Update app.py

app.py

@@ -11,9 +11,12 @@ import spaces
 import time
 import subprocess
 
-# from transformers_modules.apple.OpenELM-3B-Instruct.d3c76da586450c73898e6bed70cecf3376300fb3.configuration_openelm import OpenELMConfig
+from transformers_modules.apple.OpenELM-3B-Instruct.d3c76da586450c73898e6bed70cecf3376300fb3.configuration_openelm import OpenELMConfig
 
 from transformers_modules.apple.OpenELM_3B_Instruct.d3c76da586450c73898e6bed70cecf3376300fb3.configuration_openelm import OpenELMConfig
+from transformers import AutoConfig, AutoTokenizer
+
+config = AutoConfig.from_pretrained("apple/OpenELM-270M", trust_remote_code=True)
 
 
 # flash-attn 라이브러리 설치. CUDA 빌드는 건너뜀.
@@ -30,14 +33,14 @@ token = os.environ["HF_TOKEN"]
 # beomi/Llama-3-KoEn-8B-Instruct-preview 모델과 토크나이저 같이 로드
 model = AutoModelForCausalLM.from_pretrained(
     # "microsoft/Phi-3-mini-128k-instruct",
-    "apple/OpenELM-3B-Instruct",
+    "apple/OpenELM-270M",
     token=token,
     trust_remote_code=True,
 )
 # bug fix
 # tok = AutoTokenizer.from_pretrained("apple/OpenELM-3B-Instruct", token=token, trust_remote_code=True)
 tok = AutoTokenizer.from_pretrained(
-    "apple/OpenELM-3B-Instruct",
+    "apple/OpenELM-270M",
     token=token,
     trust_remote_code=True,
     config=OpenELMConfig(),