mirror of
https://github.com/RYDE-WORK/Langchain-Chatchat.git
synced 2026-01-26 00:33:35 +08:00
33bbb4779e
loader.py: 模型重载 定义 generatorAnswer 增加 AnswerResultStream 定义generate_with_callback收集器，在每次响应时将队列数据同步到AnswerResult requirements.txt 变更项目依赖
97 lines
3.5 KiB
Python
97 lines
3.5 KiB
Python
|
|
from abc import ABC
|
|
|
|
from langchain.llms.base import LLM
|
|
from typing import Optional, List
|
|
from models.loader import LoaderCheckPoint
|
|
from models.base import (BaseAnswer,
|
|
AnswerResult,
|
|
AnswerResultStream,
|
|
AnswerResultQueueSentinelTokenListenerQueue)
|
|
|
|
|
|
import transformers
|
|
|
|
|
|
class ChatGLM(BaseAnswer, LLM, ABC):
|
|
max_token: int = 10000
|
|
temperature: float = 0.01
|
|
top_p = 0.9
|
|
checkPoint: LoaderCheckPoint = None
|
|
# history = []
|
|
history_len: int = 10
|
|
|
|
def __init__(self, checkPoint: LoaderCheckPoint = None):
|
|
super().__init__()
|
|
self.checkPoint = checkPoint
|
|
|
|
@property
|
|
def _llm_type(self) -> str:
|
|
return "ChatGLM"
|
|
|
|
@property
|
|
def _check_point(self) -> LoaderCheckPoint:
|
|
return self.checkPoint
|
|
|
|
@property
|
|
def _history_len(self) -> int:
|
|
return self.history_len
|
|
|
|
def set_history_len(self, history_len: int = 10) -> None:
|
|
self.history_len = history_len
|
|
|
|
def _call(self, prompt: str, stop: Optional[List[str]] = None) -> str:
|
|
pass
|
|
|
|
def _generate_answer(self, prompt: str,
|
|
history: List[List[str]] = [],
|
|
streaming: bool = False,
|
|
generate_with_callback: AnswerResultStream = None) -> None:
|
|
# Create the StoppingCriteriaList with the stopping strings
|
|
stopping_criteria_list = transformers.StoppingCriteriaList()
|
|
# 定义模型stopping_criteria 队列，在每次响应时将 torch.LongTensor, torch.FloatTensor同步到AnswerResult
|
|
listenerQueue = AnswerResultQueueSentinelTokenListenerQueue()
|
|
stopping_criteria_list.append(listenerQueue)
|
|
|
|
if streaming:
|
|
|
|
for inum, (stream_resp, _) in enumerate(self.checkPoint.model.stream_chat(
|
|
self.checkPoint.tokenizer,
|
|
prompt,
|
|
history=history[-self.history_len:-1] if self.history_len > 0 else [],
|
|
max_length=self.max_token,
|
|
temperature=self.temperature,
|
|
stopping_criteria=stopping_criteria_list
|
|
)):
|
|
self.checkPoint.clear_torch_cache()
|
|
if inum == 0:
|
|
history += [[prompt, stream_resp]]
|
|
else:
|
|
history[-1] = [prompt, stream_resp]
|
|
answer_result = AnswerResult()
|
|
answer_result.history = history
|
|
answer_result.llm_output = {"answer": stream_resp}
|
|
if listenerQueue.listenerQueue.__len__() > 0:
|
|
answer_result.listenerToken = listenerQueue.listenerQueue.pop()
|
|
generate_with_callback(answer_result)
|
|
else:
|
|
response, _ = self.checkPoint.model.chat(
|
|
self.checkPoint.tokenizer,
|
|
prompt,
|
|
history=history[-self.history_len:] if self.history_len > 0 else [],
|
|
max_length=self.max_token,
|
|
temperature=self.temperature,
|
|
stopping_criteria=stopping_criteria_list
|
|
)
|
|
self.checkPoint.clear_torch_cache()
|
|
history += [[prompt, response]]
|
|
answer_result = AnswerResult()
|
|
answer_result.history = history
|
|
answer_result.llm_output = {"answer": response}
|
|
if listenerQueue.listenerQueue.__len__() > 0:
|
|
answer_result.listenerToken = listenerQueue.listenerQueue.pop()
|
|
|
|
generate_with_callback(answer_result)
|
|
|
|
|