Disable lazy loader when using GPT-2
This commit is contained in:
parent
f9732eb143
commit
1380eb0bb0
|
@ -1804,6 +1804,7 @@ def load_model(use_gpu=True, gpu_layers=None, initial_load=False, online_model="
|
||||||
# feature yet
|
# feature yet
|
||||||
if(vars.model_type == "gpt2"):
|
if(vars.model_type == "gpt2"):
|
||||||
lowmem = {}
|
lowmem = {}
|
||||||
|
vars.lazy_load = False # Also, lazy loader doesn't support GPT-2 models
|
||||||
|
|
||||||
# If we're using torch_lazy_loader, we need to get breakmodel config
|
# If we're using torch_lazy_loader, we need to get breakmodel config
|
||||||
# early so that it knows where to load the individual model tensors
|
# early so that it knows where to load the individual model tensors
|
||||||
|
|
Loading…
Reference in New Issue