Keep Minimax's indexer tensors at F32 for speed and accuracy (#26144)

* Keep Minimax's indexer tensors at F32 for speed and accuracy * name -> new_name
2026-07-26 18:02:56 -04:00
parent 55b7d6c4c7
commit 7657a6c26a
2 changed files with 9 additions and 0 deletions
@@ -58,6 +58,11 @@ class MiniMaxM2Model(TextModel):
 class MiniMaxM3Model(MiniMaxM2Model):
    model_arch = gguf.MODEL_ARCH.MINIMAXM3

+    def tensor_force_quant(self, name, new_name, bid, n_dims):
+        if ".indexer." in new_name:
+            return gguf.GGMLQuantizationType.F32
+        return super().tensor_force_quant(name, new_name, bid, n_dims)
+
    def set_gguf_parameters(self):
        super().set_gguf_parameters()