Hey @muellerzr, were you able to get SOTA on ImageWoof? I tried running your notebook and tried replicating lessw2020, but no success…
There are issues (like I mentioned in the notebook). We’re working on figuring it out slowly. We think it has something to do with the transforms. Know that all the implementations are correct.
Also there are issues with the tabular notebook. I’m trying to get to it this week. (They updated the tabular API)
Is there any specific thread where you are discussing what might be the problems there? I’m very interested in participating
See here: Meet Ranger - RAdam + Lookahead optimizer
@morgan has been working hard at cracking this
We had a problem in the cnn head that was fixed yesterday. So things might work better now.
Can confirm it did help a bit Got about 72% on average of five runs, I’ll update the SOTA notebook too. Thanks Jeremy!
So a quick update to these, (I promise I’m not ignoring your requests!) I am setting up my lecture material for my study group. The new version will be live streamed so anyone can get involved and to my knowledge will be the first using fastai 2.0. I plan on starting in mid-January next year. I will be including all of the types of models mentioned here and walking through them. For those interested, keep an eye out here in the next week or so for an official mega-thread to discuss it (and if you have any questions feel free to DM me!) This new version for streaming will be new to me, but I’m hoping that it’ll go swell
Edit: Also these current notebooks are outdated In terms of how to install the library, I’ll do my best to update them in the next week or two depending time. In the meantime, follow the install directions here: Fastai-v2 - read this before posting please! 😊
Edit x2: All notebooks are updated and Rossman is there too
Hi muellerzr Hope your having a fun day!
Your output and support always inspire me.
Long may it continue mrfabulous1
count me in!
Absolutely looking forward to this. I had started out with Jeremy’s walkthrough lectures but had to drop off after the last walkthrough. So haven’t kept pace with fastai 2.0. would be great to learn it from you. Please keep us all in the loop
@sgugger just finished up all of his notebooks as well for the course! You can find all of those here:
Once I’m back from San Francisco I’ll make the megathread for the study group (expect Friday or Saturday)
Wait you’re in SF, and didn’t come to visit!?! I hope you had a nice trip. Sorry about all the rain…
Rain is much nicer here than Florida! (Less dense. I love it!)
I’ve made the study group thread, you can find it here: A walk with fastai2 - Study Group and Online Lectures Megathread
I‘m looking forward to seeing everyone there
As I update the course nb’s I’ll bleed over any other notebooks/techniques into the Practical repo so anyone that’s starred this one won’t get confused on which to follow
Hi muellerzr it sounds like you have been busy!
Question: How many hours a day you sleep
Thanks for the notebooks they will help many people who are little slower than yourself! (That includes me )
Cheers mrfabulous1
I promise a decent amount (5-6 hrs or so, college!) I’m happy to hear they help
For anyone looking at this later, I moved all the notebooks to a different repo for the study group:
Hi @muellerzr, these walk throughs are very helpful. I am currently attempting your 09a_IMDB_Sample.ipynb walkthough, and receive an error at the following line
dbunch_lm = imdb_lm.dataloaders(df_tok, bs=64, seq_len=72)
The error is below. Any thoughts on a possible solution?
dbunch_lm = imdb_lm.dataloaders(df_tok, bs=64, seq_len=72)
File "/home/cdparks/anaconda3/envs/fastai-v2/lib/python3.7/site-packages/fastai2/data/block.py", line 98, in dataloaders
dsets = self.datasets(source)
File "/home/cdparks/anaconda3/envs/fastai-v2/lib/python3.7/site-packages/fastai2/data/block.py", line 95, in datasets
return Datasets(items, tfms=self._combine_type_tfms(), splits=splits, dl_type=self.dl_type, n_inp=self.n_inp, verbose=verbose)
File "/home/cdparks/anaconda3/envs/fastai-v2/lib/python3.7/site-packages/fastai2/data/core.py", line 261, in __init__
self.tls = L(tls if tls else [TfmdLists(items, t, **kwargs) for t in L(ifnone(tfms,[None]))])
File "/home/cdparks/anaconda3/envs/fastai-v2/lib/python3.7/site-packages/fastai2/data/core.py", line 261, in <listcomp>
self.tls = L(tls if tls else [TfmdLists(items, t, **kwargs) for t in L(ifnone(tfms,[None]))])
File "/home/cdparks/anaconda3/envs/fastai-v2/lib/python3.7/site-packages/fastcore/foundation.py", line 41, in __call__
res = super().__call__(*((x,) + args), **kwargs)
File "/home/cdparks/anaconda3/envs/fastai-v2/lib/python3.7/site-packages/fastai2/data/core.py", line 202, in __init__
self.setup(train_setup=train_setup)
File "/home/cdparks/anaconda3/envs/fastai-v2/lib/python3.7/site-packages/fastai2/data/core.py", line 215, in setup
self.tfms.setup(self, train_setup)
File "/home/cdparks/anaconda3/envs/fastai-v2/lib/python3.7/site-packages/fastcore/transform.py", line 179, in setup
for t in tfms: self.add(t,items, train_setup)
File "/home/cdparks/anaconda3/envs/fastai-v2/lib/python3.7/site-packages/fastcore/transform.py", line 182, in add
t.setup(items, train_setup)
File "/home/cdparks/anaconda3/envs/fastai-v2/lib/python3.7/site-packages/fastcore/transform.py", line 78, in setup
return self.setups(getattr(items, 'train', items) if train_setup else items)
File "/home/cdparks/anaconda3/envs/fastai-v2/lib/python3.7/site-packages/fastcore/dispatch.py", line 98, in __call__
return f(*args, **kwargs)
File "/home/cdparks/anaconda3/envs/fastai-v2/lib/python3.7/site-packages/fastai2/text/data.py", line 35, in setups
count = dsets.counter if hasattr(dsets, 'counter') else Counter(p for o in dsets for p in o)
File "/home/cdparks/anaconda3/envs/fastai-v2/lib/python3.7/collections/__init__.py", line 568, in __init__
self.update(*args, **kwds)
File "/home/cdparks/anaconda3/envs/fastai-v2/lib/python3.7/collections/__init__.py", line 655, in update
_count_elements(self, iterable)
File "/home/cdparks/anaconda3/envs/fastai-v2/lib/python3.7/site-packages/fastai2/text/data.py", line 35, in <genexpr>
count = dsets.counter if hasattr(dsets, 'counter') else Counter(p for o in dsets for p in o)
File "/home/cdparks/anaconda3/envs/fastai-v2/lib/python3.7/site-packages/fastai2/data/core.py", line 208, in <genexpr>
def __iter__(self): return (self[i] for i in range(len(self)))
File "/home/cdparks/anaconda3/envs/fastai-v2/lib/python3.7/site-packages/fastai2/data/core.py", line 242, in __getitem__
return self._after_item(res) if is_indexer(idx) else res.map(self._after_item)
File "/home/cdparks/anaconda3/envs/fastai-v2/lib/python3.7/site-packages/fastai2/data/core.py", line 206, in _after_item
def _after_item(self, o): return self.tfms(o)
File "/home/cdparks/anaconda3/envs/fastai-v2/lib/python3.7/site-packages/fastcore/transform.py", line 185, in __call__
def __call__(self, o): return compose_tfms(o, tfms=self.fs, split_idx=self.split_idx)
File "/home/cdparks/anaconda3/envs/fastai-v2/lib/python3.7/site-packages/fastcore/transform.py", line 138, in compose_tfms
x = f(x, **kwargs)
File "/home/cdparks/anaconda3/envs/fastai-v2/lib/python3.7/site-packages/fastcore/transform.py", line 72, in __call__
def __call__(self, x, **kwargs): return self._call('encodes', x, **kwargs)
File "/home/cdparks/anaconda3/envs/fastai-v2/lib/python3.7/site-packages/fastcore/transform.py", line 83, in _call
if not _is_tuple(x): return self._do_call(f, x, **kwargs)
File "/home/cdparks/anaconda3/envs/fastai-v2/lib/python3.7/site-packages/fastcore/transform.py", line 88, in _do_call
return x if f is None else retain_type(f(x, **kwargs), x, f.returns_none(x))
File "/home/cdparks/anaconda3/envs/fastai-v2/lib/python3.7/site-packages/fastcore/dispatch.py", line 98, in __call__
return f(*args, **kwargs)
TypeError: 'list' object is not callable
The language notebooks have not been updated whatsoever, I wouldn’t expect them to be until a few weeks before the language section sorry! (Just what I can handle university time wise). I’d recommend looking at the course notebooks under fastai2/course
Have a look at my notebooks here that I used for a recent Kaggle competition, they’re based on the IMDB tutorial on dev.fast.ai, so you should give that tutorial a go too