Anypy

Highlights
:rocket: Ensure equal importance of loss penalties in terms of grad norms in a specific layer :rocket:

> Other weights can then be applied, e.g. to have a loss term that influences the weights 50% more than the other.

python
Given two losses
loss_a, loss_b

Compute the loss_b weight
loss_b_weight = calculate_adaptive_weight(loss_a, loss_b, layer)
loss_b_weighted = loss_b_weight * loss_b

So that loss_a and loss_b have equal importance in the specified layer grads
loss_a_gradnorm = torch.norm(torch.autograd.grad(loss_a, layer, retain_graph=True)[0])
loss_b_weighted_gradnorm = torch.norm(torch.autograd.grad(loss_b_weighted, layer, retain_graph=True)[0])
assert torch.allclose(loss_a_gradnorm, loss_b_weighted_gradnorm)



What's Changed
* Disable pivot sorting in latex df by lucmos in https://github.com/lucmos/anypy/pull/12
* Add function to compute adaptive loss weights by lucmos in https://github.com/lucmos/anypy/pull/13


**Full Changelog**: https://github.com/lucmos/anypy/compare/0.0.5...0.0.6

What's Changed
* Add Chrono utility by lucmos in https://github.com/lucmos/anypy/pull/8
* Add recipe of hydra cfg in notebook by lucmos in https://github.com/lucmos/anypy/pull/10
* Add recipe to convert svg fig into pdf by lucmos in https://github.com/lucmos/anypy/pull/9
* Increase flexibility of df tolatex fun by lucmos in https://github.com/lucmos/anypy/pull/11


**Full Changelog**: https://github.com/lucmos/anypy/compare/0.0.4...0.0.5

What's Changed
* Implement funcions to infer missing parameters for Conv2d and ConvTranspose2d by lucmos in https://github.com/lucmos/anypy/pull/5
* Optimize test for MetadataDatasetDict by lucmos in https://github.com/lucmos/anypy/pull/6
* Add dynamic nn builder and simplify dynamic AE by lucmos in https://github.com/lucmos/anypy/pull/7


**Full Changelog**: https://github.com/lucmos/anypy/compare/0.0.3...0.0.4

What's Changed
* add metadata dataset dict by crisostomi in https://github.com/lucmos/anypy/pull/2

New Contributors
* crisostomi made their first contribution in https://github.com/lucmos/anypy/pull/2

**Full Changelog**: https://github.com/lucmos/anypy/compare/0.0.2...0.0.3

What's Changed
* Use torch-cpu in github actions by lucmos in https://github.com/lucmos/anypy/pull/4
* Add dynamic AE wrt input size by lucmos in https://github.com/lucmos/anypy/pull/3


**Full Changelog**: https://github.com/lucmos/anypy/compare/0.0.1...0.0.2

What's Changed
* Add df to LaTeX table with meanstd conversion by lucmos in https://github.com/lucmos/anypy/pull/1

New Contributors
* lucmos made their first contribution in https://github.com/lucmos/anypy/pull/1

**Full Changelog**: https://github.com/lucmos/anypy/commits/0.0.1