Update app.py

app.py

@@ -31,41 +31,19 @@ import xml.etree.ElementTree as ET
 import torch
 import mysql.connector
 from mysql.connector import errorcode, pooling
-from dotenv import load_dotenv
-from huggingface_hub import login
-from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
+import nltk
+import importlib
 
+st.title("CEEMEESEEK with Model Selection")
 
-def load_model():  # Define load_model() first
-    """
-    Loads the openLlama model and tokenizer once and returns the pipeline.
-    """
-    try:
-        model_name = "openlm-research/open_llama_3b_v2"
-        tokenizer = AutoTokenizer.from_pretrained(model_name, use_fast=False, legacy=False)
-        model = AutoModelForCausalLM.from_pretrained(model_name)
-
-        max_supported_length = 2048
-
-        openllama_pipeline = pipeline(
-            "text-generation",
-            model=model,
-            tokenizer=tokenizer,
-            truncation=True,
-            max_length=max_supported_length,
-            temperature=0.7,
-            top_p=0.95,
-            device=0 if torch.cuda.is_available() else -1,
-        )
-        logging.info("Model loaded successfully.")
-        return openllama_pipeline
-    except Exception as e:
-        logging.error(f"Error loading google/flan-t5-xl model: {e}")
-        return None
-
-chat_pipeline = load_model()  # Now call load_model()
+model_option = st.selectbox("Select a Model", ["Falcon", "Flan-T5", "Other Model"])  # Add your model names
 
-nlp = AutoTokenizer.from_pretrained("bert-base-uncased")
+if model_option == "Falcon":
+    model_module = importlib.import_module("model_falcon")  # Assuming you create model_falcon.py
+    model = model_module.load_falcon_model() 
+elif model_option == "Flan-T5":
+    model_module = importlib.import_module("model_flan_t5")
+    model = model_module.load_flan_t5_model()
 
 HUGGINGFACE_TOKEN = os.getenv("HUGGINGFACE_TOKEN")