Why not accumulate loss and then take derivative in MAML?

Why do you not do this:

def inner_loop2():
    n_inner_iter = 5
    inner_opt = torch.optim.SGD(net.parameters(), lr=1e-1)

    qry_losses = []
    qry_accs = []
    meta_opt.zero_grad()
    meta_loss = 0
    for i in range(task_num):
        with higher.innerloop_ctx(
            net, inner_opt, copy_initial_weights=False
        ) as (fnet, diffopt):
            # Optimize the likelihood of the support set by taking
            # gradient steps w.r.t. the model's parameters.
            # This adapts the model's meta-parameters to the task.
            # higher is able to automatically keep copies of
            # your network's parameters as they are being updated.
            for _ in range(n_inner_iter):
                spt_logits = fnet(x_spt[i])
                spt_loss = F.cross_entropy(spt_logits, y_spt[i])
                diffopt.step(spt_loss)

            # The final set of adapted parameters will induce some
            # final loss and accuracy on the query dataset.
            # These will be used to update the model's meta-parameters.
            qry_logits = fnet(x_qry[i])
            qry_loss = F.cross_entropy(qry_logits, y_qry[i])
            qry_losses.append(qry_loss.detach())
            qry_acc = (qry_logits.argmax(
                dim=1) == y_qry[i]).sum().item() / querysz
            qry_accs.append(qry_acc)

            # Update the model's meta-parameters to optimize the query
            # losses across all of the tasks sampled in this batch.
            # This unrolls through the gradient steps.
            #qry_loss.backward()
            meta_loss += qry_loss

    qry_losses = sum(qry_losses) / task_num
    qry_losses.backward()
    meta_opt.step()
    qry_accs = 100. * sum(qry_accs) / task_num
    i = epoch + float(batch_idx) / n_train_iter
    iter_time = time.time() - start_time

instead of what you have:

def inner_loop1():
    n_inner_iter = 5
    inner_opt = torch.optim.SGD(net.parameters(), lr=1e-1)

    qry_losses = []
    qry_accs = []
    meta_opt.zero_grad()
    for i in range(task_num):
        with higher.innerloop_ctx(
            net, inner_opt, copy_initial_weights=False
        ) as (fnet, diffopt):
            # Optimize the likelihood of the support set by taking
            # gradient steps w.r.t. the model's parameters.
            # This adapts the model's meta-parameters to the task.
            # higher is able to automatically keep copies of
            # your network's parameters as they are being updated.
            for _ in range(n_inner_iter):
                spt_logits = fnet(x_spt[i])
                spt_loss = F.cross_entropy(spt_logits, y_spt[i])
                diffopt.step(spt_loss)

            # The final set of adapted parameters will induce some
            # final loss and accuracy on the query dataset.
            # These will be used to update the model's meta-parameters.
            qry_logits = fnet(x_qry[i])
            qry_loss = F.cross_entropy(qry_logits, y_qry[i])
            qry_losses.append(qry_loss.detach())
            qry_acc = (qry_logits.argmax(
                dim=1) == y_qry[i]).sum().item() / querysz
            qry_accs.append(qry_acc)

            # Update the model's meta-parameters to optimize the query
            # losses across all of the tasks sampled in this batch.
            # This unrolls through the gradient steps.
            qry_loss.backward()

    meta_opt.step()
    qry_losses = sum(qry_losses) / task_num
    qry_accs = 100. * sum(qry_accs) / task_num
    i = epoch + float(batch_idx) / n_train_iter
    iter_time = time.time() - start_time

https://github.com/facebookresearch/higher/blob/e45c1a059e39a16fa016d37bc15397824c65547c/examples/maml-omniglot.py#L130

---

https://stackoverflow.com/questions/62394411/why-not-accumulate-query-loss-and-then-take-derivative-in-maml-with-pytorch-and

[egrefen]

The answer to your stack overflow question is the right answer. For any x and theta, grad(sum(x), theta) = sum(grad(x, theta)), or in latex:

\nabla_\theta \sum_i f_\theta(x_i) = \sum_i \nabla_\theta f_\theta(x_i)

[renesax14]

The answer to your stack overflow question is the right answer. For any x and theta, grad(sum(x), theta) = sum(grad(x, theta)), or in latex:

\nabla_\theta \sum_i f_\theta(x_i) = \sum_i \nabla_\theta f_\theta(x_i)

also, why don’t you do:

meta_loss += qrt_loss.detach()

and instead you track them all in a list?

No particular reason. I guess this allows you to plot task losses separately if you wanted, or measure mean/variance rather than just mean?

No, what I mean is why not use .item() or .data.

Why .detach() in particular