From 82677070ff275fc4c8c1ebe93225a0863c07a9db Mon Sep 17 00:00:00 2001
From: Christian Medina <37550954+cmedinasoriano@users.noreply.github.com>
Date: Thu, 2 Jul 2026 21:41:44 -0400
Subject: [PATCH] fix: weights_only=False for safetensors loading

---
 quantize_streaming.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/quantize_streaming.py b/quantize_streaming.py
index f75eb58..971a779 100644
--- a/quantize_streaming.py
+++ b/quantize_streaming.py
@@ -33,7 +33,7 @@ def streaming_quantize(model_path, output_path):
         
         # Load shard to CPU
         print("   Loading shard to CPU...")
-        shard_state_dict = torch.load(shard_file, map_location="cpu", weights_only=True)
+        shard_state_dict = torch.load(shard_file, map_location="cpu", weights_only=False)
         
         # Quantize Linear layers in this shard using both GPUs
         print("   Quantizing Linear layers (both GPUs)...")