lingyun1/Hare-zh-ckpt-190000

from transformers import AutoTokenizer,GenerationConfig,AutoConfig,GenerationConfig,AutoModelForCausalLM
import torch
from transformers import AutoTokenizer
model_path = "model_path"
model = AutoModelForCausalLM.from_pretrained(model_path,trust_remote_code = True)
tokenizer = AutoTokenizer.from_pretrained(model_path)

_inputs = """your input"""

gen_conf = GenerationConfig(
   num_beams=1,
   do_sample=False,
   max_length=128,
   max_new_tokens=128,
   no_repeat_ngram_size=4,
   eos_token_id=tokenizer.eos_token_id,
   pad_token_id=tokenizer.pad_token_id,
)

tokend = tokenizer.encode_plus(text=_inputs, add_special_tokens=False)
input_ids_lengths = len(tokend.input_ids)
print(_inputs)
input_ids = torch.LongTensor([tokend.input_ids])

outputs = model.generate(
   inputs=input_ids,
   generation_config=gen_conf,
)

outs = tokenizer.decode(outputs[0][input_ids_lengths:],skip_special_tokens=False)
print(outs)