Hugging Face's logo

Spaces:
vilarin
/
Llama-3.1-8B-Instruct
Running on Zero

App Files Community
vilarin commited on
Commit
b179e70
•
1 Parent(s): abf1b38

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +1 -0
app.py CHANGED
@@ -36,6 +36,7 @@ device = "cuda" # for GPU usage or "cpu" for CPU usage
36
  tokenizer = AutoTokenizer.from_pretrained(MODEL)
37
  model = AutoModelForCausalLM.from_pretrained(
38
  MODEL,
 
39
  low_cpu_mem_usage=True,
40
  device_map="auto")
41
 
 
36
  tokenizer = AutoTokenizer.from_pretrained(MODEL)
37
  model = AutoModelForCausalLM.from_pretrained(
38
  MODEL,
39
+ torch_dtype=torch.bfloat16,
40
  low_cpu_mem_usage=True,
41
  device_map="auto")
42