Use llama.cpp compatible split for Q8_0 variant

command-r-plus-Q8_0-00001-of-00003.gguf

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:29715ebc252bc0306e65125ad01ad4ab3a04a4d6f407a0e5598ef22700d57618
-size 37754571886
+oid sha256:efa0bf92a22229e063037d366e7b28fee5b4e7e4e1f8d63591b2a04d68b6e5e3
+size 48070727520

command-r-plus-Q8_0-00002-of-00003.gguf

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ebf24fc8c49847bc78522e6df9b6b2a2e6c117a7dbdf9f7b6128a44561829be1
-size 37754571949
+oid sha256:d64e7d340b4f21575399acf1eea7818bc0312b3e103a70aa8c378e40cf40459f
+size 41353956448

command-r-plus-Q8_0-00003-of-00003.gguf

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:296183bd08de09f4a341fd9e36ce57dd25be17b761939850f43e679dd631eb73
-size 37754571749
+oid sha256:625a9f703921be3684934026f14b64c470d953a9e390a423f381f59f53c56024
+size 23839031936