diff --git a/.github/workflows/nv-ds-chat.yml b/.github/workflows/nv-ds-chat.yml
index 94571eb101bb..cf8756fbd528 100644
--- a/.github/workflows/nv-ds-chat.yml
+++ b/.github/workflows/nv-ds-chat.yml
@@ -10,6 +10,10 @@ on:
         required: false
         default: 'master'
         type: string
+  pull_request:
+    paths:
+      - "deepspeed/runtime/zero/stage_1_and_2.py"
+      - "deepspeed/runtime/zero/stage3.py"
 
 concurrency:
   group: ${{ github.workflow }}-${{ github.ref }}
diff --git a/build_win.bat b/build_win.bat
index ec8c8a362a78..af5c5103fa4b 100644
--- a/build_win.bat
+++ b/build_win.bat
@@ -1,6 +1,10 @@
 @echo off
 
 set DS_BUILD_AIO=0
+set DS_BUILD_CUTLASS_OPS=0
+set DS_BUILD_EVOFORMER_ATTN=0
+set DS_BUILD_FP_QUANTIZER=0
+set DS_BUILD_RAGGED_DEVICE_OPS=0
 set DS_BUILD_SPARSE_ATTN=0
 
 echo Administrative permissions required. Detecting permissions...
diff --git a/deepspeed/runtime/zero/stage3.py b/deepspeed/runtime/zero/stage3.py
index 68cab13c4a93..c6ff216edfcb 100644
--- a/deepspeed/runtime/zero/stage3.py
+++ b/deepspeed/runtime/zero/stage3.py
@@ -15,7 +15,7 @@
 from deepspeed.utils import logger
 from deepspeed.runtime.fp16.loss_scaler import CreateLossScaler
 from deepspeed.runtime.comm.coalesced_collectives import reduce_scatter_coalesced, all_to_all_quant_reduce
-from deepspeed.runtime.utils import inf, is_model_parallel_parameter, get_only_unique_item
+from deepspeed.runtime.utils import inf, get_global_norm, is_model_parallel_parameter, get_only_unique_item
 from deepspeed.runtime.zero.partition_parameters import *
 from deepspeed.runtime.zero.config import ZeroStageEnum
 from deepspeed.runtime.zero.offload_config import OffloadDeviceEnum
@@ -2027,7 +2027,7 @@ def step(self, closure=None):
             return
 
         norm_groups = self._get_norm_groups()
-        scaled_global_grad_norm = torch.norm(torch.stack(norm_groups))
+        scaled_global_grad_norm = get_global_norm(norm_list=norm_groups)
 
         # Stash unscaled gradient norm
         self._global_grad_norm = scaled_global_grad_norm / self.loss_scale