From be2c0e7bff1735c5771d007ed5bcc04e9ea86a96 Mon Sep 17 00:00:00 2001
From: Mohamed Mekkouri <93391238+MekkCyber@users.noreply.github.com>
Date: Tue, 25 Mar 2025 10:43:27 +0100
Subject: [PATCH] Fixing _pre_quantization_dtype when torch_dtype is None
 (#36930)

fix
---
 src/transformers/modeling_utils.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/transformers/modeling_utils.py b/src/transformers/modeling_utils.py
index a710c5071a..ec3b37404d 100644
--- a/src/transformers/modeling_utils.py
+++ b/src/transformers/modeling_utils.py
@@ -4454,7 +4454,7 @@ class PreTrainedModel(nn.Module, ModuleUtilsMixin, GenerationMixin, PushToHubMix
             # once the weights have been quantized
             # Note that once you have loaded a quantized model, you can't change its dtype so this will
             # remain a single source of truth
-            config._pre_quantization_dtype = torch_dtype
+            config._pre_quantization_dtype = torch_dtype if torch_dtype is not None else torch.get_default_dtype()
 
         # Prepare the full device map
         if device_map is not None: