Upload app.py

app.py

@@ -1,17 +1,23 @@
 import gradio as gr
 import torch
-from transformers import AutoModelForCausalLM, AutoTokenizer
+from transformers import AutoModelForCausalLM, AutoTokenizer, BitsAndBytesConfig
 from einops import einsum
 from tqdm import tqdm
 
 device = "cuda" if torch.cuda.is_available() else "cpu"
 model_name = 'microsoft/Phi-3-mini-4k-instruct'
 
+quantization_config = BitsAndBytesConfig(
+    load_in_4bit=True,
+    bnb_4bit_compute_dtype=torch.float16
+)
+
 model = AutoModelForCausalLM.from_pretrained(
     model_name, 
     device_map=device, 
     torch_dtype="auto", 
     trust_remote_code=True, 
+    quantization_config=quantization_config,
 )
 
 tokenizer = AutoTokenizer.from_pretrained(model_name)