meta-llama_Llama-2-13b-chat-hf.yaml

model:
  init:
    torch_dtype: "float16"
  inference:
    max_new_tokens: 2024
    do_sample: True
tokenizer:
  padding_side: "left"
prompt:
  system_message: True