fix

Author: zewenli98

.github/scripts/generate-tensorrt-test-matrix.py

@@ -28,6 +28,10 @@
 # please update the future tensorRT version you want to test here
 TENSORRT_VERSIONS_DICT = {
     "windows": {
+        "10.3.0": {
+            "urls": "https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.3.0/zip/TensorRT-10.3.0.26.Windows.win10.cuda-12.5.zip",
+            "strip_prefix": "TensorRT-10.3.0.26",
+        },
         "10.7.0": {
             "urls": "https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.7.0/zip/TensorRT-10.7.0.23.Windows.win10.cuda-12.6.zip",
             "strip_prefix": "TensorRT-10.7.0.23",
@@ -42,6 +46,10 @@
         },
     },
     "linux": {
+        "10.3.0": {
+            "urls": "https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.3.0/tars/TensorRT-10.3.0.26.Linux.x86_64-gnu.cuda-12.5.tar.gz",
+            "strip_prefix": "TensorRT-10.3.0.26",
+        },
         "10.7.0": {
             "urls": "https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.7.0/tars/TensorRT-10.7.0.23.Linux.x86_64-gnu.cuda-12.6.tar.gz",
             "strip_prefix": "TensorRT-10.7.0.23",

py/torch_tensorrt/dynamo/conversion/_TRTInterpreter.py

@@ -425,10 +425,10 @@ def check_weight_equal(
             network_weight = torch.from_numpy(network_weight).to(device)
         try:
             return sd_weight.shape == network_weight.shape and torch.all(
-                torch.abs(sd_weight - network_weight) < 0.01
+                torch.abs(sd_weight.cpu() - network_weight.cpu()) < 0.01
             )
         except Exception:
-            return torch.all(sd_weight == network_weight)
+            return torch.all(sd_weight.cpu() == network_weight.cpu())
 
     def _save_weight_mapping(self) -> None:
         """