About the first-order approximation

[tht106]

Hi, thank you for this fascinating work and providing a demo of MLDG.

Two quick questions:

1. Did you use the first-order approximation in the MLP version of MLDG. The codes in ops.py look like an operation of the first-order approximation.

`

    if not stop_gradient:
        grad_weight = autograd.grad(meta_loss, weight, create_graph=True)[0]

        if bias is not None:
            grad_bias = autograd.grad(meta_loss, bias, create_graph=True)[0]
            bias_adapt = bias - grad_bias * meta_step_size
        else:
            bias_adapt = bias

    else:
        grad_weight = Variable(autograd.grad(meta_loss, weight, create_graph=True)[0].data, requires_grad=False)

        if bias is not None:
            grad_bias = Variable(autograd.grad(meta_loss, bias, create_graph=True)[0].data, requires_grad=False)
            bias_adapt = bias - grad_bias * meta_step_size
        else:
            bias_adapt = bias

    return F.linear(inputs,
                    weight - grad_weight * meta_step_size,
                    bias_adapt)
else:
    return F.linear(inputs, weight, bias)`

2. I am also wondering the meaning of the parameter "--stop_gradient". What would happen when we set it ture?

[CinKKKyo]

Hi, thank you for this fascinating work and providing a demo of MLDG.

Two quick questions:

1. Did you use the first-order approximation in the MLP version of MLDG. The codes in ops.py look like an operation of the first-order approximation.

`

    if not stop_gradient:
        grad_weight = autograd.grad(meta_loss, weight, create_graph=True)[0]

        if bias is not None:
            grad_bias = autograd.grad(meta_loss, bias, create_graph=True)[0]
            bias_adapt = bias - grad_bias * meta_step_size
        else:
            bias_adapt = bias

    else:
        grad_weight = Variable(autograd.grad(meta_loss, weight, create_graph=True)[0].data, requires_grad=False)

        if bias is not None:
            grad_bias = Variable(autograd.grad(meta_loss, bias, create_graph=True)[0].data, requires_grad=False)
            bias_adapt = bias - grad_bias * meta_step_size
        else:
            bias_adapt = bias

    return F.linear(inputs,
                    weight - grad_weight * meta_step_size,
                    bias_adapt)
else:
    return F.linear(inputs, weight, bias)`

2. I am also wondering the meaning of the parameter "--stop_gradient". What would happen when we set it true?

The meaning of the parameter "--stop_gradient" also make me confused. Did you figure it out?

[CharmsGraker]

Hi, thank you for this fascinating work and providing a demo of MLDG.
Two quick questions:

1. Did you use the first-order approximation in the MLP version of MLDG. The codes in ops.py look like an operation of the first-order approximation.

`

    if not stop_gradient:
        grad_weight = autograd.grad(meta_loss, weight, create_graph=True)[0]

        if bias is not None:
            grad_bias = autograd.grad(meta_loss, bias, create_graph=True)[0]
            bias_adapt = bias - grad_bias * meta_step_size
        else:
            bias_adapt = bias

    else:
        grad_weight = Variable(autograd.grad(meta_loss, weight, create_graph=True)[0].data, requires_grad=False)

        if bias is not None:
            grad_bias = Variable(autograd.grad(meta_loss, bias, create_graph=True)[0].data, requires_grad=False)
            bias_adapt = bias - grad_bias * meta_step_size
        else:
            bias_adapt = bias

    return F.linear(inputs,
                    weight - grad_weight * meta_step_size,
                    bias_adapt)
else:
    return F.linear(inputs, weight, bias)`

2. I am also wondering the meaning of the parameter "--stop_gradient". What would happen when we set it true?

The meaning of the parameter "--stop_gradient" also make me confused. Did you figure it out?
setting stop_gradient=True is to avoid large budget when excuting meta-optimization.
In my opinion, if stop_gradient=True, the whole algorithm could be reckoned as training objections of F(theta), G(theta) alternatively.