Code Monkey home page Code Monkey logo

meta-learning-lstm-pytorch's Issues

Optimization of `self.cI`

Thank you for sharing the reimplementation and for the clean and organized code. There is an issue with the state of the meta learner self.cI. It is created as a parameter of the metalarner, and thusly getting optimized by Adam here. It should be passed as a state to avoid such behavior.

Thanks

AttributeError

grad = torch.cat([p.grad.data.view(-1) / args.batch_size for p in learner_w_grad.parameters()], 0)
AttributeError: 'NoneType' object has no attribute 'data'
Has anyone had this problem？

Why are is loss being preprocessed?

Why are is loss being preprocessed? it's already a log quantity and applying something like log(|x|)/p seems very weird to me. Any insights?

issues

Hello, I am reproducing your code and the following question appears. Could the author help to solve it?Thank you very much!

self.stats['train']['loss'].append(kwargs['loss'])
KeyError: 'train'

Meta Test Function

Hi Mr. Mark,

Thank you first for sharing your organized code and reimplementing an important paper.
I have a question regarding the meta-test function, though. Why is there a call for meta-training there? What did I miss?

Thanks again,
Nora

Why aren't all images processed the same in the meta-sets?

see there is a brightness thing for the meta-train-set:

def prepare_data(args):

    normalize = transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
    
    train_set = EpisodeDataset(args.data_root, 'train', args.n_shot, args.n_eval,
        transform=transforms.Compose([
            transforms.RandomResizedCrop(args.image_size),
            transforms.RandomHorizontalFlip(),
            transforms.ColorJitter(
                brightness=0.4,
                contrast=0.4,
                saturation=0.4,
                hue=0.2),
            transforms.ToTensor(),
            normalize]))

    val_set = EpisodeDataset(args.data_root, 'val', args.n_shot, args.n_eval,
        transform=transforms.Compose([
            transforms.Resize(args.image_size * 8 // 7),
            transforms.CenterCrop(args.image_size),
            transforms.ToTensor(),
            normalize]))

    test_set = EpisodeDataset(args.data_root, 'test', args.n_shot, args.n_eval,
        transform=transforms.Compose([
            transforms.Resize(args.image_size * 8 // 7),
            transforms.CenterCrop(args.image_size),
            transforms.ToTensor(),
            normalize]))

    train_loader = data.DataLoader(train_set, num_workers=args.n_workers, pin_memory=args.pin_mem,
        batch_sampler=EpisodicSampler(len(train_set), args.n_class, args.episode))

    val_loader = data.DataLoader(val_set, num_workers=2, pin_memory=False,
        batch_sampler=EpisodicSampler(len(val_set), args.n_class, args.episode_val))

    test_loader = data.DataLoader(test_set, num_workers=2, pin_memory=False,
        batch_sampler=EpisodicSampler(len(test_set), args.n_class, args.episode_val))

meta-learning-lstm-pytorch/dataloader.py

Line 82 in fcc68ba

def prepare_data(args):

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.