scripts : Use mmap in torch load (#4202)

* Use mmap in torch load, prefer .bin files when loading

* Revert .bin > .safetensors preference

diff --git a/convert-hf-to-gguf.py b/convert-hf-to-gguf.py
index 147d5717e7e573dd97327f654fff9eff5c49989c..53ce76c70b0894e0369ae6c6790f9c7f61a7b1bf 100755 (executable)
--- a/convert-hf-to-gguf.py
+++ b/convert-hf-to-gguf.py
@@ -59,7 +59,7 @@ class Model:
                 from safetensors import safe_open
                 ctx = cast(ContextManager[Any], safe_open(self.dir_model / part_name, framework="pt", device="cpu"))
             else:
-                ctx = contextlib.nullcontext(torch.load(self.dir_model / part_name, map_location="cpu"))
+                ctx = contextlib.nullcontext(torch.load(str(self.dir_model / part_name), map_location="cpu", mmap=True, weights_only=True))
 
             with ctx as model_part:
                 for name in model_part.keys():