support GDN context parallel

Author: yuzhongw-nvidia

tests/unit_tests/ssm/test_gated_delta_net.py

@@ -146,7 +146,7 @@ def test_gpu_forward(self):
 @pytest.mark.skipif(not HAVE_FLA, reason="FLA is not installed.")
 def test_parallel_gated_delta_net_correctness(tmp_path_dist_ckpt, tp, sp, cp):
     transformer_config = TransformerConfig(
-        hidden_size=128,
+        hidden_size=hidden_size,
         linear_conv_kernel_dim=2,
         linear_key_head_dim=32,
         linear_value_head_dim=32,