updated demo

util/utils.py

@@ -35,7 +35,7 @@ def load_tok(model_name="gpt2-xl"):
 
     elif model_name == 'llama-3-8b':
 
-        model = "meta-llama/Meta-Llama-3-8B"
+        model = "/data/meta-llama/Meta-Llama-3-8B"
         tok = AutoTokenizer.from_pretrained(model)
         tok.pad_token = tok.eos_token
 
@@ -75,7 +75,7 @@ def load_model_tok(model_name="gpt2-xl"):
 
     elif model_name == 'llama-3-8b':
 
-        model = "meta-llama/Meta-Llama-3-8B"
+        model = "/data/meta-llama/Meta-Llama-3-8B"
         tok = AutoTokenizer.from_pretrained(model)
         model = AutoModelForCausalLM.from_pretrained(
             model, 
@@ -463,4 +463,4 @@ def print_request(rs):
         print_single_request(rs)
     else:
         for r in rs:
-            print_single_request(r)
+            print_single_request(r)