutoModelForSequenceClassification
 import spaces
 
 # Ensure you have GPU support
@@ -20,11 +20,11 @@ embeddings = torch.tensor(df['embedding'].tolist(), device=device)
 model = SentenceTransformer('all-MiniLM-L6-v2', device=device)
 
 # Load the ai model for response generation
-tokenizer = AutoTokenizer.from_pretrained("distilbert/distilbert-base-uncased-distilled-squad")
-model_response = AutoModelForQuestionAnswering.from_pretrained("distilbert/distilbert-base-uncased-distilled-squad").to(device)
+tokenizer = AutoTokenizer.from_pretrained("openai-community/gpt2-medium")
+model_response = AutoModelForCausalLM.from_pretrained("openai-community/gpt2-medium").to(device)
 
 # Load the NLU model for intent detection
-nlu_model = AutoModelForSequenceClassification.from_pretrained("bert-base-uncased-finetuned-sst-2-english").to(device)
+nlu_model = AutoModelForSequenceClassification.from_pretrained("distilbert/distilbert-base-uncased-finetuned-sst-2-english").to(device)
 
 # Define the function to find the most relevant document
 @spaces.GPU(duration=120)