Baichuan2-13B-Chat / model.py
jZoNg's picture
fix model.chat
3ccf1a4
raw
history blame
2.36 kB
from threading import Thread
from typing import Iterator
import torch
from transformers import AutoModelForCausalLM, AutoTokenizer
from transformers.generation.utils import GenerationConfig
model_id = 'baichuan-inc/Baichuan2-13B-Chat'
if torch.cuda.is_available():
model = AutoModelForCausalLM.from_pretrained(
model_id,
# device_map='auto',
torch_dtype=torch.float16,
trust_remote_code=True
)
model = model.quantize(4).cuda()
model.generation_config = GenerationConfig.from_pretrained(model_id)
else:
model = None
tokenizer = AutoTokenizer.from_pretrained(
model_id,
use_fast=False,
trust_remote_code=True
)
def get_prompt(
message: str,
chat_history: list[tuple[str, str]],
system_prompt: str
) -> str:
texts = [f'<s>[INST] <<SYS>>\n{system_prompt}\n<</SYS>>\n\n']
# The first user input is _not_ stripped
do_strip = False
for user_input, response in chat_history:
user_input = user_input.strip() if do_strip else user_input
do_strip = True
texts.append(f'{user_input} [/INST] {response.strip()} </s><s>[INST] ')
message = message.strip() if do_strip else message
texts.append(f'{message} [/INST]')
return ''.join(texts)
def get_input_token_length(
message: str,
chat_history: list[tuple[str, str]],
system_prompt: str
) -> int:
prompt = get_prompt(message, chat_history, system_prompt)
input_ids = tokenizer([prompt], return_tensors='np', add_special_tokens=False)['input_ids']
return input_ids.shape[-1]
def run(
message: str,
chat_history: list[tuple[str, str]],
system_prompt: str,
max_new_tokens: int = 1024,
temperature: float = 1.0,
top_p: float = 0.95,
top_k: int = 5
) -> Iterator[str]:
print(chat_history)
history = []
result=""
for i in chat_history:
history.append({"role": "user", "content": i[0]})
history.append({"role": "assistant", "content": i[1]})
print(history)
history.append({"role": "user", "content": message})
for response in model.chat(
tokenizer,
history,
# stream=True,
# max_new_tokens=max_new_tokens,
# temperature=temperature,
# top_p=top_p,
# top_k=top_k,
):
print(response)
result = result + response
yield result
# if "content" in response["choices"][0]["delta"]:
# result = result + response["choices"][0]["delta"]["content"]
# yield result