migration to `pydantic>=2.*` (#105)

* Unify gaussian likelihoods

* remove print

* `pydantic` migration

* fix VAEGAN and planar flows

* black and isort

* cleaning

* prepare release

prepare release

distributed training (#74)

* [WIP] distributed

* [WIP] distributed training

* add script

* add smaller script

* fix device

* fix DDP model setting

* reorder model on device

* remove model.model...

* set print on process 0

* remove duplicate

* model folder set on main process

* test

* model.update fix

* exists_ok=True

* add exist_ok

* jz multinode fix

* fix batch size splitting

* isort and black

* [WIP] work on trainers

* [WIP] work on Distributed and trainers

* add tets example with adversarial trainer

* fix small issue

* fix master addr environ

* fix typo

* fix update with DDP

* udpate callback for distributed training

* diplay progress per process

* enhance display

* [WIP] make CoupledAdv distributed

* Cealn up trainers and add distributed training

* fix piwae tests

* fix test piwae

* fix some tests

* increase coverage

* increase coverage

* add predict on main process

* apply balck and isort

* update notebooks with batch_size

* update reproducibility scripts

* clean up

* isort & black

* update README

* remove assert 0

* update distributed script

* add wandb

* update script

* log only on main process

* test batch size

* loss dubugging

* test with AE

* test with adaptive batchsize

* test with larger batch size

* benchmark

* benchmark perf

* remove debug prints

* redece learning rate

* show results

* new net

* lr

* remove sigm

* lr

* epochs

* batch_size

* new test

* with sigm

* test

* test

* retest

* retest

* with rank

* test in trainer

* retest

* test

* test

* test no embedding

* test

* test distributed

* debuggin

* debug

* not learnable codebook

* fix typo

* contiguous

* fix issue

* test inplace

* no_grad(

* debug

* find unused

* debug

* test with dist_nn

* remove find_unused

* test with dist.nn

* chekc rank

* remove all_reduce

* test with ddp

* second all_reduce

* async

* add detach

* add detach

* test

* debug

* change

* with einsum

* contiguous

* remove parameter

* new test

* debug

* debug

* add barrier

* remove embeddings

* update code

* update

* update

* mass sanity check on all process

* revert to good VQVAE

* remove prints

* add dist backend to script

* reduce number of epoch in example

* udpate doc

* increase batch size in example

* add other script

* remove find_unused

* test without unused

* fix ununsed

* add num_workers option to Training config

* add num_workers to scripts

* test with embedding

* remove learned codebook

* grad accumulation for benchmark

* beanchmark

* add grad accumulation

* remove print

* benchmark

* remove num_workers

* add FFHQ to benchmark

* fix predict

* fix predict

* reduce number of samples in predict

* add parser

* add sigmoid

* update config

* add imagenet script

* convert img to RGB

* add sigmoid to decoder

* increase batch size

* change nets

* change nets

* add new script

* add convert to RGB

* update tests

* clean up

* prepare release

* update doc

* fix input_dim

* last figures

* doc fix

Integration of `comet_ml` (#56)

* add comet_ml callback

* fix typo

* add comet to readme

* add link to notebook

* train for more epochs

* add TrainingCallbacks in doc

* apply isort and black

* prepare release

* typing

* Update README.md

* Update README.md

* update gitignore

* add viz in notebooks

Multiply/Partially/Combination IWAE (#54)

* work on CIWAE and MIWAE

* add PIWAE

* minor change in Coupled Optimizer Trainer

* update repro script

* prepare repro

* add DynBinDataset to scripts

* fix number of training samples

* update test calbacks with trainer update

* fix issue in callback

* add eval model

* add eval mode

* add examples and update readme

* add reproduction results

* apply black and isort

* clean up scripts

* clean up scripts

* clean up scripts

* removed unecessary

* prepare release

PoincaréVAE (#51)

* [WIP] work on pvae

* [WIP] fix Hyperbolic geometry and distributions

* [WIP] add tests and work on PVAE

* increase coverage

* [WIP] fix device

* [WIP] work on PVAE

* [WIP] tesing the model

* test

* test

* test

* fix pvae

* work on repro

* change data processing in repro experiment

* Add PoincaréDisk Sampler

* add sampler tests

* fix device issue

* add tutorials and fix device setting

* add PVAE to doc

* minor change in sampler

* minor change

* small change in docs

* work on reproducibility

* add ref to readme

* clean up readme

* add Wrapped PVAE results

* black and isort

* update README

* remove not needed

* update repro

* isort

Interpolations (#38)

* add interpolate and reconstruct methods

* update doc

* update tests

* update demo

* black & isort

* prepare release

* update README

Fix hub model card (#33)

* fix missing model_card.md

* fix missing model_card

* update manifest.in

bumping to python3.7+ (#29)

* bump to python3.7+

* prepare release

* update wandb example

Integration with HuggingFace Hub (#28)

* vaegan

* remove test_vae

* add auto config

* start hf hub integration

* model reloading refacto

* add hf hub to AutoModel

* add nf to AutoConfig

* remove typo

* add wandb tutorial

* add hf hub tutorial

* update gitignore

* Update README

* Update README

* Update README

* Update README

* remove hf hub and wandb from coverage compute

* update demo

* update tests with AutoModel

* remove wandb and hf hub from coverage compute

* update README

* black and isort formatting

* update demo

* add install command to demo

* add finish to wandb callback

* add test AutoModel for NF

* add install comands

* typo correction

* black

* fix typo

* fix logging in load_from_hf_hub

* add ModelOutput to AutoModel

* switch from dill to cloudpickle

* replace pickle by pickle5

* add pickle security to load_from_hf_hub

* fix pickling outside of __main__ scope

* black & isort

* add env config

* update test with env saving

* update setup

* isort & black

* add check to hf_load

* fix typo

* prepare release 0.0.2

* fix typo