From a979927727c43337361df888f7fa6158a59f0b4f Mon Sep 17 00:00:00 2001
From: Alison Shao <54658187+alisonshao@users.noreply.github.com>
Date: Thu, 8 Jan 2026 15:13:13 -0800
Subject: [PATCH] Skip causal_conv1d test with padded batches due to Triton
 kernel bug (#16715)

---
 test/registered/layers/mamba/test_causal_conv1d.py | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/test/registered/layers/mamba/test_causal_conv1d.py b/test/registered/layers/mamba/test_causal_conv1d.py
index 6d408d03a..a4a65b799 100644
--- a/test/registered/layers/mamba/test_causal_conv1d.py
+++ b/test/registered/layers/mamba/test_causal_conv1d.py
@@ -195,6 +195,9 @@ def test_causal_conv1d_update_with_batch_gather(
     if not torch.cuda.is_available():
         pytest.skip("CUDA device not available")
 
+    # TODO: Fix Triton kernel bug - see issue #16714
+    pytest.skip("Triton kernel bug - see issue #16714")
+
     device = "cuda"
     rtol, atol = (3e-4, 1e-3) if itype == torch.float32 else (3e-3, 5e-3)
     if itype == torch.bfloat16:
@@ -275,6 +278,9 @@ def test_causal_conv1d_varlen(
     if not torch.cuda.is_available():
         pytest.skip("CUDA device not available")
 
+    # TODO: Fix Triton kernel bug - see issue #16714
+    pytest.skip("Triton kernel bug - see issue #16714")
+
     device = "cuda"
     torch.cuda.empty_cache()
     rtol, atol = (3e-4, 1e-3) if itype == torch.float32 else (3e-3, 5e-3)