mindspore 2.3.0rc1__cp38-cp38-manylinux1_x86_64.whl → 2.3.0rc2__cp38-cp38-manylinux1_x86_64.whl

mindspore/communication/management.py

@@ -138,14 +138,14 @@ def init(backend_name=None):
 
             For the Ascend devices, users need to prepare the rank table, set rank_id and device_id.
             Please see the `rank table Startup
-            <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/rank_table.html>`_
+            <https://www.mindspore.cn/tutorials/experts/en/master/parallel/rank_table.html>`_
             for more details.
 
             For the GPU devices, users need to prepare the host file and mpi, please see the `mpirun Startup
-            <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/mpirun.html>`_ .
+            <https://www.mindspore.cn/tutorials/experts/en/master/parallel/mpirun.html>`_ .
 
             For the CPU device, users need to write a dynamic cluster startup script, please see the `Dynamic Cluster
-            Startup <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/dynamic_cluster.html>`_ .
+            Startup <https://www.mindspore.cn/tutorials/experts/en/master/parallel/dynamic_cluster.html>`_ .
 
         >>> from mindspore.communication import init
         >>> init()
@@ -226,14 +226,14 @@ def release():
 
             For the Ascend devices, users need to prepare the rank table, set rank_id and device_id.
             Please see the `rank table Startup
-            <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/rank_table.html>`_
+            <https://www.mindspore.cn/tutorials/experts/en/master/parallel/rank_table.html>`_
             for more details.
 
             For the GPU devices, users need to prepare the host file and mpi, please see the `mpirun Startup
-            <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/mpirun.html>`_ .
+            <https://www.mindspore.cn/tutorials/experts/en/master/parallel/mpirun.html>`_ .
 
             For the CPU device, users need to write a dynamic cluster startup script, please see the `Dynamic Cluster
-            Startup <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/dynamic_cluster.html>`_ .
+            Startup <https://www.mindspore.cn/tutorials/experts/en/master/parallel/dynamic_cluster.html>`_ .
 
         >>> from mindspore.communication import init, release
         >>> init()
@@ -270,14 +270,14 @@ def get_rank(group=GlobalComm.WORLD_COMM_GROUP):
 
             For the Ascend devices, users need to prepare the rank table, set rank_id and device_id.
             Please see the `rank table Startup
-            <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/rank_table.html>`_
+            <https://www.mindspore.cn/tutorials/experts/en/master/parallel/rank_table.html>`_
             for more details.
 
             For the GPU devices, users need to prepare the host file and mpi, please see the `mpirun Startup
-            <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/mpirun.html>`_ .
+            <https://www.mindspore.cn/tutorials/experts/en/master/parallel/mpirun.html>`_ .
 
             For the CPU device, users need to write a dynamic cluster startup script, please see the `Dynamic Cluster
-            Startup <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/dynamic_cluster.html>`_ .
+            Startup <https://www.mindspore.cn/tutorials/experts/en/master/parallel/dynamic_cluster.html>`_ .
 
         >>> from mindspore.communication import init, get_rank
         >>> init()
@@ -320,14 +320,14 @@ def get_local_rank(group=GlobalComm.WORLD_COMM_GROUP):
 
             For the Ascend devices, users need to prepare the rank table, set rank_id and device_id.
             Please see the `rank table Startup
-            <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/rank_table.html>`_
+            <https://www.mindspore.cn/tutorials/experts/en/master/parallel/rank_table.html>`_
             for more details.
 
             For the GPU devices, users need to prepare the host file and mpi, please see the `mpirun Startup
-            <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/mpirun.html>`_ .
+            <https://www.mindspore.cn/tutorials/experts/en/master/parallel/mpirun.html>`_ .
 
             For the CPU device, users need to write a dynamic cluster startup script, please see the `Dynamic Cluster
-            Startup <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/dynamic_cluster.html>`_ .
+            Startup <https://www.mindspore.cn/tutorials/experts/en/master/parallel/dynamic_cluster.html>`_ .
 
         >>> import mindspore as ms
         >>> from mindspore.communication import init, get_rank, get_local_rank
@@ -373,14 +373,14 @@ def get_group_size(group=GlobalComm.WORLD_COMM_GROUP):
 
             For the Ascend devices, users need to prepare the rank table, set rank_id and device_id.
             Please see the `rank table Startup
-            <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/rank_table.html>`_
+            <https://www.mindspore.cn/tutorials/experts/en/master/parallel/rank_table.html>`_
             for more details.
 
             For the GPU devices, users need to prepare the host file and mpi, please see the `mpirun Startup
-            <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/mpirun.html>`_ .
+            <https://www.mindspore.cn/tutorials/experts/en/master/parallel/mpirun.html>`_ .
 
             For the CPU device, users need to write a dynamic cluster startup script, please see the `Dynamic Cluster
-            Startup <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/dynamic_cluster.html>`_ .
+            Startup <https://www.mindspore.cn/tutorials/experts/en/master/parallel/dynamic_cluster.html>`_ .
 
         >>> import mindspore as ms
         >>> from mindspore.communication import init, get_group_size
@@ -425,14 +425,14 @@ def get_local_rank_size(group=GlobalComm.WORLD_COMM_GROUP):
 
             For the Ascend devices, users need to prepare the rank table, set rank_id and device_id.
             Please see the `rank table Startup
-            <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/rank_table.html>`_
+            <https://www.mindspore.cn/tutorials/experts/en/master/parallel/rank_table.html>`_
             for more details.
 
             For the GPU devices, users need to prepare the host file and mpi, please see the `mpirun Startup
-            <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/mpirun.html>`_ .
+            <https://www.mindspore.cn/tutorials/experts/en/master/parallel/mpirun.html>`_ .
 
             For the CPU device, users need to write a dynamic cluster startup script, please see the `Dynamic Cluster
-            Startup <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/dynamic_cluster.html>`_ .
+            Startup <https://www.mindspore.cn/tutorials/experts/en/master/parallel/dynamic_cluster.html>`_ .
 
         >>> import mindspore as ms
         >>> from mindspore.communication import init, get_local_rank_size
@@ -480,14 +480,14 @@ def get_world_rank_from_group_rank(group, group_rank_id):
 
             For the Ascend devices, users need to prepare the rank table, set rank_id and device_id.
             Please see the `rank table Startup
-            <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/rank_table.html>`_
+            <https://www.mindspore.cn/tutorials/experts/en/master/parallel/rank_table.html>`_
             for more details.
 
             For the GPU devices, users need to prepare the host file and mpi, please see the `mpirun Startup
-            <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/mpirun.html>`_
+            <https://www.mindspore.cn/tutorials/experts/en/master/parallel/mpirun.html>`_
 
             For the CPU device, users need to write a dynamic cluster startup script, please see the `Dynamic Cluster
-            Startup <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/dynamic_cluster.html>`_ .
+            Startup <https://www.mindspore.cn/tutorials/experts/en/master/parallel/dynamic_cluster.html>`_ .
 
         >>> import mindspore as ms
         >>> from mindspore import set_context
@@ -539,14 +539,14 @@ def get_group_rank_from_world_rank(world_rank_id, group):
 
             For the Ascend devices, users need to prepare the rank table, set rank_id and device_id.
             Please see the `rank table Startup
-            <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/rank_table.html>`_
+            <https://www.mindspore.cn/tutorials/experts/en/master/parallel/rank_table.html>`_
             for more details.
 
             For the GPU devices, users need to prepare the host file and mpi, please see the `mpirun Startup
-            <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/mpirun.html>`_
+            <https://www.mindspore.cn/tutorials/experts/en/master/parallel/mpirun.html>`_
 
             For the CPU device, users need to write a dynamic cluster startup script, please see the `Dynamic Cluster
-            Startup <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/dynamic_cluster.html>`_ .
+            Startup <https://www.mindspore.cn/tutorials/experts/en/master/parallel/dynamic_cluster.html>`_ .
 
         >>> import mindspore as ms
         >>> from mindspore import set_context
@@ -595,14 +595,14 @@ def create_group(group, rank_ids):
 
             For the Ascend devices, users need to prepare the rank table, set rank_id and device_id.
             Please see the `rank table Startup
-            <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/rank_table.html>`_
+            <https://www.mindspore.cn/tutorials/experts/en/master/parallel/rank_table.html>`_
             for more details.
 
             For the GPU devices, users need to prepare the host file and mpi, please see the `mpirun Startup
-            <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/mpirun.html>`_ .
+            <https://www.mindspore.cn/tutorials/experts/en/master/parallel/mpirun.html>`_ .
 
             For the CPU device, users need to write a dynamic cluster startup script, please see the `Dynamic Cluster
-            Startup <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/dynamic_cluster.html>`_ .
+            Startup <https://www.mindspore.cn/tutorials/experts/en/master/parallel/dynamic_cluster.html>`_ .
 
         >>> import mindspore as ms
         >>> from mindspore import set_context
@@ -648,14 +648,14 @@ def destroy_group(group):
 
             For the Ascend devices, users need to prepare the rank table, set rank_id and device_id.
             Please see the `rank table startup
-            <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/rank_table.html>`_
+            <https://www.mindspore.cn/tutorials/experts/en/master/parallel/rank_table.html>`_
             for more details.
 
             For the GPU devices, users need to prepare the host file and mpi, please see the `mpirun startup
-            <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/mpirun.html>`_ .
+            <https://www.mindspore.cn/tutorials/experts/en/master/parallel/mpirun.html>`_ .
 
             For the CPU device, users need to write a dynamic cluster startup script, please see the `Dynamic Cluster
-            Startup <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/parallel/dynamic_cluster.html>`_ .
+            Startup <https://www.mindspore.cn/tutorials/experts/en/master/parallel/dynamic_cluster.html>`_ .
 
         >>> import mindspore as ms
         >>> from mindspore import set_context

mindspore/context.py

@@ -701,6 +701,7 @@ class _Context:
                             "enable_task_opt": (ms_ctx_param.enable_task_opt, bool),
                             "enable_grad_comm_opt": (ms_ctx_param.enable_grad_comm_opt, bool),
                             "interleaved_matmul_comm": (ms_ctx_param.interleaved_matmul_comm, bool),
+                            "bias_add_comm_swap": (ms_ctx_param.bias_add_comm_swap, bool),
                             "enable_opt_shard_comm_opt": (ms_ctx_param.enable_opt_shard_comm_opt, bool),
                             "enable_begin_end_inline_opt": (ms_ctx_param.enable_begin_end_inline_opt, bool),
                             "enable_concat_eliminate_opt": (ms_ctx_param.enable_concat_eliminate_opt, bool),
@@ -758,7 +759,8 @@ def _context():
                  strategy_ckpt_save_file=str, full_batch=bool, enable_parallel_optimizer=bool, enable_alltoall=bool,
                  all_reduce_fusion_config=list, pipeline_stages=int, pipeline_segments=int,
                  pipeline_result_broadcast=bool, parallel_optimizer_config=dict,
-                 comm_fusion=dict, strategy_ckpt_config=dict)
+                 pipeline_config=dict,
+                 comm_fusion=dict, strategy_ckpt_config=dict, force_fp32_communication=bool)
 def set_auto_parallel_context(**kwargs):
     r"""
     Set auto parallel context, only data parallel supported on CPU.
@@ -782,11 +784,10 @@ def set_auto_parallel_context(**kwargs):
     parallel_mode                parameter_broadcast
     all_reduce_fusion_config     strategy_ckpt_load_file
     enable_parallel_optimizer    strategy_ckpt_save_file
-    parallel_optimizer_config    full_batch
-    enable_alltoall              dataset_strategy
-               \                 pipeline_stages
-               \                 pipeline_result_broadcast
-               \                 auto_parallel_search_mode
+    parallel_optimizer_config    dataset_strategy
+    enable_alltoall              pipeline_stages
+    pipeline_config              auto_parallel_search_mode
+    force_fp32_communication     pipeline_result_broadcast
                \                 comm_fusion
                \                 strategy_ckpt_config
                \                 group_ckpt_save_file
@@ -850,6 +851,9 @@ def set_auto_parallel_context(**kwargs):
                        data parallel training in the benefit of time and memory saving. Currently, auto and semi auto
                        parallel mode support all optimizers in both Ascend and GPU. Data parallel mode only supports
                        `Lamb` and `AdamWeightDecay` in Ascend . Default: ``False`` .
+        force_fp32_communication (bool): A switch that determines whether reduce operators (AllReduce, ReduceScatter)
+                        are forced to use the fp32 data type for communication during communication. True is the enable
+                        switch. Default: ``False`` .
         enable_alltoall (bool): A switch that allows AllToAll operators to be generated during communication. If its
                         value is ``False`` , there will be a combination of operators such as AllGather, Split and
                         Concat instead of AllToAll. Default: ``False`` .
@@ -861,6 +865,12 @@ def set_auto_parallel_context(**kwargs):
                         Default: ``1`` .
         pipeline_result_broadcast (bool): A switch that broadcast the last stage result to all other stage in pipeline
                         parallel inference. Default: ``False`` .
+        pipeline_config (dict): A dict contains the keys and values for setting the pipeline parallelism configuration.
+                        It supports the following keys:
+
+                        - pipeline_interleave(bool): Indicates whether to enable the interleaved execution mode.
+                        - pipeline_scheduler(str): Indicates the scheduling mode for pipeline parallelism. Only support
+                          ``gpipe/1f1b``.
         parallel_optimizer_config (dict): A dict contains the keys and values for setting the parallel optimizer
                         configure. The configure provides more detailed behavior control about parallel training
                         when parallel optimizer is enabled. The configure will be effective when we use
@@ -999,6 +1009,7 @@ def reset_auto_parallel_context():
     - strategy_ckpt_save_file: ''.
     - full_batch: False.
     - enable_parallel_optimizer: False.
+    - force_fp32_communication: False
     - enable_alltoall: False.
     - pipeline_stages: 1.
     - pipeline_result_broadcast: False.
@@ -1289,7 +1300,7 @@ def set_context(**kwargs):
             If enable_graph_kernel is set to ``True`` , acceleration can be enabled.
             For details of graph kernel fusion, please check
             `Enabling Graph Kernel Fusion
-            <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/optimize/graph_fusion_engine.html>`_.
+            <https://www.mindspore.cn/tutorials/experts/en/master/optimize/graph_fusion_engine.html>`_.
         graph_kernel_flags (str):
             Optimization options of graph kernel fusion, and the priority is higher when it conflicts
             with enable_graph_kernel. Only for experienced users.
@@ -1373,7 +1384,7 @@ def set_context(**kwargs):
             range of ['ON', 'OFF'], and the default value is ``'OFF'`` .
 
             - ON: Enable the memory Offload function. On Ascend hardware platform, this parameter does not take effect
-              when jit_level of JitConfig is not set 'O0'; This parameter does not take effect when
+              when the graph compilation level is not 'O0'; This parameter does not take effect when
               memory_optimize_level is set 'O1'.
             - OFF: Turn off the memory Offload function.
         ascend_config (dict): Set the parameters specific to Ascend hardware platform. It is not set by default.
@@ -1384,16 +1395,18 @@ def set_context(**kwargs):
               is ``force_fp16`` . The value range is as follows:
 
               - force_fp16: When the operator supports both float16 and float32, select float16 directly.
-              - allow_fp32_to_fp16: When the operator does not support the float32 data type, directly reduce
-                the precision of float16.
+              - allow_fp32_to_fp16: For cube operators, use the float16. For vector operators,
+                prefer to keep the origin dtype, if the operator in model can support float32,
+                it will keep original dtype, otherwise it will reduce to float16.
               - allow_mix_precision: Automatic mixing precision, facing the whole network operator, according
                 to the built-in optimization strategy, automatically reduces the precision of some operators
                 to float16 or bfloat16.
               - must_keep_origin_dtype: Keep the accuracy of the original drawing.
               - force_fp32: When the input of the matrix calculation operator is float16 and the output supports
                 float16 and float32, output is forced to float32.
-              - allow_fp32_to_bf16: When the operator does not support the float32 data type, directly reduce
-                the precision of bfloat16.
+              - allow_fp32_to_bf16: For cube operators, use the bfloat16. For vector operators,
+                prefer to keep the origin dtype, if the operator in model can support float32,
+                it will keep original dtype, otherwise it will reduce to bfloat16.
               - allow_mix_precision_fp16: Automatic mixing precision, facing the whole network operator, automatically
                 reduces the precision of some operators to float16 according to the built-in optimization strategy.
               - allow_mix_precision_bf16: Automatic mixing precision, facing the whole network operator, according to
@@ -1434,7 +1447,7 @@ def set_context(**kwargs):
 
             - parallel_speed_up_json_path(Union[str, None]): The path to the parallel speed up json file, configuration
               can refer to `parallel_speed_up.json
-              <https://gitee.com/mindspore/mindspore/blob/r2.3.q1/config/parallel_speed_up.json>`_ .
+              <https://gitee.com/mindspore/mindspore/blob/master/config/parallel_speed_up.json>`_ .
               If its value is None or '', it does not take effect. Default None.
 
               - recompute_comm_overlap (bool): Enable overlap between recompute ops and communication ops if True.
@@ -1445,11 +1458,11 @@ def set_context(**kwargs):
                 Default: False.
               - enable_grad_comm_opt (bool): Enable overlap between dx ops and data parallel communication ops if True.
                 Currently, do not support
-                `LazyInline <https://www.mindspore.cn/docs/en/r2.3.q1/api_python/mindspore/mindspore.lazy_inline.html>`
+                `LazyInline <https://www.mindspore.cn/docs/en/master/api_python/mindspore/mindspore.lazy_inline.html>`
                 Default: False.
               - enable_opt_shard_comm_opt (bool): Enable overlap between forward ops
                 and optimizer parallel allgather communication if True. Currently, do not support
-                `LazyInline <https://www.mindspore.cn/docs/en/r2.3.q1/api_python/mindspore/mindspore.lazy_inline.html>`
+                `LazyInline <https://www.mindspore.cn/docs/en/master/api_python/mindspore/mindspore.lazy_inline.html>`
                 Default: False.
               - compute_communicate_fusion_level (int): Enable the fusion between compute and communicate.
                 Default: ``0``.

mindspore/dataset/__init__.py

@@ -21,7 +21,7 @@ Besides, this module provides APIs to sample data while loading.
 
 We can enable cache in most of the dataset with its key arguments 'cache'. Please notice that cache is not supported
 on Windows platform yet. Do not use it while loading and processing data on Windows. More introductions and limitations
-can refer `Single-Node Tensor Cache <https://www.mindspore.cn/tutorials/experts/en/r2.3.q1/dataset/cache.html>`_ .
+can refer `Single-Node Tensor Cache <https://www.mindspore.cn/tutorials/experts/en/master/dataset/cache.html>`_ .
 
 Common imported modules in corresponding API examples are as follows:
 
@@ -55,11 +55,11 @@ The specific steps are as follows:
 - Dataset operation: The user uses the dataset object method `.shuffle` / `.filter` / `.skip` / `.split` /
   `.take` / ... to further shuffle, filter, skip, and obtain the maximum number of samples of datasets;
 - Dataset sample transform operation: The user can add data transform operations
-  ( `vision transform <https://mindspore.cn/docs/en/r2.3.q1/api_python/mindspore.\
+  ( `vision transform <https://mindspore.cn/docs/en/master/api_python/mindspore.\
   dataset.transforms.html#module-mindspore.dataset.vision>`_ ,
-  `NLP transform <https://mindspore.cn/docs/en/r2.3.q1/api_python/mindspore.\
+  `NLP transform <https://mindspore.cn/docs/en/master/api_python/mindspore.\
   dataset.transforms.html#module-mindspore.dataset.text>`_ ,
-  `audio transform <https://mindspore.cn/docs/en/r2.3.q1/api_python/mindspore.\
+  `audio transform <https://mindspore.cn/docs/en/master/api_python/mindspore.\
   dataset.transforms.html#module-mindspore.dataset.audio>`_ ) to the map
   operation to perform transformations. During data preprocessing, multiple map operations can be defined to
   perform different transform operations to different fields. The data transform operation can also be a
@@ -73,7 +73,7 @@ Quick start of Dataset Pipeline
 -------------------------------
 
 For a quick start of using Dataset Pipeline, download `Load & Process Data With Dataset Pipeline
-<https://www.mindspore.cn/docs/en/r2.3.q1/api_python/samples/dataset/dataset_gallery.html>`_
+<https://www.mindspore.cn/docs/en/master/api_python/samples/dataset/dataset_gallery.html>`_
 to local and run in sequence.
 
 """

mindspore/dataset/audio/__init__.py

@@ -40,10 +40,10 @@ Descriptions of common data processing terms are as follows:
 The data transform operation can be executed in the data processing pipeline or in the eager mode:
 
 - Pipeline mode is generally used to process big datasets. Examples refer to
-  `introduction to data processing pipeline <https://www.mindspore.cn/docs/en/r2.3.q1/api_python/
+  `introduction to data processing pipeline <https://www.mindspore.cn/docs/en/master/api_python/
   mindspore.dataset.html#introduction-to-data-processing-pipeline>`_ .
 - Eager mode is more like a function call to process data. Examples refer to
-  `Lightweight Data Processing <https://www.mindspore.cn/tutorials/en/r2.3.q1/advanced/dataset/eager.html>`_ .
+  `Lightweight Data Processing <https://www.mindspore.cn/tutorials/en/master/advanced/dataset/eager.html>`_ .
 """
 from __future__ import absolute_import