Value Error when importing blurr.modeling.all and blurr.data.all

[JeremyOBrien16]

After pip install in Colab, import blurr.modeling.all and blurr.data.all are throwing errors. Did not encounter errors last week, not sure when they began. Has there been a syntax change for setup?

---

ValueError Traceback (most recent call last)
/usr/local/lib/python3.6/dist-packages/pandas/core/indexes/range.py in get_loc(self, key, method, tolerance)
354 try:
--> 355 return self._range.index(new_key)
356 except ValueError as err:

ValueError: 1 is not in range

The above exception was the direct cause of the following exception:

KeyError Traceback (most recent call last)
6 frames
in ()
----> 1 from blurr.modeling.all import *

/usr/local/lib/python3.6/dist-packages/blurr/modeling/all.py in ()
----> 1 from ..utils import *
2 from .core import *
3 from .question_answering import *
4 from .token_classification import *
5 from .text2text.core import *

/usr/local/lib/python3.6/dist-packages/blurr/utils.py in ()
182
183 # Cell
--> 184 BLURR_MODEL_HELPER = ModelHelper()
185
186 # Cell

/usr/local/lib/python3.6/dist-packages/blurr/utils.py in call(self, *args, **kwargs)
25
26 def call(self, *args, **kwargs):
---> 27 if self._instance == None: self._instance = self._cls(*args, **kwargs)
28 return self._instance
29

/usr/local/lib/python3.6/dist-packages/blurr/utils.py in init(self)
59 model_type_df = self._df[(self._df.functional_area == 'modeling')].class_name.str.split('For', n=1, expand=True)
60
---> 61 model_type_df[1] = np.where(model_type_df[1].notnull(),
62 'For' + model_type_df[1].astype(str),
63 model_type_df[1])

/usr/local/lib/python3.6/dist-packages/pandas/core/frame.py in getitem(self, key)
2904 if self.columns.nlevels > 1:
2905 return self._getitem_multilevel(key)
-> 2906 indexer = self.columns.get_loc(key)
2907 if is_integer(indexer):
2908 indexer = [indexer]

/usr/local/lib/python3.6/dist-packages/pandas/core/indexes/range.py in get_loc(self, key, method, tolerance)
355 return self._range.index(new_key)
356 except ValueError as err:
--> 357 raise KeyError(key) from err
358 raise KeyError(key)
359 return super().get_loc(key, method=method, tolerance=tolerance)

KeyError: 1

[LR-Silva]

Same error here. It was working fine last week and now I cannot import.

[ohmeow]

I'll take a look ... either something in fastai or huggingface has changed. What versions of each are you both working with?

[LR-Silva]

I tried both the default version from pypi (0.0.18) or installing from source (0.0.19) with pip install -e ".[dev]".
I got the same error when using on Colab and on my personal machine with both versions.

[JeremyOBrien16]

Thanks for the quick response!

fastai 1.0.61
transformers 4.0.0

[JeremyOBrien16]

Looks like huggingface pushed a transformers update a week ago: https://github.com/huggingface/transformers/releases

[JeremyOBrien16]

Rolled back to transformer==3.5.1 and install / imports working fine.

[ohmeow]

Thanks for the info. Btw, fastai needs to be >= 2.15 as well. I'll find some time to fix things to work against the latest this week. wg

…

On Mon, Dec 7, 2020 at 6:29 PM JeremyOBrien16 ***@***.***> wrote: Rolled back to transformer==3.5.1 and install / imports working fine. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#21 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAADNMEXAK5YXEEJ5VMXNJ3STWFPNANCNFSM4UPP5H6A> .

[LR-Silva]

It worked here for me.
Thanks 👍

[tyoc213]

It worked here for me.
Thanks +1

How?

For me is still failing?... well, installed from pip

python        : 3.8.5
fastai        : 2.1.8
fastprogress  : 0.2.7
torch         : 1.7.0
transformers  : 3.5.1

Oh yeah, it works with

Thx
pip install transformers==3.5.1 
pip install ohmeow-blurr==0.0.18

[HenryDashwood]

It's actually quite an easy fix I think. Huggingface have just buried their model files in a couple of extra layers of directories since the src folder was getting quite big. So in ModelHelper's __init__() you need to split self._df.module_part_3 instead of self._df.module_part_1. I got it to work on my fork here: https://github.com/HenryDashwood/blurr/commit/dececd4e706129694b25ee80c15dfb61ffadf9a9#diff-45d6eb3657aeaa14c3dacc16d9278dda8d6aa21c61456f9a03d80e34d368eeec

Unfortunately there are a few more version 4 breaking changes which come up when I try to test the notebooks. I might have some time to take a crack at them this week if it would be helpful?

[ohmeow]

Done.

The latest fast.ai and transformers libs actually broke a few things ... but all is working now; all tests are passing.

Closing this out ... lmk if anything is still broken for you all.

[bappctl]

[Works fine with]
python: 3.7.13
pytorch: 1.7.1+cu110
fastai: 2.2.5
transformers: 4.3.3
ohmeow-blurr==0.0.24

and to use BLURR_MODEL_HELPER. instead of BLURR.

[ohmeow]

@bappctl : try installing the latest version of blurr ... pip install ohmeow-blurr==1.0.2

There were some of the old imports being cached in there that nbdev didn't clear out when I made the initial release. That's been fixed and the namespaces have changed. All the NLP/Text bits will be under blurr.text.

Feel free to close this out once things are working (or lmk if you are still having issues).

[waisyousofi]

You are right ,

ohmeow-blurr==0.0.24
and
BLURR_MODEL_HELPER instead of BLURR

worked for me.

thanks✌️

[ohmeow]

Closing this out as v.1 is out and the code above is obsolete.