Releases: Transpile-AI/ivy
Releases · Transpile-AI/ivy
Test Release v0.0.9.0
What's Changed
- fix: torch frontend allowing negative padding values by @Sam-Armstrong in #28501
- fix: ensure consant values are correct dtype in tf pad backend by @Sam-Armstrong in #28502
- fix: enabled the casting modes to type cast bools to ints when bool isn't supported by a function by @vedpatwardhan in #28503
- fix: ensure traced function return is turned to ivy arrays by @Ishticode in #28494
- fix: add unsupported dtypes to some tf backend fns by @Sam-Armstrong in #28504
- fix: tf backend range supporting float args with int dtype by @Sam-Armstrong in #28507
- fix: Ivy Failing Test: paddle - sorting.msort by @MuhammadNizamani in #28460
- fix: Ivy Failing Test: paddle - searching.where by @MuhammadNizamani in #28410
- fix: tf.function working with set item by @mattbarrett98 in #28481
- Fixed test for test_paddle_softplus by @ZJay07 in #28476
- fixed test: tensor.math.less by @ZJay07 in #28468
- fix: support return_indices=True in torch max_pool2d frontend by @Sam-Armstrong in #28537
- fix: remove incorrect unsupported dtype decorator from tf backend reshape by @Sam-Armstrong in #28538
- fix: remove key error in jax frontend tests by @Ishticode in #28513
- fix: Fixed
pointwise_ops.torch.real
for paddle backend by @Sai-Suraj-27 in #28115 - Fix Frontend Failing Test: numpy - tensor.paddle.Tensor.mean by @ZJay07 in #28486
- feat: Integrate F1 score function to frontend, align with sklearn met… by @muzakkirhussain011 in #28487
- fix(paddle_test): fixed dtype errors in
manipulation.paddle.tile
by @alt-shreya in #28478 - fix: fixed the logical_not not support the bfloat16 by @samthakur587 in #28455
- fix: fixed some failures in the README by @vedpatwardhan in #28548
- fix: paddle backend scatter_nd indices error and test by @Sam-Armstrong in #28552
- fix: added cuda installation to the GPU dockerfile for paddlepaddle to work in the image by @vedpatwardhan in #28554
- fix: added the missing tee command to record the logs of the tests that were run during the pre-release workflow by @vedpatwardhan in #28555
- fix: for jax backend type hint,
NativeArray
shouldn't be a tuple by @mattbarrett98 in #28543 - fix: downgraded tensorflow to 2.15.1 in the GPU docker image as 2.16.1 can't detect a GPU by @vedpatwardhan in #28578
- fix: numpy indexing in tf backend getitem by @mattbarrett98 in #28556
- fix: disable copy, out and inplace flags when trace testing by @Ishticode in #28595
- Fixed paddle.math.heaviside for all frontends by @ZJay07 in #28518
- Fixed math.paddle.conj for jax, numpy, tensorflow and torch by @ZJay07 in #28512
- Fix Frontend Failing Test: all frontends- math.paddle.trunc by @ZJay07 in #28506
- fix: haiku ivy module build working without needing backend set + extended test by @mattbarrett98 in #28586
- chore: removed the version hardcoding of ml-dtypes as tensorflow now works with the latest one by @vedpatwardhan in #28598
- fix: explicitly add haiku as target in demos, since flax is now used by default by @mattbarrett98 in #28599
- docs: updated the README to make the page more compact and clear by @vedpatwardhan in #28597
- fix: fixed the logical_or at tf frontend by @samthakur587 in #28585
- fix: fixed the greater_equal to tf frontend by @samthakur587 in #28588
- fix: ivy.linalg.matrix_norm() for paddle backend by @Kacper-W-Kozdon in #28500
- Update compiler.py 3c27196 by @vedpatwardhan in #28612
Full Changelog: 0.0.8.0...0.0.9.0
Test Release v0.0.8.0
What's Changed
- fix: Fixed Frontend Failing Test: paddle - non_linear_activation_functions.torch.nn.functional.leaky_relu by @aibenStunner in #28392
- Fixed the tests failure: numpy - pooling_functions.torch.nn.functional.max_pool2d by @Msahoo876 in #28401
- fix: Removed un-necessary workflows by @Sai-Suraj-27 in #28449
- feat: Updated
Ruff
to the latest version and fixed lint failure by @Sai-Suraj-27 in #28463 - fix: Frontend Failing Test: torch.linalg.matrix_rank by @ZenithFlux in #28390
- fix: torch frontend max/min to support dim and keepdim as arg or kwarg by @Sam-Armstrong in #28469
- feat(frontends): Implemented numpy fftn with tests by @AwkNinja in #26917
- feat: gradient numpy frontend by @sineha17 in #23104
- fix: use primary
get_item
implementation for 1d tensor queries with tf, and added a test for partial_mixed_handler to check for this by @mattbarrett98 in #28456 - fix: setting torch frontend tensor 'data' attribute by @Sam-Armstrong in #28473
- fix: torch.Tensor.to frontend failing with certain arg/kwarg combinations by @Sam-Armstrong in #28477
- fix: remove unnecessary dim expansion from ivy.interpolate by @Sam-Armstrong in #28484
- feat: Implement precision_score function and test aligned with sklearn metrics by @muzakkirhussain011 in #28407
- fix: updated ivy.Shape, added interpolate with the paddle backend and updated expand by @vedpatwardhan in #28471
- fix: fixing the tests for
ivy.Shape
method by @fnhirwa in #28470 - fix: Fixed
paddle.math.floor
for jax, torch, and tensorflow backends by @Sai-Suraj-27 in #28420 - fix: Docstring Edits by @Adlai-1 in #28424
- feat(backends): Added a primary implementation for flatten in tensorflow backend by @hmahmood24 in #28488
- fix: torch frontend max pooling to support optional batch dim by @Sam-Armstrong in #28490
- fix: tf backend unflatten not supporting dim size inference by @Sam-Armstrong in #28496
- feat: added more runners for the pre-release workflow to run tests for all priority functions by @vedpatwardhan in #28498
- fix(get-item): Use dynamic shape rather than static (x.shape) when possible in
_parse_query
by @hmahmood24 in #28489 - Update compiler.py 32c4e44 by @vedpatwardhan in #28499
New Contributors
- @Msahoo876 made their first contribution in #28401
- @ZenithFlux made their first contribution in #28390
- @sineha17 made their first contribution in #23104
Full Changelog: 0.0.7.5...0.0.8.0
Test Release v0.0.7.5
What's Changed
- fix: fixed
ivy.equal
for paddle backend support all dtype by @samthakur587 in #28426 - remove to_list call in ivy's get item by @Ishticode in #28453
- fix: fixed the ivy.not_equal at paddle backend by @samthakur587 in #28431
- feat(docs/api-standards): fix placeholders in api_standards.rst by @shephinphilip in #23404
- Add dropout function for MindSpore Frontend by @savita921 in #21362
- fix: correct tensorflow avg pool2d test failure by @adityadeshmukh369 in #26116
- feat: Add
negative_
method to PyTorch frontend by @daspartho in #23766 - fix: added max_pool2d accessible through the root namespace of the torch frontend and updated the condition to work when stride = [] by @vedpatwardhan in #28457
New Contributors
- @shephinphilip made their first contribution in #23404
- @savita921 made their first contribution in #21362
- @daspartho made their first contribution in #23766
Full Changelog: 0.0.7.4...0.0.7.5
Test Release v0.0.7.4
What's Changed
- fix: support 4d input for torch max_pool3d frontend by @Sam-Armstrong in #28451
- fix: added the missing init files to the tensorflow frontend to get them uploaded to pypi by @vedpatwardhan in #28452
Full Changelog: 0.0.7.3...0.0.7.4
Test Release v0.0.7.3
What's Changed
- fix: remove numpy use from tf backend split by @Sam-Armstrong in #28377
- fix: remove numpy use from tf bincount backend by @Sam-Armstrong in #28378
- fix: fixing the tests for the Shape
__ge__
method by @fnhirwa in #28384 - feat: Updated autoflake version by @Sai-Suraj-27 in #28386
- feat: add tf ResourceVariable to frontend by @Sam-Armstrong in #28393
- fix: fixed the hypothisis requirement error by @samthakur587 in #28412
- Fix Ivy Failing Test: tensorflow - elementwise.pow by @MuhammadNizamani in #28153
- fix: Fixed
ivy.zero_pad
for paddle backend by @Vismay-dev in #27994 - fix: Fix
ivy.permute_dims
for all backends by @Vismay-dev in #28009 - fix: ivy elementwise not_equal for paddle backend by @ksheroz in #28249
- fix: Fix Ivy Failing Test: numpy - manipulation.reshape by @sgalpha01 in #28016
- fix: Fix Ivy Failing Test: all backends - manipulation.squeeze by @sgalpha01 in #28035
- fix: Fix Ivy Failing Test: [jax,numpy,torch] - general.get_item by @sgalpha01 in #28049
- Fix Frontend Failing Test: torch - linalg.torch.linalg.norm by @shruzki in #28106
- fix: Fix Frontend Failing Test: torch - tensor.torch.Tensor.gt (#28219) by @jacksondm33 in #28221
- fix ivy creation.asarray for torch and tensorflow backend by @fleventy-5 in #28226
- fix Ivy Failing Test: paddle - activations.softmax by @fleventy-5 in #28241
- fix Ivy Failing Test: paddle - elementwise.equal by @fleventy-5 in #28243
- Fixed Ivy Frontend Test: numpy - linalg.scipy.linalg.norm by @rp097 in #28244
- fix creation_ops.torch.full for paddle frontend by @fleventy-5 in #28267
- Fixed failing tests for torch - linalg.vector_norm by @BilgehanMehmet in #28273
- Add recall_score function in Ivy with Test by @muzakkirhussain011 in #27986
- feat: Paddle Frontend implemented special method mod by @druvdub in #28245
- fix: fixed cummin not passing test at some backend by @samthakur587 in #28400
- fix: removed code modifying ndarray/torch tensor methods (caused an issue with torch.compile recognising dunder methods). but added the same functionality using
__torch_function__
/__array_ufunc__
which are the proper mechanisms to enable this behaviour. also simplified the test by @mattbarrett98 in #28394 - fix: Fixed Frontend Failing Test: paddle - tensor.torch.Tensor.repeat by @aibenStunner in #28324
- fix: Fixed
tf.math.zero_fraction
for jax backend by @Sai-Suraj-27 in #28356 - feat(jax-backend) - Implementing native jax while loop in
nms
body by @mobley-trent in #28274 - fix: Fixed
ivy.argsort
for jax backend by @Sai-Suraj-27 in #28422 - fix: Variable frontend methods setting ivy_array to frontend tensor by @Sam-Armstrong in #28423
- fix: undocumented args for torch frontend- added
axis
support forstack
, andvalue
support forscatter_
+ extended tests by @mattbarrett98 in #28252 - fix: fixed the maximum at tensorflow frontend by @samthakur587 in #28375
- fix: added a check for when the dtype argument is None in check_unsupported_dtype by @vedpatwardhan in #28434
- fix: fixed the
less_equal
to support the complex dtype by @samthakur587 in #28429 - Update meta.py<_train_tasks_batched by @MahadShahid8 in #22628
- fix: dealing with boolean arrays with tuple query in ivy.get_item by @vedpatwardhan in #28435
- fix: removed the jax.config import from the ivy_tests init as it's deprecated in the recent release by @vedpatwardhan in #28444
- feat: added the test_trace_each flag to the tests to also have tracer tests that trace each example separately by @vedpatwardhan in #28436
- fix: fixed the IvyValueError not import error by @samthakur587 in #28441
- fix: fixed logical_and not passing tests at paddle backend by @samthakur587 in #28433
- Update compiler.py ccb0a3d by @vedpatwardhan in #28450
New Contributors
- @rp097 made their first contribution in #28244
- @BilgehanMehmet made their first contribution in #28273
- @druvdub made their first contribution in #28245
Full Changelog: 0.0.7.2...0.0.7.3
Test Release v0.0.7.2
What's Changed
- fix: fixed argmax for ivy functional api by @samthakur587 in #28329
- fix: Ivy Failing Test: paddle - manipulation.expand_dims by @MuhammadNizamani in #28309
- fix: Fixed
tf.math.argmax
for all backends by @Sai-Suraj-27 in #28347 - fix: fixed the complex dtype support for ivy.maximum at paddle backend by @samthakur587 in #28353
- fix: ensure all lstm weights are contiguous, so they will work on gpu by @Sam-Armstrong in #28358
- fix: fixing tests for
Shape
__mul__
and__rmul__
methods by @fnhirwa in #28337 - fix: improved fix for making lstm weights contiguous by @Sam-Armstrong in #28368
- fix: fixed lint error in adaptive_max_pool3d in layer by @samthakur587 in #28367
- fix: fixed the docstring test for ivy.hardsilu by @samthakur587 in #28370
- fix: minimum for complex dtype by @samthakur587 in #28364
- fix: Ivy Failing Test: paddle - manipulation.permute_dims by @MuhammadNizamani in #28310
- fix: Frontend Failing Test: torch - linalg.torch.linalg.matrix_norm by @Kacper-W-Kozdon in #28323
- fix: fixed the minimum at tf frontend by @samthakur587 in #28373
- Update compiler.py eadf392 by @vaatsalya123 in #28376
Full Changelog: 0.0.7.1...0.0.7.2
Test Release v0.0.7.1
What's Changed
- fix: Fixed
tensorflow.math.is_strictly_increasing
for all backends by @Sai-Suraj-27 in #28336 - fix: Fixed
paddle.take_along_axis
for all backends by @Sai-Suraj-27 in #28321 - fix compositonal get_item removing to_numpy and scalar calls while keeping it general enough by @Ishticode in #28077
- fix: Fixed
tf.math.argmin
for torch, jax, and paddle backends by @Sai-Suraj-27 in #28342 - fix: fixed the relu6 for complex dtype by @samthakur587 in #28326
- fix: Ivy Failing Test: paddle - activations.relu for all backend by @fleventy-5 in #28233
- Update compiler.py 04244f0 by @vedpatwardhan in #28351
Full Changelog: 0.0.7.0...0.0.7.1
Test Release v0.0.7.0
What's Changed
- fix: Fixed
ivy.fix
for paddle backend by @Sai-Suraj-27 in #27988 - fix: fixed ivy.cumsum tests at all backend by @samthakur587 in #27974
- fix: fixed ivy.minimum tests not passing at tensorflow paddle backend by @samthakur587 in #27963
- fix: fixed floor_divide tests not passing at paddle backend by @samthakur587 in #27960
- fix: fixed tests not passing at paddle backend by @samthakur587 in #27958
- fix: fixed unsupported complex dtype at jax and torch backend by @samthakur587 in #27948
- fix: edited the x1 multiple check at complex dtype by @samthakur587 in #27939
- feat: added unflatten frontend and backend support by @Kacper-W-Kozdon in #27416
- FIX! Fix Ivy Failing Test: paddle - elementwise.not_equal by @MuhammadNizamani in #28000
- feat: added an example for using the profiler in utils by @vedpatwardhan in #28004
- fix: Fixed Frontend Failing Test: paddle - tensor.paddle.Tensor.any by @Sai-Suraj-27 in #27992
- fix: Updated the OpenSSL download link in Using venv section of Setting Up by @sgalpha01 in #27954
- refactor: Removed
Optional typehint
when the default value of an argument is non-None value by @Sai-Suraj-27 in #28034 - fix: Fixed inconsistent
naming
of some tests by @Sai-Suraj-27 in #28028 - fix: Fixed creation
ivy.zeros_like
failing test for all backends by @Vismay-dev in #28032 - fix: Fixed
ivy.remainder()
function call which contains wrong argument by @Sai-Suraj-27 in #28045 - feat: Updated
black
formatter version and modified all the files according to it by @Sai-Suraj-27 in #28040 - fix: fixed sigmoid test function by adding atol and rtol to support f… by @samthakur587 in #28093
- fix: fix the tests on bool dtype by @samthakur587 in #28092
- fix: Fixed paddle backend for
comparison_ops.torch.equal
andraw_ops.tensorflow.raw_ops.Equal
by @Sai-Suraj-27 in #28096 - fix: Fixed
attribute.paddle.real
for all the backends by @Sai-Suraj-27 in #28103 - fix: Fixed
math.paddle.pow
for paddle backend by @Sai-Suraj-27 in #28091 - fix: Fixed
creation.paddle.eye
for all backends by @Sai-Suraj-27 in #28068 - fix: fixed cumprod not support bool dtype by @samthakur587 in #28059
- fix: Fix Ivy Failing Test: torch - manipulation.reshape by @sgalpha01 in #28018
- Fix! Fix Ivy Failing Test: paddle jax tensoflow- general.inplace_update by @MuhammadNizamani in #28022
- fix: implement torch frontend
eq
, and remove it as alias for `equal… by @mattbarrett98 in #28121 - fix: Fixed
paddle_backend.divide
by @Sai-Suraj-27 in #28063 - fix: fixed fmod at paddle backend by @samthakur587 in #28124
- fix: fixed rad2deg for all the backend by @samthakur587 in #28131
- feat: added a workflow to be able to trigger all or priority failing tests in the CI, given that tests don't get triggered intelligently at times by @vedpatwardhan in #28134
- fix: added the tensorflow backend implementation for bernoulli and fixed the backend implementations for the other backends regarding default shape and default dtype by @vedpatwardhan in #28139
- feat: superset ivy.lstm so it can be used equivalently to tf.keras.layers.lstm by @Sam-Armstrong in #28140
- fix: Fixed
stat.paddle.mean
for all backends by @Sai-Suraj-27 in #28087 - fix: Fixed the lint warnings, failures and Updated black version by @Sai-Suraj-27 in #28141
- fix: fixed
paddle_backend.greater
andpaddle_backend.greater_equal
by @Sai-Suraj-27 in #28111 - fix: fixed
paddle_backend.less
andpaddle_backend.less_equal
by @Sai-Suraj-27 in #28109 - fix: fixed the failing test case at paddle backend by @samthakur587 in #27831
- Cov by @samthakur587 in #22811
- Fix: Fix Ivy Failing Test: paddle - manipulation.concat by @MuhammadNizamani in #28003
- FIX! Fix Ivy Failing Test: torch - searching.argmax by @MuhammadNizamani in #28117
- feat: change ivy.lstm to always return sequences and states by @Sam-Armstrong in #28155
- fix: fixed fmax at paddle frontend dtype problem by @samthakur587 in #28149
- fix: fixed fmin at paddle frontend facing unsupported dtype error by @samthakur587 in #28148
- feat: added the erfinv function to ivy's experimental API by @vedpatwardhan in #28159
- feat: added Ivy.unflatten by @Kacper-W-Kozdon in #28079
- fix: Fix Ivy Failing Test: torch - elementwise.deg2rad by @MuhammadNizamani in #28171
- fix: Fix Ivy Failing Test: numpy - manipulation.swapaxes by @sgalpha01 in #28042
- fix: Fix Ivy Failing Test: paddle - elementwise.divide by @MuhammadNizamani in #28074
- fix: Fixed
paddle_backend.divide
by adding an if condition to separate dtypes by @Sai-Suraj-27 in #28191 - FIX! Fix Ivy Failing Test: paddle - elementwise.allclose by @MuhammadNizamani in #28166
- fix: Fixed failing docstring tests by @Sai-Suraj-27 in #28172
- fix: fixed the inefficiency when using ivy.conv_general_dilated in a depthwise manner, replaced the loop by using tf.nn.depthwise_conv2d by @vedpatwardhan in #28208
- fix: Fixing pytest discovery issues by @Vismay-dev in #28158
- feat: Add tensorflow GatherNd raw_ops by @TalhaKhalil in #27745
- Revert "feat: Add tensorflow GatherNd raw_ops" by @NripeshN in #28210
- fix: broadcasting issue in compositional set_item by @mattbarrett98 in #28181
- fix: Fix Ivy Failing Test: paddle - elementwise.multiply by @sgalpha01 in #28056
- fix: allow torch backend split to infer final split size by @Sam-Armstrong in #28230
- Fix the reduce method for Ivy Shape by @Dsantra92 in #28212
- fix:
Shape.__bool__
method tests by @fnhirwa in #28218 - Revert "fix:
Shape.__bool__
method tests" by @fnhirwa in #28231 - fix: use of ivy.Shape in conditions by @mattbarrett98 in #28224
- Fix Frontend Failing Test: torch - miscellaneous_ops.torch.clone by @fleventy-5 in #28202
- add reduce method to IvyGlobalProps by @Dsantra92 in #28234
- fix: Ivy Failing Test: paddle - creation.ones_like by @MuhammadNizamani in #28188
- feat: Updated
jax
version mapping from0.4.23
to0.4.24
by @Sai-Suraj-27 in #28237 - feat: Updated
torch
version mapping form2.1.2
to2.2
by @Sai-Suraj-27 in #28236 - fix: changed docstring by @Adlai-1 in #28201
- feat: Updated
ruff
configuration and it's pre-commit version according to the latest release by @Sai-Suraj-27 in #28247 - fix: Fixed triggering of workflows due to
labeling
of the PR's by @Sai-Suraj-27 in #28250 - fix: fixing tests for
Shape
__add__
and__radd__
methods by @fnhirwa in #28263 - feat: Add tensorflow Gather nd raw ops by @TalhaKhalil in #28211
- fix: Fixed
paddle_backend.i0
by @Sai-Suraj-27 in #28184 - fix: Fixed
paddle_backend.copysign
by @Sai-Suraj-27 in #28178 - fix: Fixed
paddle_backend.signbit
by @Sai-Suraj-27 in #28180 - feat: Updated
black
formatter to latest version and modified files accordingly by @Sai-Suraj-27 in #28282 - feat: Added a decorator function, which will be used to test
explicit examples
by @Sai-Suraj-27 in #28251 - fix: Frontend Failing Test: torch - tensor.torch.Tensor.unique by @ogbanugot in #28283
- fix: array_indices_axis helper updated to use args to control vals from dtypes_and_values helper by @Kacper-W-Kozdon in #28168
- fix: handling of None in torch frontend pad + extended test by @m...
Test Release v0.0.6.2
Full Changelog: 0.0.6.1...0.0.6.2
Test Release v0.0.6.1
fix: added wrappers.json to the MANIFEST.in