Update app.py

app.py

@@ -17,26 +17,14 @@ from datetime import date, datetime, timedelta
 
 access_token = os.environ["TOKEN"]
 
-base_model = AutoModelForCausalLM.from_pretrained(
-    'NousResearch/Llama-2-7b-chat-hf',
-    token=access_token,
-    trust_remote_code=True, 
-    device_map="auto",
-    load_in_8bit=True,
-    offload_folder="offload/"
-)
-model = PeftModel.from_pretrained(
-    base_model,
-    'FinGPT/fingpt-forecaster_dow30_llama2-7b_lora',
-    offload_folder="offload/"
-)
+base_model = "NousResearch/Llama-2-13b-hf" 
+peft_model = "FinGPT/fingpt-sentiment_llama2-13b_lora"
+tokenizer = LlamaTokenizerFast.from_pretrained(base_model, trust_remote_code=True)
+tokenizer.pad_token = tokenizer.eos_token
+model = LlamaForCausalLM.from_pretrained(base_model, trust_remote_code=True, device_map = "cuda:0", load_in_8bit = True,)
+model = PeftModel.from_pretrained(model, peft_model)
 model = model.eval()
 
-tokenizer = AutoTokenizer.from_pretrained(
-    'NousResearch/Llama-2-7b-chat-hf',
-    token=access_token
-)
-
 # Inference Data
 # get company news online