_weights = [[glumpy.Image(w, vmin=-m, vmax=m) for w in ws]
                for ws in weights]

    W = 100 * (opts.cols + 1) + 4
    H = 100 * opts.rows + 4

    win = glumpy.Window(W, H)

    loaded = False
    updates = -1
    batches = 0.
    recent = collections.deque(maxlen=20)
    errors = [collections.deque(maxlen=20) for _ in range(10)]
    testset = [None] * 10
    trainset = dict((i, []) for i in range(10))
    loader = idx_reader.iterimages(opts.labels, opts.images, False)

    rbm = opts.model and pickle.load(open(opts.model, 'rb')) or rbm.RBM(
        28 * 28, opts.rows * opts.cols, opts.binary)

    trainer = rbm.Trainer(
        rbm,
        momentum=opts.momentum,
        target_sparsity=opts.sparsity,
    )

    def get_pixels():
        global loaded
        if not loaded and numpy.all([len(trainset[t]) > 10
                                     for t in range(10)]):
            loaded = True
Exemplo n.º 2
0
    visible_frames = [
        fig.add_figure(args.n + 1, args.n,
                       position=(args.n, r)).add_frame(aspect=1)
        for r in range(args.n)
    ]

    weight_frames = [
        fig.add_figure(args.n + 1, args.n, position=(c, r)).add_frame(aspect=1)
        for r in range(args.n) for c in range(args.n)
    ]

    loaded = False
    recent = collections.deque(maxlen=20)
    errors = [collections.deque(maxlen=20) for _ in range(10)]
    trainset = dict((i, []) for i in range(10))
    loader = idx_reader.iterimages(args.images, args.labels, True)

    Model = lmj.rbm.Convolutional if args.conv else lmj.rbm.RBM
    rbm = args.model and pickle.load(open(args.model, 'rb')) or Model(
        28 * 28, args.n * args.n, not args.gaussian)

    Trainer = lmj.rbm.ConvolutionalTrainer if args.conv else lmj.rbm.Trainer
    trainer = Trainer(rbm,
                      l2=args.l2,
                      momentum=args.momentum,
                      target_sparsity=args.sparsity)

    def get_pixels():
        global loaded
        if not loaded and all(len(v) > 50 for v in trainset.itervalues()):
            loaded = True
Exemplo n.º 3
0
    hiddens = glumpy.image.Image(_hiddens)
    weights = [glumpy.image.Image(w) for w in _weights]

    visible_frames = [
        fig.add_figure(args.n + 1, args.n, position=(args.n, r)).add_frame(aspect=1)
        for r in range(args.n)]

    weight_frames = [
        fig.add_figure(args.n + 1, args.n, position=(c, r)).add_frame(aspect=1)
        for r in range(args.n) for c in range(args.n)]

    loaded = False
    recent = collections.deque(maxlen=20)
    errors = [collections.deque(maxlen=20) for _ in range(10)]
    trainset = dict((i, []) for i in range(10))
    loader = idx_reader.iterimages(args.images, args.labels, True)

    Model = lmj.rbm.Convolutional if args.conv else lmj.rbm.RBM
    rbm = args.model and pickle.load(open(args.model, 'rb')) or Model(
        28 * 28, args.n * args.n, not args.gaussian)

    Trainer = lmj.rbm.ConvolutionalTrainer if args.conv else lmj.rbm.Trainer
    trainer = Trainer(rbm, l2=args.l2, momentum=args.momentum, target_sparsity=args.sparsity)

    def get_pixels():
        global loaded
        if not loaded and all(len(v) > 50 for v in trainset.itervalues()):
            loaded = True

        if loaded:
            t = rng.randint(10)
    _hiddens = [glumpy.Image(h) for h in hiddens]
    _weights = [[glumpy.Image(w, vmin=-m, vmax=m) for w in ws] for ws in weights]

    W = 100 * (opts.cols + 1) + 4
    H = 100 * opts.rows + 4

    win = glumpy.Window(W, H)

    loaded = False
    updates = -1
    batches = 0.0
    recent = collections.deque(maxlen=20)
    errors = [collections.deque(maxlen=20) for _ in range(10)]
    testset = [None] * 10
    trainset = dict((i, []) for i in range(10))
    loader = idx_reader.iterimages(opts.labels, opts.images, False)

    rbm = opts.model and pickle.load(open(opts.model, "rb")) or rbm.RBM(28 * 28, opts.rows * opts.cols, opts.binary)

    trainer = rbm.Trainer(rbm, momentum=opts.momentum, target_sparsity=opts.sparsity)

    def get_pixels():
        global loaded
        if not loaded and numpy.all([len(trainset[t]) > 10 for t in range(10)]):
            loaded = True

        if loaded and rng.random() < 0.99:
            t = rng.randint(10)
            pixels = trainset[t][rng.randint(len(trainset[t]))]
        else:
            t, pixels = loader.next()