use base and new models

app.py

@@ -1,8 +1,20 @@
 import gradio as gr
 
-from transformers import pipeline
+from transformers import AutoModelForCausalLM, pipeline
 
-pipe = pipeline("translation", model="emya/vicuna-7b-v1.5-steve-jobs-8bit-v1")
+model_name = "lmsys/vicuna-7b-v1.5"
+base_model = AutoModelForCausalLM.from_pretrained(
+    model_name,
+    low_cpu_mem_usage=True,
+    return_dict=True,
+    #torch_dtype=torch.float16,
+    device_map="auto",
+    load_in_8bit=True,
+)
+new_model = "emya/vicuna-7b-v1.5-steve-jobs-8bit-v1"
+model = PeftModel.from_pretrained(base_model, new_model, load_in_8bit=True)
+
+pipe = pipeline("translation", model=model)
 
 def predict(text):
     prompt = f"{text} (Answer in a few sentences)"