zhengjl/text-generation-webui
Public Access
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Update GPTQ_Loader.py

Browse Source 
Correcting decoder layer for renamed class.
This commit is contained in:
askmyteapot
2023-03-17 18:34:13 +10:00
committed by GitHub
parent ee164d1821
commit 53b6a66beb
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
modules/GPTQ_loader.py
View File

@@ -61,7 +61,7 @@ def load_quantized(model_name):
max_memory[i] = f"{shared.args.gpu_memory[i]}GiB" max_memory[i] = f"{shared.args.gpu_memory[i]}GiB"
max_memory['cpu'] = f"{shared.args.cpu_memory or '99'}GiB" max_memory['cpu'] = f"{shared.args.cpu_memory or '99'}GiB"
device_map = accelerate.infer_auto_device_map(model, max_memory=max_memory, no_split_module_classes=["LLaMADecoderLayer"]) device_map = accelerate.infer_auto_device_map(model, max_memory=max_memory, no_split_module_classes=["LlamaDecoderLayer"])
model = accelerate.dispatch_model(model, device_map=device_map) model = accelerate.dispatch_model(model, device_map=device_map)
# Single GPU # Single GPU