Mirrors/text-generation-webui
1
0
Fork 0
mirror of https://github.com/oobabooga/text-generation-webui.git synced 2024-09-20 10:35:10 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 1

Bitsandbytes: load preconverted 4bit models without additional flags

Browse source
This commit is contained in:
oobabooga 2024-04-04 18:09:34 -07:00
parent d423021a48
commit 308452b783
1 changed files with 10 additions and 9 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

19
modules/models.py
View file

@ -145,17 +145,18 @@ def huggingface_loader(model_name):
else:
LoaderClass = AutoModelForCausalLM
# Load the model in simple 16-bit mode by default
# Load the model without any special settings
if not any([shared.args.cpu, shared.args.load_in_8bit, shared.args.load_in_4bit, shared.args.auto_devices, shared.args.disk, shared.args.deepspeed, shared.args.gpu_memory is not None, shared.args.cpu_memory is not None, shared.args.compress_pos_emb > 1, shared.args.alpha_value > 1, shared.args.disable_exllama, shared.args.disable_exllamav2]):
model = LoaderClass.from_pretrained(path_to_model, **params)
if torch.backends.mps.is_available():
device = torch.device('mps')
model = model.to(device)
elif is_xpu_available():
device = torch.device("xpu")
model = model.to(device)
else:
model = model.cuda()
if not (hasattr(model, 'is_loaded_in_4bit') and model.is_loaded_in_4bit):
if torch.backends.mps.is_available():
device = torch.device('mps')
model = model.to(device)
elif is_xpu_available():
device = torch.device("xpu")
model = model.to(device)
else:
model = model.cuda()
# DeepSpeed ZeRO-3
elif shared.args.deepspeed: