AttributeError: 'Learner' object has no attribute 'pbar'

I use editable installations of fastai2 and fastcore, following the instruction: https://github.com/fastai/fastai2

The following code ran into the error:
Learner' object has no attribute 'pbar'

Any one encountered the same error?

fastcore version: 0.1.17
fastai2 version: 0.0.17
pytorch version: 1.4.0

Appendix:
Full code:

import fastai2
from fastai2.vision.all import *
import numpy as np
import torch

path = untar_data(URLs.MNIST_SAMPLE)
threes = (path/'train'/'3').ls().sorted()
sevens = (path/'train'/'7').ls().sorted()


stacked_threes = torch.stack(
    tensors=[torch.tensor(np.array(Image.open(i))).float() / 255. for i in threes],
    dim=0
)

stacked_sevens = torch.stack(
    tensors=[torch.tensor(np.array(Image.open(i))).float() / 255. for i in sevens],
    dim=0
)

train_x = torch.cat([stacked_threes, stacked_sevens]).view(-1, 28 * 28)
train_y = torch.tensor([1] * stacked_threes.shape[0] + [0] * stacked_sevens.shape[0]).unsqueeze(1)

train_dset = list(zip(train_x, train_y))

train_dl = DataLoader(train_dset, batch_size=256, shuffle=True)


def mnist_binary_loss(logits, targets):
    preds = logits.sigmoid()
    return torch.where(targets == 1, 1.0 - preds, preds).mean()


def batch_accuracy(logits, yb):
    correct = (logits > 0) == yb
    return correct.float().mean()


dls = DataLoader(train_dl)
learn = Learner(
    dls, 
    nn.Linear(28 * 28, 1), 
    opt_func=SGD,
    loss_func=mnist_binary_loss,
    metrics=batch_accuracy
)


learn.fit(20, 0.1)

@sgugger @jeremy any insight on this?

To get help and people answering your posts, please follow these guidelines. In this case you did not post the full stack trace, only the end, which seems to a fallback but not he actual cause of the error.

Thanks Sylvain! Will post a more complete description! Apology for the @

Any development on this issue as I am facing the same issue right now.

Any help will be appreciated.