Skip quant for linear_attn.in_proj_a/b

Signed-off-by: Anurag Mukkara <134339030+amukkara@users.noreply.github.com>

Author: amukkara

modelopt/torch/quantization/config.py

@@ -228,6 +228,8 @@ def find_quant_cfg_entry_by_path(
         "enable": False,
     },  # Skip the MOE router
     {"quantizer_name": "*linear_attn.conv1d*", "enable": False},
+    {"quantizer_name": "*linear_attn.in_proj_a*", "enable": False},
+    {"quantizer_name": "*linear_attn.in_proj_b*", "enable": False},
     {"quantizer_name": "*mixer.conv1d*", "enable": False},  # Skip mamba conv1d
     {"quantizer_name": "*output_layer*", "enable": False},
     {"quantizer_name": "output.*", "enable": False},