pie

pie copied to clipboard

Fixed AttentionalDecoder truncating to 20 chars

4

When labels for a task are longer than 20 characters, they would be decoded as truncated to 20 characters. So instead we first look at the length of the longest...

PrinsINT

Referencing pull request #85 for the fix.

PrinsINT

dependency problems

5

Hi, I am trying to install PIE (python version 3.10) but I run into dependency problems: Is there a workaround? (venv) jesse@svprre10:~/workspace/pie$ pip install nlp-pie==0.3.8 Collecting nlp-pie==0.3.8 Using cached nlp_pie-0.3.8-py2.py3-none-any.whl...

JessedeDoes

Accuracy function

8

Actually duplicates in part #53 because I forgot to go back to master... This adds the ability to chose on which metric to fit in regard to devset: accuracy, f1,...

PonteIneptique

(improvement/AttentionalDecoder.predict_max) Avoid computation on rea…

7

…ched EOS for prediction (See #73) The current prediction time is quite slow, we agree that there might be room for improvement. After having a good look at it, it...

PonteIneptique

Bumps [pyyaml](https://github.com/yaml/pyyaml) from 5.1b3 to 5.4. Changelog Sourced from pyyaml's changelog. 5.4 (2021-01-19) yaml/pyyaml#407 -- Build modernization, remove distutils, fix metadata, build wheels, CI to GHA yaml/pyyaml#472 -- Fix for...

dependabot[bot]

Allow for using other Learning Rate Schedulers and Optimizers

5

Hey ! I started reading about some other optimizers, as things went through my news feed (stuff like [this](https://penseeartificielle.fr/meilleur-optimizer-ranger-radam-lookahead-fastai/) or [that](https://medium.com/@lessw/new-deep-learning-optimizer-ranger-synergistic-combination-of-radam-lookahead-for-the-best-of-2dc83f79a48d)). I ended up trying to implement it in pie...

PonteIneptique

Feature: allow picking whatever pytorch lr_scheduler

11

Here is a draft for the lr_scheduler params (related to #78) and how to avoid breaking backwards compatibility. Could you use it to test the results of the cosineannealing that...

emanjavacas

Improve prediction time (predict_max)

12

The current prediction time is quite slow, we agree that there might be room for improvement. After having a good look at it, it seemed clear that we were computing...

PonteIneptique

Allow for other Loss systems

2

Akin to https://github.com/emanjavacas/pie/issues/76, I would like to try Dice Loss for Linear tasks. It's a bit more tricky to implement as it is done in model, and not in Trainer....

PonteIneptique