diff --git a/ai_edge_quantizer/algorithms/uniform_quantize/naive_min_max_quantize.py b/ai_edge_quantizer/algorithms/uniform_quantize/naive_min_max_quantize.py
index e2276d7..a04f32a 100644
--- a/ai_edge_quantizer/algorithms/uniform_quantize/naive_min_max_quantize.py
+++ b/ai_edge_quantizer/algorithms/uniform_quantize/naive_min_max_quantize.py
@@ -28,6 +28,9 @@ def check_op_quantization_config(
   Raises:
     ValueError: If the op quantization config is invalid.
   """
+  if op_quant_config.skip_checks:
+    return
+
   if op_quant_config.weight_tensor_config.dtype != qtyping.TensorDataType.INT:
     raise ValueError(
         "Weights need to have integer type for min/max uniform quantization. If"
diff --git a/ai_edge_quantizer/qtyping.py b/ai_edge_quantizer/qtyping.py
index ba8d6b8..94f8e95 100644
--- a/ai_edge_quantizer/qtyping.py
+++ b/ai_edge_quantizer/qtyping.py
@@ -246,6 +246,7 @@ class OpQuantizationConfig:
     weight_tensor_config: The quantization configuration for weight tensor in
       the op.
     execution_mode: How to execute the op after quantization.
+    skip_checks: Skip op quantization config checks.
   """
 
   # Quant config for activation tensors in the op (i.e., runtime tensors).
@@ -262,6 +263,11 @@ class OpQuantizationConfig:
   # How to execute the op after quantization.
   execution_mode: OpExecutionMode = OpExecutionMode.WEIGHT_ONLY
 
+  # For advanced users only. If set, the quantizer will ignore all op
+  # configuration checks and forcefully quantize this op according to the user
+  # instructions even if it's not supported in the TFLite runtime.
+  skip_checks: bool = False
+
   def __post_init__(self):
     if self.activation_tensor_config is None:
       return