Mirrors/text-generation-webui
1
0
Fork 0
mirror of https://github.com/oobabooga/text-generation-webui.git synced 2024-11-23 00:18:20 +01:00
Code Issues Packages Projects Releases Wiki Activity

Disable pre_layer when the model type is not llama

Browse Source
This commit is contained in:
oobabooga 2023-04-05 01:19:26 -03:00
parent ca8bb38949
commit f3a2e0b8a9
1 changed files with 7 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
modules/GPTQ_loader.py
View File

@ -65,8 +65,12 @@ def load_quantized(model_name):
else: else:
model_type = shared.args.model_type.lower() model_type = shared.args.model_type.lower()
if model_type == 'llama' and shared.args.pre_layer: if shared.args.pre_layer:
load_quant = llama_inference_offload.load_quant if model_type == 'llama':
load_quant = llama_inference_offload.load_quant
else:
print("Warning: ignoring --pre_layer because it only works for llama model type.")
load_quant = _load_quant
elif model_type in ('llama', 'opt', 'gptj'): elif model_type in ('llama', 'opt', 'gptj'):
load_quant = _load_quant load_quant = _load_quant
else: else:
@ -107,7 +111,7 @@ def load_quantized(model_name):
exit() exit()
# qwopqwop200's offload # qwopqwop200's offload
if shared.args.pre_layer: if model_type == 'llama' and shared.args.pre_layer:
model = load_quant(str(path_to_model), str(pt_path), shared.args.wbits, shared.args.groupsize, shared.args.pre_layer) model = load_quant(str(path_to_model), str(pt_path), shared.args.wbits, shared.args.groupsize, shared.args.pre_layer)
else: else:
threshold = False if model_type == 'gptj' else 128 threshold = False if model_type == 'gptj' else 128