fn
dequantize_op
int dequantize_op(const int & w, const int & scales, const int & biases, int group_size, int bits)Python wrappers
Public Python APIs implemented by this engine symbol.Reconstruct a float weight from its packed form (inverse of quantize_op).