Skip to content

Latest commit

 

History

History
137 lines (101 loc) · 4.52 KB

CHANGELOG.md

File metadata and controls

137 lines (101 loc) · 4.52 KB
Raw

Change log

Best viewed here.

wax 0.6.4 (April 22 2023)

  • upgrade CI to python 3.9
  • remove numba functions from the core package

wax 0.6.3 (November 18 2022)

  • import tree_map from tree_util
  • fix VMap module

wax 0.6.2 (November 14 2022)

  • Fix deployement of 0.6.1

wax 0.6.1 (November 14 2022)

  • patch vmap_lift_with_state (now rely on hk.experimental.lift_with_state)

wax 0.6.0 (November 13 2022)

  • add stateful.py module with vmap_lift_with_state and unroll_lift_with… (#65)
  • remove tensorflow and pytorch from requirements
  • remove dependency with eagerpy (#62)
  • correct imports of tree utils to avoid FutureWarnings (#63)
  • replace jax.tree_multi_map (removed in JAX release v0.3.16) with jax.tree_map
  • faster unroll: do not propagate params in scan state (#61)

wax 0.5.0 (May 16 2022)

  • update to Jax v0.3.10 (#59)
  • add optimizers (#56), (#58)

wax 0.4.0 (April 5 2022)

  • EWMA alignement with pandas and speedup (#53) This adds the options: * com * min_periods * ignore_na * return_info

  • [wax_numba] add an implementation of the ewma in numba extending the one of pandas with the additional modes we have in wax:

    • adjust='linear'
    • initial_value parameter
    • a state management for online usages and warm-start of the ewma.
    • add numba to requirements

  • [EWMA] use log1com as a haiku parameter to ease training with gradient descent.

  • Align EWMCov and EWMVar with EWMA (#55)

  • [PctChange] correct PctChange module to align with pandas behavior. Introduce fillna_zero option.

wax 0.3.2 (February 25 2022)

  • [modules] faster EWMA in adjust=True mode.

wax 0.3.1 (January 4 2022)

  • [unroll] split rng in two rng keys.

wax 0.3.0 (December 16 2021)

  • [VMap] VMap module works in contexts without PRNG key

  • [online optimizer] ; refactor

    • refactor OnlineOptimizer outputs: only return loss, model_info, opt_loss by default. New option 'return_params' to return params in outputs
    • OnlineOptimizer returns updated params if return_params is set to True

  • [newton optimizer]: use NamedTuple instead of base.OptState

  • [unroll] propagate pbar argument to static_scan

  • [unroll] Renew the PRNG key in the unroll operations

  • refactor usage of OnlineOptimizer in notebooks

  • format with laster version of black

  • require jax<=0.2.21

  • add graphviz to optional dependencies

  • upgrade jupytext to 1.13.3

  • use python 3.8 in CI and documentation

wax 0.2.0 (October 20 2021)

  • Documentation:

    • New notebook : 07_Online_Time_Series_Prediction
    • New notebook : 08_Online_learning_in_non_stationary_environments

  • API modifications:

    • refactor accessors and stream
    • GymFeedback now assumes that agent and env return info object
    • OnlineSupervisedLearner action is y_pred, loss and params are returned as info

  • Improvements:

    • introduce general unroll transformation.
    • dynamic_unroll can handle Callable objects
    • UpdateOnEvent can handle any signature for functions
    • EWMCov can handle the x and y arguments explicitly
    • add initial action option to GymFeedback

  • New Features:

    • New module UpdateParams
    • New module SNARIMAX, ARMA
    • New module OnlineOptimizer
    • New module VMap
    • add grads_fill_nan_inf option to OnlineSupervisedLearner
    • Introduce unroll_transform_with_state following Haiku API.
    • New function auto_format_with_shape and tree_auto_format_with_shape
    • New module Ffill
    • New module Counter

  • Deprecate:

    • deprecate dynamic_unroll and static_unroll, refactor their usages.

  • Fixes:

    • Simplify Buffer to work only on ndarrays (implementation on pytrees were too complex)
    • EWMA behave corectly with gradient
    • MaskStd behave correctly with gradient
    • correct encode_int64 when working on int32
    • update notebook 06_Online_Linear_Regression and add it to run-notebooks rule
    • correct pct_change to behave correctly when input data has nan values.
    • correct eagerpy test for update of tensorflow, pytorch and jax
    • remove duplicate license comments
    • use numpy.allclose instsead of jax.numpy.allclose for comparaison of non Jax objects
    • update comment in notebooks : jaxlib==0.1.67+cuda111 to jaxlib==0.1.70+cuda111
    • fix jupytext dependency
    • add seaborn as optional dependency

wax 0.1.0 (June 14 2021)

  • First realease.