Merge branch 'unifyai:main' into main

.github/pull_request_template.md

@@ -22,8 +22,21 @@ Close #
 
 - [ ] Did you add a function?
 - [ ] Did you add the tests?
+- [ ] Did you run your tests and are your tests passing?
+- [ ] Did pre-commit not fail on any check?
 - [ ] Did you follow the steps we provided?
 
+<!--
+Please mark your PR as a draft if you realise after the fact that your tests are not passing or
+that your pre-commit check has some failures.
+
+Here are some relevant resources regarding tests and pre-commit:
+
+https://unify.ai/docs/ivy/overview/deep_dive/ivy_tests.html
+https://unify.ai/docs/ivy/overview/deep_dive/formatting.html#pre-commit
+
+-->
+
 ### Socials:
 
 <!--

.github/workflows/dockerfile-multiversion-push.yml

@@ -0,0 +1,26 @@
+name: Dockerfile Mutiversion Push
+
+on:
+  schedule:
+    - cron: '0 0 * * *'
+  workflow_dispatch:
+
+jobs:
+
+  build:
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Checkout Ivy 🛎
+        uses: actions/checkout@v3
+
+      - name: Login to Docker Hub
+        uses: docker/login-action@v2
+        with:
+          username: ${{ secrets.DOCKERHUB_USERNAME }}
+          password: ${{ secrets.DOCKERHUB_TOKEN }}
+
+      - name: Build and push Dockerfile
+        run: |
+          docker build --progress=plain --no-cache -t unifyai/multiversion:latest -f docker/DockerfileMultiversion .
+          docker push unifyai/multiversion:latest

.github/workflows/welcome-message.yml

@@ -0,0 +1,56 @@
+name: Check Semantic and welcome new contributors
+
+on:
+  pull_request_target:
+    types:
+      - opened
+      - edited
+      - synchronize
+      - reopened
+  workflow_call:
+
+permissions:
+  pull-requests: write
+
+jobs:
+  pr-compliance-checks:
+    name: PR Compliance Checks
+    runs-on: ubuntu-latest
+    timeout-minutes: 10
+    steps:
+      - uses: mtfoley/[email protected]
+        with:
+          body-auto-close: false
+          protected-branch-auto-close: false
+          body-comment: >
+            ## Issue Reference
+
+            In order to be considered for merging, the pull request description must refer to a
+            specific issue number. This is described in our
+            [contributing guide](https://unify.ai/docs/ivy/overview/contributing/the_basics.html#todo-list-issues) and our PR template.
+
+            This check is looking for a phrase similar to: "Fixes #XYZ" or "Resolves #XYZ" where XYZ is the issue
+            number that this PR is meant to address.
+
+  welcome:
+    name: Welcome
+    runs-on: ubuntu-latest
+    timeout-minutes: 10
+    needs: semantics
+    if: github.event.action == 'opened'
+    steps:
+      - uses: actions/first-interaction@v1
+        with:
+          repo-token: ${{ secrets.GITHUB_TOKEN }}
+          pr-message: |-
+            Congrats on making your first Pull Request and thanks for supporting Ivy! 🎉
+            Join the conversation in our [Discord](https://discord.com/invite/sXyFF8tDtm)
+
+            Here are some notes to understand our tests:
+            - We have merged all the tests in one file called \`display_test_results\` job. 👀 It contains the following two sections:
+              - **Combined Test Results:** This shows the results of all the ivy tests that ran on the PR. ✔️
+              - **New Failures Introduced:** This lists the tests that fails on this PR.
+           
+            Please make sure they are passing. 💪
+
+            Keep in mind that we will assign an engineer for this task and they will look at it based on the workload that they have, so be patient.

docker/DockerfileMultiversion

@@ -2,7 +2,7 @@ FROM debian:buster
 WORKDIR /ivy
 
 ARG fw
-ARG pycon=3.8.10
+ARG pycon=3.10
 # Install miniconda
 ENV CONDA_DIR /opt/miniconda/
 

docker/requirement_mappings_multiversion.json

@@ -1,11 +1,34 @@
-
 {
-    "tensorflow": [
-        {"tensorflow-probability":{"2.12.0":"0.20.0","2.11.0":"0.19.0"}}
-    ],
-    "jax": ["dm-haiku", "flax",{"jaxlib": {"0.4.10": "0.4.10","0.4.8": "0.4.7"}}],
-    "numpy": ["numpy"],
-    "paddle": ["paddlepaddle"],
-    "mxnet": ["mxnet"],
-    "torch": ["torch-scatter"]
+  "tensorflow": [
+    {
+      "tensorflow-probability": {
+        "2.13.0": "0.21.0",
+        "2.12.0": "0.20.0",
+        "2.11.0": "0.19.0"
+      }
+    }
+  ],
+  "jax": [
+    "dm-haiku",
+    "flax",
+    {
+      "jaxlib": {
+        "0.4.14": "0.4.14",
+        "0.4.10": "0.4.10",
+        "0.4.8": "0.4.7"
+      }
+    }
+  ],
+  "numpy": [
+    "numpy"
+  ],
+  "paddle": [
+    "paddlepaddle"
+  ],
+  "mxnet": [
+    "mxnet"
+  ],
+  "torch": [
+    "torch-scatter"
+  ]
 }

ivy/data_classes/array/activations.py

@@ -131,7 +131,13 @@ def gelu(
             self._data, approximate=approximate, complex_mode=complex_mode, out=out
         )
 
-    def sigmoid(self: ivy.Array, /, *, out: Optional[ivy.Array] = None) -> ivy.Array:
+    def sigmoid(
+        self: ivy.Array,
+        /,
+        *,
+        complex_mode: Literal["split", "magnitude", "jax"] = "jax",
+        out: Optional[ivy.Array] = None,
+    ) -> ivy.Array:
         """
         ivy.Array instance method variant of ivy.sigmoid.
 
@@ -142,6 +148,9 @@ def sigmoid(self: ivy.Array, /, *, out: Optional[ivy.Array] = None) -> ivy.Array
         ----------
         self
             Input array
+        complex_mode
+            optional specifier for how to handle complex data types. See
+            ``ivy.func_wrapper.handle_complex_input`` for more detail.
         out
             optional output array for writing the result to. It must have the same shape
             the input broadcast to default: None
@@ -159,7 +168,7 @@ def sigmoid(self: ivy.Array, /, *, out: Optional[ivy.Array] = None) -> ivy.Array
         >>> print(y)
         ivy.array([0.269, 0.731, 0.881])
         """
-        return ivy.sigmoid(self._data, out=out)
+        return ivy.sigmoid(self._data, complex_mode=complex_mode, out=out)
 
     def softmax(
         self: ivy.Array,

ivy/data_classes/container/activations.py

@@ -419,6 +419,7 @@ def _static_sigmoid(
         to_apply: Union[bool, ivy.Container] = True,
         prune_unapplied: Union[bool, ivy.Container] = False,
         map_sequences: Union[bool, ivy.Container] = False,
+        complex_mode: Literal["split", "magnitude", "jax"] = "jax",
         out: Optional[ivy.Container] = None,
     ) -> ivy.Container:
         """
@@ -441,6 +442,9 @@ def _static_sigmoid(
         map_sequences
             Whether to also map method to sequences (lists, tuples).
             Default is ``False``.
+        complex_mode
+            optional specifier for how to handle complex data types. See
+            ``ivy.func_wrapper.handle_complex_input`` for more detail.
         out
             optional output container, for writing the result to. It must have a shape
             that the inputs broadcast to.
@@ -467,6 +471,7 @@ def _static_sigmoid(
             to_apply=to_apply,
             prune_unapplied=prune_unapplied,
             map_sequences=map_sequences,
+            complex_mode=complex_mode,
             out=out,
         )
 
@@ -478,6 +483,7 @@ def sigmoid(
         to_apply: Union[bool, ivy.Container] = True,
         prune_unapplied: Union[bool, ivy.Container] = False,
         map_sequences: Union[bool, ivy.Container] = False,
+        complex_mode: Literal["split", "magnitude", "jax"] = "jax",
         out: Optional[ivy.Container] = None,
     ) -> ivy.Container:
         """
@@ -500,6 +506,9 @@ def sigmoid(
         map_sequences
             Whether to also map method to sequences (lists, tuples).
             Default is ``False``.
+        complex_mode
+            optional specifier for how to handle complex data types. See
+            ``ivy.func_wrapper.handle_complex_input`` for more detail.
         out
             optional output container, for writing the result to. It must have a shape
             that the inputs broadcast to.
@@ -525,6 +534,7 @@ def sigmoid(
             to_apply=to_apply,
             prune_unapplied=prune_unapplied,
             map_sequences=map_sequences,
+            complex_mode=complex_mode,
             out=out,
         )
 

ivy/functional/backends/jax/activations.py

@@ -40,7 +40,9 @@ def relu(
     return jnp.maximum(x, 0)
 
 
-def sigmoid(x: JaxArray, /, *, out: Optional[JaxArray] = None) -> JaxArray:
+def sigmoid(
+    x: JaxArray, /, *, complex_mode="jax", out: Optional[JaxArray] = None
+) -> JaxArray:
     return 1 / (1 + jnp.exp(-x))
 
 

ivy/functional/backends/jax/experimental/layers.py

@@ -209,6 +209,7 @@ def max_pool2d(
     out: Optional[JaxArray] = None,
 ) -> JaxArray:
     dims = 2
+    odtype = x.dtype
     kernel, strides, padding, dilation = _validate_max_pool_params(
         kernel, strides, padding, dilation, ceil_mode, dims=dims
     )
@@ -234,9 +235,9 @@ def max_pool2d(
     )
 
     if data_format == "NCHW":
-        return jnp.transpose(res, (0, 3, 1, 2))
+        res = jnp.transpose(res, (0, 3, 1, 2))
 
-    return res
+    return res.astype(odtype)
 
 
 def max_pool3d(

ivy/functional/backends/numpy/activations.py

@@ -46,7 +46,9 @@ def gelu(
     return ivy.astype(ret, x.dtype, copy=False)
 
 
-def sigmoid(x: np.ndarray, /, *, out: Optional[np.ndarray] = None) -> np.ndarray:
+def sigmoid(
+    x: np.ndarray, /, *, complex_mode="jax", out: Optional[np.ndarray] = None
+) -> np.ndarray:
     if not ivy.is_array(x):
         return np.asarray(1 / (1 + np.exp(-x)))
     return np.asarray(1 / (1 + np.exp(-x))).astype(x.dtype)

ivy/functional/backends/paddle/activations.py

@@ -86,12 +86,15 @@ def gelu(
     return F.gelu(x, approximate=approximate)
 
 
+@with_unsupported_device_and_dtypes(
+    {"2.5.1 and below": {"cpu": ("bfloat16",)}}, backend_version
+)
 def sigmoid(
-    x: paddle.Tensor, /, *, out: Optional[paddle.Tensor] = None
+    x: paddle.Tensor, /, *, complex_mode="jax", out: Optional[paddle.Tensor] = None
 ) -> paddle.Tensor:
+    if paddle.is_complex(x):
+        return 1.0 / (1.0 + paddle_backend.exp(-x))
     if x.dtype in unsupported_dtypes:
-        if paddle.is_complex(x):
-            return 1 / (1 + paddle_backend.exp(-x))
         return F.sigmoid(x.cast("float32")).cast(x.dtype)
     return F.sigmoid(x)
 

ivy/functional/backends/tensorflow/activations.py

@@ -46,7 +46,9 @@ def relu(x: Tensor, /, *, complex_mode="jax", out: Optional[Tensor] = None) -> T
     return tf.nn.relu(x)
 
 
-def sigmoid(x: Tensor, /, *, out: Optional[Tensor] = None) -> Tensor:
+def sigmoid(
+    x: Tensor, /, *, complex_mode="jax", out: Optional[Tensor] = None
+) -> Tensor:
     if not ivy.is_array(x):
         x = float(x)
     return tf.nn.sigmoid(x)

ivy/functional/backends/torch/activations.py

@@ -54,7 +54,9 @@ def gelu(
 
 
 @with_unsupported_dtypes({"2.0.1 and below": ("float16",)}, backend_version)
-def sigmoid(x: torch.Tensor, /, *, out: Optional[torch.Tensor] = None) -> torch.Tensor:
+def sigmoid(
+    x: torch.Tensor, /, *, complex_mode="jax", out: Optional[torch.Tensor] = None
+) -> torch.Tensor:
     if not ivy.is_array(x):
         x = torch.tensor(x)
     return torch.sigmoid(x, out=out)

ivy/functional/frontends/jax/nn/non_linear_activations.py

@@ -286,7 +286,7 @@ def selu(x):
 @to_ivy_arrays_and_back
 def sigmoid(x):
     x = _type_conversion(x)
-    ret = ivy.sigmoid(x)
+    ret = ivy.sigmoid(x, complex_mode="jax")
     return ivy.astype(ret, x.dtype)
 
 

ivy/functional/frontends/paddle/tensor/math.py

@@ -48,3 +48,9 @@ def rsqrt_(x, name=None):
 @to_ivy_arrays_and_back
 def sqrt_(x, name=None):
     return ivy.inplace_update(x, sqrt(x))
+
+
+@with_supported_dtypes({"2.5.1 and below": ("float32", "float64")}, "paddle")
+@to_ivy_arrays_and_back
+def subtract_(x, y, name=None):
+    return ivy.inplace_update(x, subtract(x, y))