static public Transformer Transformer(long d_model = 512, long nhead = 8, long num_encoder_layers = 6, long num_decoder_layers = 6, long dim_feedforward = 2048, double dropout = 0.1, Activations activation = nn.Activations.ReLU)
{
var res = THSNN_Transformer_ctor(d_model, nhead, num_encoder_layers, num_decoder_layers, dim_feedforward, dropout, (long)activation, out var boxedHandle);
if (res == IntPtr.Zero)
{
torch.CheckForErrors();
}
return(new Transformer(res, boxedHandle));
}
public void activate_array_kernel(XArray x, int offset, int n, Activations a)
{
amp.Ops.activate_array_kernel(x, offset, n, (int)a);
}
public void gradient_array_kernel(XArray x, int offset, int n, Activations a, XArray delta)
{
amp.Ops.gradient_array_kernel(x, offset, n, (int)a, delta);
}
public Activation(Activations act_type)
{
Set("ActType", act_type);
}
