AQLM CUDA support (#3287)

Co-authored-by: mgoin <michael@neuralmagic.com>

csrc/pybind.cpp

@@ -63,6 +63,8 @@ PYBIND11_MODULE(TORCH_EXTENSION_NAME, m) {
 
 // Quantization ops
 #ifndef USE_ROCM
+  ops.def("aqlm_gemm", &aqlm_gemm, "Quantized GEMM for AQLM");
+  ops.def("aqlm_dequant", &aqlm_dequant, "Decompression method for AQLM");
   ops.def("awq_gemm", &awq_gemm, "Quantized GEMM for AWQ");
   ops.def("marlin_gemm", &marlin_gemm, "Marlin Optimized Quantized GEMM for GPTQ");
   ops.def("awq_dequantize", &awq_dequantize, "Dequantization for AWQ");