|
model_name = "ChatGLM" |
|
cmd_to_install = "`pip install -r request_llms/requirements_chatglm.txt`" |
|
|
|
|
|
from toolbox import get_conf, ProxyNetworkActivate |
|
from .local_llm_class import LocalLLMHandle, get_local_llm_predict_fns |
|
|
|
|
|
|
|
|
|
|
|
|
|
class GetGLM2Handle(LocalLLMHandle): |
|
|
|
def load_model_info(self): |
|
|
|
self.model_name = model_name |
|
self.cmd_to_install = cmd_to_install |
|
|
|
def load_model_and_tokenizer(self): |
|
|
|
import os, glob |
|
import os |
|
import platform |
|
from transformers import AutoModel, AutoTokenizer |
|
LOCAL_MODEL_QUANT, device = get_conf('LOCAL_MODEL_QUANT', 'LOCAL_MODEL_DEVICE') |
|
|
|
if LOCAL_MODEL_QUANT == "INT4": |
|
_model_name_ = "THUDM/chatglm2-6b-int4" |
|
elif LOCAL_MODEL_QUANT == "INT8": |
|
_model_name_ = "THUDM/chatglm2-6b-int8" |
|
else: |
|
_model_name_ = "THUDM/chatglm2-6b" |
|
|
|
with ProxyNetworkActivate('Download_LLM'): |
|
chatglm_tokenizer = AutoTokenizer.from_pretrained(_model_name_, trust_remote_code=True) |
|
if device=='cpu': |
|
chatglm_model = AutoModel.from_pretrained(_model_name_, trust_remote_code=True).float() |
|
else: |
|
chatglm_model = AutoModel.from_pretrained(_model_name_, trust_remote_code=True).half().cuda() |
|
chatglm_model = chatglm_model.eval() |
|
|
|
self._model = chatglm_model |
|
self._tokenizer = chatglm_tokenizer |
|
return self._model, self._tokenizer |
|
|
|
def llm_stream_generator(self, **kwargs): |
|
|
|
def adaptor(kwargs): |
|
query = kwargs['query'] |
|
max_length = kwargs['max_length'] |
|
top_p = kwargs['top_p'] |
|
temperature = kwargs['temperature'] |
|
history = kwargs['history'] |
|
return query, max_length, top_p, temperature, history |
|
|
|
query, max_length, top_p, temperature, history = adaptor(kwargs) |
|
|
|
for response, history in self._model.stream_chat(self._tokenizer, |
|
query, |
|
history, |
|
max_length=max_length, |
|
top_p=top_p, |
|
temperature=temperature, |
|
): |
|
yield response |
|
|
|
def try_to_import_special_deps(self, **kwargs): |
|
|
|
|
|
import importlib |
|
|
|
|
|
|
|
|
|
|
|
|
|
predict_no_ui_long_connection, predict = get_local_llm_predict_fns(GetGLM2Handle, model_name) |