Update app.py

app.py

@@ -8,8 +8,8 @@ import lightning as L
 import torch
 
 # support running without installing as a package
-# wd = Path(__file__).parent.parent.resolve()
-# sys.path.append(str(wd))
+wd = Path(__file__).parent.parent.resolve()
+sys.path.append(str(wd))
 
 from generate import generate
 from lit_llama import Tokenizer
@@ -32,7 +32,7 @@ torch.set_float32_matmul_precision("high")
 def model_load(
     adapter_path: Path = Path("out/adapter/alpaca/lit-llama-adapter-finetuned_15k.pth"),
     pretrained_path: Path = Path("checkpoints/lit-llama/7B/lit-llama.pth"),
-    quantize: Optional[str] = "llm.int8",
+    quantize: Optional[str] = None,
 ):
 
     fabric = L.Fabric(devices=1)