fastaidocsprint / fastai

Documentation Sprint for the fastai deep learning library
http://fastaidocsprint.github.io/fastai
Apache License 2.0
15 stars 16 forks source link

Docment base optimizer optimizer #68

Closed tyoc213 closed 2 years ago

tyoc213 commented 2 years ago

This PR is only to add docmetns for _BaseOptimizer and Optimizer, hope it is a good starting point.

Q: I have added all the changes taht did pop up after nbdev_clean_nbs; nbdev_build_lib; nbdev_build_docs --fname nbs/12_optimizer.ipynb of line numbers of links to src, should I only let the table generation? (I can remove them if needed)

I will be making amends over same commit so all goes in 1 commit, thanks for the observations and feedback!

review-notebook-app[bot] commented 2 years ago

Check out this pull request on  ReviewNB

See visual diffs & provide feedback on Jupyter Notebooks.


Powered by ReviewNB

tyoc213 commented 2 years ago

This is for https://github.com/muellerzr/fastai-docment-sprint/issues/33

tyoc213 commented 2 years ago

mmm, trying to use delegates I go this err

converting: nbs/12_optimizer.ipynb
An error occurred while executing the following cell:
------------------
from nbdev.showdoc import show_doc
from fastai.optimizer import *
------------------

---------------------------------------------------------------------------
NameError                                 Traceback (most recent call last)
Input In [1], in <module>
      1 from nbdev.showdoc import show_doc
----> 2 from fastai.optimizer import *

File ~/Documents/git/hub/fastai-docment-sprint/fastai/optimizer.py:12, in <module>
      9 from .torch_basics import *
     11 # Cell
---> 12 class _BaseOptimizer():
     13     "Common functionality between `Optimizer` and `OptimWrapper`"
     14     def all_params(self,
     15         n=slice(None), # Where to slice the model or all as default
     16         with_grad=False # Get all param tuples. If `True` select only those with a gradient
     17     ):

File ~/Documents/git/hub/fastai-docment-sprint/fastai/optimizer.py:61, in _BaseOptimizer()
     59     self._set_hyper(k, v)
     60 def unfreeze(self): self.freeze_to(0)
---> 61 @delegates(_BaseOptimizer.set_hyper)
     62 def set_hypers(self,
     63     **kwargs
     64 ):
     65     L(kwargs.items()).starmap(self.set_hyper)
     66 def _set_hyper(self,
     67     k, # Hyper parameter key
     68     v # Hyper parameter value
     69 ):

NameError: name '_BaseOptimizer' is not defined
NameError: name '_BaseOptimizer' is not defined

Conversion failed on the following:
12_optimizer.ipynb

So Im removing it for now, how would be the correct way?

tyoc213 commented 2 years ago

Did update the PR, I left one of the comments untyped as importing Callback did throw errors like circular imports and things like that, so I leave only in the comment.

I also didnt use the delegates cause it is throwing that error above.

tyoc213 commented 2 years ago

@tmabraham @warner-benjamin hopefully this is OK

tyoc213 commented 2 years ago

latest fix @tmabraham @warner-benjamin