change model

modules/app.py

@@ -6,9 +6,11 @@ import os
 if os.path.exists('use_normal_tokenizers'):
     import tokenizers
     BIG_MODEL = False
+    CUDA = False
 else:
     import tokenizers_patch
-    BIG_MODEL = True
+    BIG_MODEL = False
+    CUDA = True
 from transformers import AutoModelForCausalLM, AutoTokenizer
 import json
 
@@ -29,10 +31,8 @@ MAX_LENGTH = 256+64
 TRUNCATION_MESSAGE = f'warning: This demo is limited to {MAX_LENGTH} tokens in the document for efficiency.'
 
 if BIG_MODEL:
-    CUDA = True
-    model_name = "./incoder-6B"
+    model_name = "facebook/incoder-6B"
 else:
-    CUDA = False
     model_name = "facebook/incoder-1B"
 
 from fastapi import FastAPI, Request