codellama

/

CodeLlama-34b-hf

like 167

Follow

Code Llama 514

here is a max_length
transformers.pipeline(
"text-generation",
model=_hf_model_dir,
torch_dtype=torch.float16,
device_map="auto",
)(_content,
do_sample=True,
top_k=5,
num_return_sequences=1,
eos_token_id=_hf_tokenizer.eos_token_id,
max_length=1000,
temperature=0.2,
top_p=0.95,
)