quant-forgeQuantKernelBridge validates contiguous memory layout (np.ascontiguousarray), positive even group_size divisibility (in_features % group_size == 0), and exact input dimension matching (x.shape[1] == qweight.in_features) in Python before crossing the C ABI boundary into csrc/quant_kernels.cpp.ge/le bounds on tensor dimensions, batch sizes, and accelerator identifiers.Please report security issues responsibly to sendemailtonagarjuna@gmail.com.