We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
There was an error while loading. Please reload this page.
1 parent 6fc153d commit c6aa83fCopy full SHA for c6aa83f
exllamav3/models/model.py
@@ -92,7 +92,12 @@ def unload(self):
92
def _load_single(self, progressbar: bool, device: torch.device):
93
with ProgressBar(f"Loading" if progressbar else None, len(self.modules)) as progress:
94
for idx, module in enumerate(self.modules):
95
+ defer = module.can_defer_load()
96
+ if defer:
97
+ self.config.stc.begin_deferred_load()
98
module.load(torch.device("cpu") if module.caps.get("prefer_cpu") else device)
99
100
+ self.config.stc.end_deferred_load()
101
progress.update(idx + 1)
102
103
0 commit comments