upgrade chatglm to chatglm2
This commit is contained in:
parent
37172906ef
commit
403667aec1
@ -152,7 +152,7 @@ model_info = {
|
|||||||
"token_cnt": get_token_num_gpt4,
|
"token_cnt": get_token_num_gpt4,
|
||||||
},
|
},
|
||||||
|
|
||||||
# chatglm
|
# 将 chatglm 直接对齐到 chatglm2
|
||||||
"chatglm": {
|
"chatglm": {
|
||||||
"fn_with_ui": chatglm_ui,
|
"fn_with_ui": chatglm_ui,
|
||||||
"fn_without_ui": chatglm_noui,
|
"fn_without_ui": chatglm_noui,
|
||||||
@ -161,6 +161,15 @@ model_info = {
|
|||||||
"tokenizer": tokenizer_gpt35,
|
"tokenizer": tokenizer_gpt35,
|
||||||
"token_cnt": get_token_num_gpt35,
|
"token_cnt": get_token_num_gpt35,
|
||||||
},
|
},
|
||||||
|
"chatglm2": {
|
||||||
|
"fn_with_ui": chatglm_ui,
|
||||||
|
"fn_without_ui": chatglm_noui,
|
||||||
|
"endpoint": None,
|
||||||
|
"max_token": 1024,
|
||||||
|
"tokenizer": tokenizer_gpt35,
|
||||||
|
"token_cnt": get_token_num_gpt35,
|
||||||
|
},
|
||||||
|
|
||||||
# newbing
|
# newbing
|
||||||
"newbing": {
|
"newbing": {
|
||||||
"fn_with_ui": newbing_ui,
|
"fn_with_ui": newbing_ui,
|
||||||
|
@ -40,12 +40,12 @@ class GetGLMHandle(Process):
|
|||||||
while True:
|
while True:
|
||||||
try:
|
try:
|
||||||
if self.chatglm_model is None:
|
if self.chatglm_model is None:
|
||||||
self.chatglm_tokenizer = AutoTokenizer.from_pretrained("THUDM/chatglm-6b", trust_remote_code=True)
|
self.chatglm_tokenizer = AutoTokenizer.from_pretrained("THUDM/chatglm2-6b", trust_remote_code=True)
|
||||||
device, = get_conf('LOCAL_MODEL_DEVICE')
|
device, = get_conf('LOCAL_MODEL_DEVICE')
|
||||||
if device=='cpu':
|
if device=='cpu':
|
||||||
self.chatglm_model = AutoModel.from_pretrained("THUDM/chatglm-6b", trust_remote_code=True).float()
|
self.chatglm_model = AutoModel.from_pretrained("THUDM/chatglm2-6b", trust_remote_code=True).float()
|
||||||
else:
|
else:
|
||||||
self.chatglm_model = AutoModel.from_pretrained("THUDM/chatglm-6b", trust_remote_code=True).half().cuda()
|
self.chatglm_model = AutoModel.from_pretrained("THUDM/chatglm2-6b", trust_remote_code=True).half().cuda()
|
||||||
self.chatglm_model = self.chatglm_model.eval()
|
self.chatglm_model = self.chatglm_model.eval()
|
||||||
break
|
break
|
||||||
else:
|
else:
|
||||||
|
Loading…
x
Reference in New Issue
Block a user