Commits

Branch selector

main

User selector

hmahmood24

All time

Commit history

Commits on Mar 6, 2024

• fix(get-item): Use dynamic shape when available rather than static x.shape e.g. when the backend is set as tensorflow in _parse_query, allows get_item to be compilable with tf.function

  hmahmood24

  committed
  3e7389dView commit details

  Copy full SHA for 3e7389d

  Browse repository at this point

• feat(backends): Added a primary implementation for flatten in tensorflow backend (#28488)

  hmahmood24

  authored
  514fedcView commit details

  Copy full SHA for 514fedc

  Browse repository at this point

• fix(stateful): Added a bypass to avoid calculating input shapes whenever we don't want to use the keras build method for the backend keras.Model class

  hmahmood24

  committed
  9c52cdbView commit details

  Copy full SHA for 9c52cdb

  Browse repository at this point

Commits on Feb 29, 2024

• feat(ivy): Add size to Ivy Functional API

  hmahmood24

  committed
  cd7cb35View commit details

  Copy full SHA for cd7cb35

  Browse repository at this point

Commits on Feb 21, 2024

• feat(tensorflow-keras): Added depthwise_conv2d to tf.keras.backend frontend namespace along with the test. Also added tf.nn.swish as an alias for tf.nn.silu

  hmahmood24

  committed
  2eb432dView commit details

  Copy full SHA for 2eb432d

  Browse repository at this point

Commits on Feb 15, 2024

• fix(keras.Model): Rewrote the _get_input_shapes method to correctly handle non-tensor like arguments and also avoid using tf.nest.flatten to avoid unpacking tuples in the arguments

  hmahmood24

  committed
  d3258a9View commit details

  Copy full SHA for d3258a9

  Browse repository at this point

Commits on Feb 7, 2024

• fix(nn.Module): Temporarily add a try-except block in model.call to deal with multiple arguments when run as a keras model

  hmahmood24

  committed
  54e3f62View commit details

  Copy full SHA for 54e3f62

  Browse repository at this point

• fix(tf.keras.Model): Added a _get_input_shapes helper func to retrieve the input shapes during model building. Also added a number of tf.autograph.experimental.do_not_convert decorators to avoid tr…

  Show description for 74238fc
  hmahmood24

  committed
  74238fcView commit details

  Copy full SHA for 74238fc

  Browse repository at this point

• fix(nn.Module): Replaced references to .built with ._built to avoid name clashes with the backend tf.keras.Model stateful subclassed module. Also modified the call method to deal with multiple inpu…

  Show description for b206d3e
  hmahmood24

  committed
  b206d3eView commit details

  Copy full SHA for b206d3e

  Browse repository at this point

• fix(dev): Added a check to fix retrieving the correct device when running with tensorflow backend and in tf's graph mode

  hmahmood24

  committed
  12647e9View commit details

  Copy full SHA for 12647e9

  Browse repository at this point

Commits on Feb 3, 2024

• feat(frontends): Added erfinv and erfinv_ to torch frontend along with tests

  hmahmood24

  committed
  ba08c56View commit details

  Copy full SHA for ba08c56

  Browse repository at this point

Commits on Feb 1, 2024

• fix(stateful): Fixed to_keras_module so that after converting a frontend nn.Module instance to a keras instance, the model parameters in the .v attribute remain as frontend torch.nn.Parameter insta…

  Show description for a66367b
  hmahmood24

  committed
  a66367bView commit details

  Copy full SHA for a66367b

  Browse repository at this point

• fix(stateful): Fixes torch.nn.Module __setattr__ logic to remove existing attributes during attribute setting for correct retrieval of overwritten attributes. Added module building check to named_m…

  Show description for 85ff35e
  hmahmood24

  committed
  85ff35eView commit details

  Copy full SHA for 85ff35e

  Browse repository at this point

Commits on Jan 24, 2024

• fix(data_type): Add missing functools.wraps to the _handle_nestable_dtype_info_wrapper decorator

  hmahmood24

  committed
  3e2da70View commit details

  Copy full SHA for 3e2da70

  Browse repository at this point

• fix(torch-frontend): Handle default case for torch.Size when no input shape is provided. Also add numel to the Size class

  hmahmood24

  committed
  e1724a8View commit details

  Copy full SHA for e1724a8

  Browse repository at this point

• feat(torch-frontend): Add finfo to torch frontend

  hmahmood24

  committed
  4a2ef0bView commit details

  Copy full SHA for 4a2ef0b

  Browse repository at this point

Commits on Jan 22, 2024

• feat(stateful): Add `NativeModule` to all backends with the implementation of a custom `tensorflow.keras.Model` subclass for the `tensorflow` backend to follow `ivy.Module`'s API (#27955)

  hmahmood24

  authored
  90bedc1View commit details

  Copy full SHA for 90bedc1

  Browse repository at this point

Commits on Jan 19, 2024

• chore(temp): Remove dynamic_backend arg for ivy.Container from frontend nn.Module

  hmahmood24

  committed
  bd1bbd9View commit details

  Copy full SHA for bd1bbd9

  Browse repository at this point

Commits on Jan 18, 2024

• fix(tensorflow-backend): Fix asarray in tensorflow backend for cases when dtype is not provided and the passed in object is also not a tf.Tensor

  hmahmood24

  committed
  2f8548dView commit details

  Copy full SHA for 2f8548d

  Browse repository at this point

Commits on Jan 4, 2024

• chore: Modify the fill_ and zero_ frontend torch methods to use ivy.inplace_update explicitly

  hmahmood24

  committed
  e12cd47View commit details

  Copy full SHA for e12cd47

  Browse repository at this point

Commits on Dec 4, 2023

• fix: Add a check in the converters to filter any NoneType model parameters when doing a to_keras_module on an ivy.Module instance

  hmahmood24

  committed
  541a68cView commit details

  Copy full SHA for 541a68c

  Browse repository at this point

Commits on Dec 2, 2023

• Add __getstate__ and __setstate__ methods to frontend torch.nn.Module

  hmahmood24

  committed
  0984daeView commit details

  Copy full SHA for 0984dae

  Browse repository at this point

Commits on Nov 28, 2023

• refactor(stateful): Refactor `torch.nn.Module` and `ivy.Module` classes (#27403)

  hmahmood24

  authored
  feb676aView commit details

  Copy full SHA for feb676a

  Browse repository at this point

Commits on Nov 17, 2023

• chore: Add test for torch.Tensor.uniform_ and remove unnecessary comments

  hmahmood24

  committed
  79d36a1View commit details

  Copy full SHA for 79d36a1

  Browse repository at this point

• feat: Add training support to `ivy.Module` converted `tf.keras.Model` instances (#27297)

  hmahmood24

  authored
  d9e8e49View commit details

  Copy full SHA for d9e8e49

  Browse repository at this point

Commits on Nov 16, 2023

• refactor: Refactor the frontend Parameter class to use gradient helper functions to correctly handle NativeVariable instances underneath

  hmahmood24

  committed
  8baac2dView commit details

  Copy full SHA for 8baac2d

  Browse repository at this point

• fix: Fix tensorflow backend conv functions with incorrect types being casted together

  hmahmood24

  committed
  35a5a4eView commit details

  Copy full SHA for 35a5a4e

  Browse repository at this point

Commits on Nov 7, 2023

• feat: Add checks to handle `tf.TensorArrays` used extensively by `tf.keras.LSTM` (#27222)

  hmahmood24

  authored
  d0fe54dView commit details

  Copy full SHA for d0fe54d

  Browse repository at this point

Commits on Nov 5, 2023

• fix: Explicitly set requires_grad=False on torch Modules until backprop through the frontends has been fixed (#27212)

  hmahmood24

  authored
  82986f4View commit details

  Copy full SHA for 82986f4

  Browse repository at this point

• Revert back recent changes made to avg_pool functions (#27211)

  hmahmood24

  authored
  786130aView commit details

  Copy full SHA for 786130a

  Browse repository at this point

Commits on Nov 4, 2023

• Add `torch.nn.Parameter`, `torch.nn.Module` and `ivy.Module.to_keras_module` for KLA (#27210)

  hmahmood24

  authored
  48044a4View commit details

  Copy full SHA for 48044a4

  Browse repository at this point

Commits on Oct 23, 2023

• fix: Fix transpiling torchvision to torch

  hmahmood24

  committed
  f7753bbView commit details

  Copy full SHA for f7753bb

  Browse repository at this point

• fix: Fix broken module converter tests (#27089)

  hmahmood24

  authored
  0665a76View commit details

  Copy full SHA for 0665a76

  Browse repository at this point

Commits on Oct 20, 2023

• fix: avoid casting shape to list to make frontend torch.lstm traceable

  hmahmood24

  committed
  0478a74View commit details

  Copy full SHA for 0478a74

  Browse repository at this point

Commits on Oct 10, 2023

• fix: Update module converters to correctly handle tuple outputs in their `_forward` methods (#26737)

  hmahmood24

  authored
  814389eView commit details

  Copy full SHA for 814389e

  Browse repository at this point