github
/
ChatGLM2-6B
mirror of https://github.com/THUDM/ChatGLM2-6B
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

Add multi-gpu support

pull/78/head
duzx16 2023-06-27 14:40:44 +08:00
parent 549cdeb054
commit b09e0c5a69
4 changed files with 12 additions and 17 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
cli_demo.py
View File

@ -3,14 +3,12 @@ import platform
import signal import signal
from transformers import AutoTokenizer, AutoModel from transformers import AutoTokenizer, AutoModel
import readline import readline
from utils import load_model_on_gpus
tokenizer = AutoTokenizer.from_pretrained("THUDM/chatglm2-6b", trust_remote_code=True) tokenizer = AutoTokenizer.from_pretrained("THUDM/chatglm2-6b", trust_remote_code=True)
model = AutoModel.from_pretrained("THUDM/chatglm2-6b", trust_remote_code=True).cuda() model = AutoModel.from_pretrained("THUDM/chatglm2-6b", trust_remote_code=True).cuda()
# 多显卡支持，使用下面三行代替上面两行，将num_gpus改为你实际的显卡数量 # 多显卡支持，使用下面两行代替上面一行，将num_gpus改为你实际的显卡数量
# model_path = "THUDM/chatglm2-6b" # from utils import load_model_on_gpus
# tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True) # model = load_model_on_gpus("THUDM/chatglm2-6b", num_gpus=2)
# model = load_model_on_gpus(model_path, num_gpus=2)
model = model.eval() model = model.eval()
os_name = platform.system() os_name = platform.system()

7
openai_api.py
View File

@ -158,10 +158,9 @@ async def predict(query: str, history: List[List[str]], model_id: str):
if __name__ == "__main__": if __name__ == "__main__":
tokenizer = AutoTokenizer.from_pretrained("THUDM/chatglm2-6b", trust_remote_code=True) tokenizer = AutoTokenizer.from_pretrained("THUDM/chatglm2-6b", trust_remote_code=True)
model = AutoModel.from_pretrained("THUDM/chatglm2-6b", trust_remote_code=True).cuda() model = AutoModel.from_pretrained("THUDM/chatglm2-6b", trust_remote_code=True).cuda()
# 多显卡支持，使用下面三行代替上面两行，将num_gpus改为你实际的显卡数量 # 多显卡支持，使用下面两行代替上面一行，将num_gpus改为你实际的显卡数量
# model_path = "THUDM/chatglm2-6b" # from utils import load_model_on_gpus
# tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True) # model = load_model_on_gpus("THUDM/chatglm2-6b", num_gpus=2)
# model = load_model_on_gpus(model_path, num_gpus=2)
model.eval() model.eval()
uvicorn.run(app, host='0.0.0.0', port=8000, workers=1) uvicorn.run(app, host='0.0.0.0', port=8000, workers=1)

7
web_demo.py
View File

@ -5,10 +5,9 @@ from utils import load_model_on_gpus
tokenizer = AutoTokenizer.from_pretrained("THUDM/chatglm2-6b", trust_remote_code=True) tokenizer = AutoTokenizer.from_pretrained("THUDM/chatglm2-6b", trust_remote_code=True)
model = AutoModel.from_pretrained("THUDM/chatglm2-6b", trust_remote_code=True).cuda() model = AutoModel.from_pretrained("THUDM/chatglm2-6b", trust_remote_code=True).cuda()
# 多显卡支持，使用下面三行代替上面两行，将num_gpus改为你实际的显卡数量 # 多显卡支持，使用下面两行代替上面一行，将num_gpus改为你实际的显卡数量
# model_path = "THUDM/chatglm2-6b" # from utils import load_model_on_gpus
# tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True) # model = load_model_on_gpus("THUDM/chatglm2-6b", num_gpus=2)
# model = load_model_on_gpus(model_path, num_gpus=2)
model = model.eval() model = model.eval()
"""Override Chatbot.postprocess""" """Override Chatbot.postprocess"""

7
web_demo2.py
View File

@ -14,10 +14,9 @@ st.set_page_config(
def get_model(): def get_model():
tokenizer = AutoTokenizer.from_pretrained("THUDM/chatglm2-6b", trust_remote_code=True) tokenizer = AutoTokenizer.from_pretrained("THUDM/chatglm2-6b", trust_remote_code=True)
model = AutoModel.from_pretrained("THUDM/chatglm2-6b", trust_remote_code=True).cuda() model = AutoModel.from_pretrained("THUDM/chatglm2-6b", trust_remote_code=True).cuda()
# 多显卡支持，使用下面三行代替上面两行，将num_gpus改为你实际的显卡数量 # 多显卡支持，使用下面两行代替上面一行，将num_gpus改为你实际的显卡数量
# model_path = "THUDM/chatglm2-6b" # from utils import load_model_on_gpus
# tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True) # model = load_model_on_gpus("THUDM/chatglm2-6b", num_gpus=2)
# model = load_model_on_gpus(model_path, num_gpus=2)
model = model.eval() model = model.eval()
return tokenizer, model return tokenizer, model