[API] Standardize MXNet NumPy creation functions (#20572)

* standardize mxnet numpy creation functions

* update

* fix linspace

* merge & add tests

* add NumpyLinspaceParam

* fix lint'

'

* fix

* add indexing test

* fix tests

* fix sanity

* fix lint

* fix tests

* disable warning

* fix

* update

* skip signature standardization

* fix lint

* update

* rm test_contants

* Add Code Signing Key

* Revert "Add Code Signing Key"

This reverts commit b09814c.

* Replace context with device & update multiarray.py/_op.py

* ctx => device

* ctx/context => device

* fix

* fix multiarray

* update ndarray.py

* fix

* fix

* fix tests

* update

* fix tests

* update rand_zipfian

* update

* device => cuda_device in util.py

* context.gpu_memory_info => device.gpu_memory_info

* fix docs

* rm context in doc

* fix lint

* remove npv

* Revert "remove npv"

This reverts commit e775844.

Author: barry-jin

benchmark/python/sparse/cast_storage.py

@@ -41,7 +41,7 @@ def measure_cost(repeat, f, *args, **kwargs):
 
 def run_cast_storage_synthetic():
     def dense_to_sparse(m, n, density, ctx, repeat, stype):
-        set_default_context(ctx)
+        set_default_device(ctx)
         data_shape = (m, n)
         dns_data = rand_ndarray(data_shape, stype, density).tostype('default')
         dns_data.wait_to_read()

benchmark/python/sparse/dot.py

@@ -26,7 +26,7 @@
 import mxnet as mx
 import numpy as np
 import numpy.random as rnd
-from mxnet.test_utils import rand_ndarray, set_default_context, assert_almost_equal, get_bz2_data
+from mxnet.test_utils import rand_ndarray, set_default_device, assert_almost_equal, get_bz2_data
 from mxnet.base import check_call, _LIB
 from util import estimate_density
 
@@ -267,7 +267,7 @@ def test_dot_synthetic(data_dict):
     # Benchmark MXNet and Scipys dot operator
     def bench_dot(lhs_shape, rhs_shape, lhs_stype, rhs_stype,
                   lhs_den, rhs_den, trans_lhs, ctx, num_repeat=10, fw="mxnet", distribution="uniform"):
-        set_default_context(ctx)
+        set_default_device(ctx)
         assert fw == "mxnet" or fw == "scipy"
         # Set funcs
         dot_func_sparse = mx.nd.sparse.dot if fw == "mxnet" else sp.spmatrix.dot

benchmark/python/sparse/sparse_op.py

@@ -155,7 +155,7 @@ def measure_cost_backward_baseline(repeat, dot, transpose, lhs, rhs):
         return diff / repeat
 
     def bench_dot_forward(m, k, n, density, ctx, repeat):
-        set_default_context(ctx)
+        set_default_device(ctx)
         dns = mx.nd.random.uniform(shape=(k, n)).copyto(ctx)
         data_shape = (m, k)
         csr_data = rand_ndarray(data_shape, 'csr', density)
@@ -184,7 +184,7 @@ def bench_dot_forward(m, k, n, density, ctx, repeat):
                      ratio_baseline, costs_baseline[0], costs_baseline[1]))
 
     def bench_dot_backward(m, k, n, density, ctx, repeat):
-        set_default_context(ctx)
+        set_default_device(ctx)
         dns = mx.nd.random.uniform(shape=(m, n)).copyto(ctx)
         data_shape = (m, k)
         csr_data = rand_ndarray(data_shape, 'csr', density)

ci/docker/runtime_functions.sh

@@ -874,6 +874,8 @@ unittest_array_api_standardization() {
     export DMLC_LOG_STACK_TRACE_DEPTH=100
     python3 -m pytest --durations=50 --cov-report xml:tests_api.xml --verbose \
         array_api_tests/test_type_promotion.py::test_elementwise_function_two_arg_bool_type_promotion
+    python3 -m pytest --durations=50 --cov-report xml:tests_api.xml --verbose array_api_tests/test_creation_functions.py
+    python3 -m pytest --durations=50 --cov-report xml:tests_api.xml --verbose array_api_tests/test_indexing.py
     popd
 }
 

…python_docs/python/api/context/index.rst …/python_docs/python/api/device/index.rstdocs/python_docs/python/api/context/index.rst renamed to docs/python_docs/python/api/device/index.rst docs/python_docs/python/api/context/index.rst renamed to docs/python_docs/python/api/device/index.rst

@@ -15,9 +15,9 @@
    specific language governing permissions and limitations
    under the License.
 
-mxnet.context
+mxnet.device
 =============
 
-.. automodule:: mxnet.context
+.. automodule:: mxnet.device
     :members:
     :autosummary:

docs/python_docs/python/api/index.rst

@@ -86,10 +86,10 @@ Gluon related modules
       Key value store interface of MXNet for parameter synchronization.
 
    .. card::
-      :title: mxnet.context
-      :link: mxnet/context/index.html
+      :title: mxnet.device
+      :link: mxnet/device/index.html
 
-      CPU and GPU context information.
+      CPU and GPU device information.
 
    .. card::
       :title: mxnet.profiler
@@ -116,10 +116,10 @@ Advanced modules
       API for querying MXNet enabled features.
 
    .. card::
-      :title: mxnet.context
-      :link: context/index.html
+      :title: mxnet.device
+      :link: device/index.html
 
-      MXNet array context for specifying in-memory storage device.
+      MXNet array device for specifying in-memory storage device.
 
    .. card::
       :title: mxnet.profiler

docs/python_docs/python/api/npx/index.rst

@@ -50,7 +50,7 @@ Devices
    cpu_pinned
    gpu
    gpu_memory_info
-   current_context
+   current_device
    num_gpus
 
 Nerual networks

docs/python_docs/python/tutorials/deploy/inference/image_classification_jetson.md

@@ -71,18 +71,18 @@ And we are done. You can test the installation now by importing mxnet from pytho
 
 ## Running a pre-trained ResNet-50 model on Jetson
 
-We are now ready to run a pre-trained model and run inference on a Jetson module. In this tutorial we are using ResNet-50 model trained on Imagenet dataset. We run the following classification script with either cpu/gpu context using python3.
+We are now ready to run a pre-trained model and run inference on a Jetson module. In this tutorial we are using ResNet-50 model trained on Imagenet dataset. We run the following classification script with either cpu/gpu device using python3.
 
 ```{.python .input}
 from mxnet import gluon
 import mxnet as mx
 
-# set context
+# set device
 gpus = mx.test_utils.list_gpus()
-ctx =  mx.gpu() if gpus else mx.cpu()
+device =  mx.gpu() if gpus else mx.cpu()
 
 # load pre-trained model
-net = gluon.model_zoo.vision.resnet50_v1(pretrained=True, ctx=ctx)
+net = gluon.model_zoo.vision.resnet50_v1(pretrained=True, device=device)
 net.hybridize(static_alloc=True, static_shape=True)
 
 # load labels
@@ -99,7 +99,7 @@ img = mx.image.color_normalize(img.astype(dtype='float32')/255,
                                std=mx.np.array([0.229, 0.224, 0.225])) # normalize
 img = img.transpose((2, 0, 1)) # channel first
 img = mx.np.expand_dims(img, axis=0) # batchify
-img = img.as_in_ctx(ctx)
+img = img.to_device(device)
 
 prob = mx.npx.softmax(net(img)) # predict and normalize output
 idx = mx.npx.topk(prob, k=5)[0] # get top 5 result

docs/python_docs/python/tutorials/extend/customop.md

@@ -101,7 +101,7 @@ class SigmoidProp(mx.operator.CustomOpProp):
         # return 3 lists representing inputs shapes, outputs shapes, and aux data shapes.
         return (data_shape,), (output_shape,), ()
 
-    def create_operator(self, ctx, in_shapes, in_dtypes):
+    def create_operator(self, device, in_shapes, in_dtypes):
         #  create and return the CustomOp class.
         return Sigmoid()
 ```
@@ -183,7 +183,7 @@ class DenseProp(mx.operator.CustomOpProp):
         # return 3 lists representing inputs shapes, outputs shapes, and aux data shapes.
         return (data_shape, weight_shape), (output_shape,), ()
 
-    def create_operator(self, ctx, in_shapes, in_dtypes):
+    def create_operator(self, device, in_shapes, in_dtypes):
         #  create and return the CustomOp class.
         return Dense(self._bias)
 ```
@@ -201,8 +201,8 @@ class DenseBlock(mx.gluon.Block):
         self.weight = gluon.Parameter('weight', shape=(channels, in_channels))
 
     def forward(self, x):
-        ctx = x.context
-        return mx.nd.Custom(x, self.weight.data(ctx), bias=self._bias, op_type='dense')
+        device = x.device
+        return mx.nd.Custom(x, self.weight.data(device), bias=self._bias, op_type='dense')
 ```
 
 ### Example usage

docs/python_docs/python/tutorials/getting-started/crash-course/1-nparray.md

@@ -195,7 +195,7 @@ a = np.array(a)
 (type(a), a)
 ```
 
-Additionally, you can move them to different GPU contexts. You will dive more
+Additionally, you can move them to different GPU devices. You will dive more
 into this later, but here is an example for now.
 
 ```{.python .input}