Update app.py

app.py

@@ -1,15 +1,14 @@
 from accelerate import Accelerator
-from transformers import pipeline, AutoTokenizer, AutoModelForCausalLM
+from transformers import pipeline, AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig
 import gradio as gr
 import os, random, gc
 import torch
 accelerator=Accelerator(cpu=True)
-mdl = "bigcode/starcoder2-3b"
-tokenizer = AutoTokenizer.from_pretrained(mdl)
-model = accelerator.prepare(AutoModelForCausalLM.from_pretrained(mdl, torch_dtype=torch.bfloat16))
+tokenizer = AutoTokenizer.from_pretrained("bigcode/starcoder2-3b")
+pipe = accelerator.prepare(AutoModelForCausalLM.from_pretrained("bigcode/starcoder2-3b", torch_dtype=torch.bfloat16))
 def plex(ynputs):
     onputs = tokenizer.encode(""+ynputs+"", return_tensors="pt").to("cpu")
-    iutputs = model.generate(onputs, max_new_tokens=350)
+    iutputs = pipe.generate(onputs, max_new_tokens=350)
     return tokenizer.decode(iutputs[0])
 with gr.Blocks() as iface:
     zutput=gr.Textbox(lines=5)