momo1986 commented on issue #14672: Is it possible for mxnet to save the best model and early stopping URL: https://github.com/apache/incubator-mxnet/issues/14672#issuecomment-482406117 > @momo1986 Glad that you are experimenting with the framework, I assume you are using a for loop setup for the epochs?could you share with us a snippet of code, on what you are trying to achieve? I guess that will help us answer this question better. > > @mxnet-label-bot add [Question] Hello, @vrakesh . Here is my code. ``` ############################################################################################# # Try use GPU for training try: a = mx.nd.zeros((1,), ctx=mx.gpu(0)) ctx = [mx.gpu(0)] except: ctx = [mx.cpu()] ############################################################################################# # Start training(finetuning) net.collect_params().reset_ctx(ctx) trainer = gluon.Trainer( net.collect_params(), 'sgd', {'learning_rate': 0.001, 'wd': 0.0005, 'momentum': 0.9}) mbox_loss = gcv.loss.SSDMultiBoxLoss() ce_metric = mx.metric.Loss('CrossEntropy') smoothl1_metric = mx.metric.Loss('SmoothL1') for epoch in range(0, 100): ce_metric.reset() smoothl1_metric.reset() tic = time.time() btic = time.time() net.hybridize(static_alloc=True, static_shape=True) for i, batch in enumerate(train_data): batch_size = batch[0].shape[0] data = gluon.utils.split_and_load(batch[0], ctx_list=ctx, batch_axis=0) cls_targets = gluon.utils.split_and_load(batch[1], ctx_list=ctx, batch_axis=0) box_targets = gluon.utils.split_and_load(batch[2], ctx_list=ctx, batch_axis=0) with autograd.record(): cls_preds = [] box_preds = [] for x in data: cls_pred, box_pred, _ = net(x) cls_preds.append(cls_pred) box_preds.append(box_pred) sum_loss, cls_loss, box_loss = mbox_loss( cls_preds, box_preds, cls_targets, box_targets) autograd.backward(sum_loss) # since we have already normalized the loss, we don't want to normalize # by batch-size anymore trainer.step(1) ce_metric.update(0, [l * batch_size for l in cls_loss]) smoothl1_metric.update(0, [l * batch_size for l in box_loss]) name1, loss1 = ce_metric.get() name2, loss2 = smoothl1_metric.get() if i % 20 == 0: print('[Epoch {}][Batch {}], Speed: {:.3f} samples/sec, {}={:.3f}, {}={:.3f}'.format( epoch, i, batch_size/(time.time()-btic), name1, loss1, name2, loss2)) btic = time.time() ############################################################################################# # Save finetuned weights to disk net.save_parameters('ssd_512_mobilenet1.0_right_hand.params') ``` It does not change a lot from official gluon code. In the MXNET/Gluon forum in Chinese, there is a discussion with some demos to save the best check-point. However, I am very curious about a better solution to save the best parameter. Thanks for your answer. Momo
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services