push

app.py

@@ -9,7 +9,7 @@ from transformers import AutoTokenizer, AutoModelForCausalLM
 
 # nstruct_pipeline_3b = pipeline(model="fnlp/moss-moon-003-sft-int4", torch_dtype=torch.float, trust_remote_code=True,
 #                                device_map="auto")
-model_path = "fnlp/moss-moon-003-sft-int8"
+model_path = "fnlp/moss-moon-003-sft-int4"
 
 # config = MossConfig.from_pretrained(model_path)
 # tokenizer = MossTokenizer.from_pretrained(model_path)