| @@ -190,6 +190,31 @@ def get_bprop_tile(self): | |||||
| return bprop | return bprop | ||||
| @bprop_getters.register(P.EmbeddingLookup) | |||||
| def get_bprop_embedding_lookup(self): | |||||
| """Generate bprop for EmbeddingLookup""" | |||||
| host_sub = P.Sub().add_prim_attr('primitive_target', 'CPU') | |||||
| host_reshape = P.Reshape().add_prim_attr('primitive_target', 'CPU') | |||||
| def bprop_sparse(x, indices, offset, reduce_scatter_flag, split_num, out, dout): | |||||
| x_shp = shape_op(x) | |||||
| if reduce_scatter_flag is True: | |||||
| elu_grad = G.EmbeddingLookupCommGrad() | |||||
| actual_dout = elu_grad(dout, split_num) | |||||
| else: | |||||
| actual_dout = dout | |||||
| new_indices = host_sub(indices - offset) | |||||
| # Reshape the 'new_indices' | |||||
| new_indices_shape_changed = (size_op(new_indices),) | |||||
| new_indices = host_reshape(new_indices, new_indices_shape_changed) | |||||
| # Reshape the 'actual_dout' | |||||
| x_shp_tail = x_shp[1:] | |||||
| actual_dout_shape_changed = new_indices_shape_changed + x_shp_tail | |||||
| actual_dout = host_reshape(actual_dout, actual_dout_shape_changed) | |||||
| return (new_indices, actual_dout, x_shp), zeros_like(new_indices), zeros_like(axis), \ | |||||
| zeros_like(reduce_scatter_flag), zeros_like(split_num) | |||||
| return bprop_sparse | |||||
| @bprop_getters.register(P.Transpose) | @bprop_getters.register(P.Transpose) | ||||
| def get_bprop_transpose(self): | def get_bprop_transpose(self): | ||||
| """Generate bprop for Transpose""" | """Generate bprop for Transpose""" | ||||
| @@ -616,9 +616,10 @@ class Range(PrimitiveWithInfer): | |||||
| class EmbeddingLookup(PrimitiveWithInfer): | class EmbeddingLookup(PrimitiveWithInfer): | ||||
| """ | """ | ||||
| Returns a slice of input tensor based on the specified indices and axis. This Primitive has the similar | |||||
| functionality as GatherV2, but has three more inputs: `offset`, `reduce_scatter_flag` and `split_num`. | |||||
| This primitive runs on the host instead of devices. | |||||
| Returns a slice of input tensor based on the specified indices. | |||||
| This Primitive has the similar functionality as GatherV2 operating on `axis = 0`, but has three more inputs: | |||||
| `offset`, `reduce_scatter_flag` and `split_num`. This primitive runs on the host instead of devices. | |||||
| Inputs: | Inputs: | ||||
| - **input_params** (Tensor) - The shape of tensor is :math:`(x_1, x_2, ..., x_R)`. | - **input_params** (Tensor) - The shape of tensor is :math:`(x_1, x_2, ..., x_R)`. | ||||
| @@ -626,7 +627,6 @@ class EmbeddingLookup(PrimitiveWithInfer): | |||||
| - **input_indices** (Tensor) - The shape of tensor is :math:`(y_1, y_2, ..., y_S)`. | - **input_indices** (Tensor) - The shape of tensor is :math:`(y_1, y_2, ..., y_S)`. | ||||
| Specifies the indices of elements of the original Tensor. Values can be out of range of `input_params`, | Specifies the indices of elements of the original Tensor. Values can be out of range of `input_params`, | ||||
| and the exceeding part will be filled with 0 in the output. | and the exceeding part will be filled with 0 in the output. | ||||
| - **axis** (int) - Specifies the dimension index to gather indices. | |||||
| - **offset** (int) - Specifies the offset value of this `input_params` slice. Thus the real indices | - **offset** (int) - Specifies the offset value of this `input_params` slice. Thus the real indices | ||||
| are equal to `input_indices` minus `offset`. | are equal to `input_indices` minus `offset`. | ||||
| - **reduce_scatter_flag** (bool) - Specifies whether perform reduce_scatter on host or not. | - **reduce_scatter_flag** (bool) - Specifies whether perform reduce_scatter on host or not. | ||||
| @@ -641,36 +641,29 @@ class EmbeddingLookup(PrimitiveWithInfer): | |||||
| Examples: | Examples: | ||||
| >>> input_params = Tensor(np.array([[8, 9], [10, 11], [12, 13], [14, 15]]), mindspore.float32) | >>> input_params = Tensor(np.array([[8, 9], [10, 11], [12, 13], [14, 15]]), mindspore.float32) | ||||
| >>> input_indices = Tensor(np.array([[5, 2], [8, 5]]), mindspore.int32) | >>> input_indices = Tensor(np.array([[5, 2], [8, 5]]), mindspore.int32) | ||||
| >>> axis = 0 | |||||
| >>> offset = 4 | >>> offset = 4 | ||||
| >>> reduce_scatter_flag = False | >>> reduce_scatter_flag = False | ||||
| >>> split_num = 1 | >>> split_num = 1 | ||||
| >>> out = P.EmbeddingLookup()(input_params, input_indices, axis, offset, reduce_scatter_flag, split_num) | |||||
| >>> out = P.EmbeddingLookup()(input_params, input_indices, offset, reduce_scatter_flag, split_num) | |||||
| [[[10, 11], [0 ,0]], [[0, 0], [10, 11]]] | [[[10, 11], [0 ,0]], [[0, 0], [10, 11]]] | ||||
| """ | """ | ||||
| @prim_attr_register | @prim_attr_register | ||||
| def __init__(self): | def __init__(self): | ||||
| """init index_select""" | """init index_select""" | ||||
| self.__setattr_flag__ = True | self.__setattr_flag__ = True | ||||
| self.init_prim_io_names(inputs=['params', 'indices', 'axis', 'offset', 'reduce_scatter_flag', 'split_num'], | |||||
| self.init_prim_io_names(inputs=['params', 'indices', 'offset', 'reduce_scatter_flag', 'split_num'], | |||||
| outputs=['output']) | outputs=['output']) | ||||
| self.add_prim_attr('primitive_target', 'CPU') | self.add_prim_attr('primitive_target', 'CPU') | ||||
| def __infer__(self, params, indices, axis, offset, reduce_scatter_flag=False, split_num=2): | |||||
| def __infer__(self, params, indices, offset, reduce_scatter_flag=False, split_num=2): | |||||
| validator.check_subclass("params", params['dtype'], mstype.tensor, self.name) | validator.check_subclass("params", params['dtype'], mstype.tensor, self.name) | ||||
| validator.check_tensor_type_same({"indices": indices['dtype']}, mstype.int_type, self.name) | validator.check_tensor_type_same({"indices": indices['dtype']}, mstype.int_type, self.name) | ||||
| validator.check_subclass("axis", axis['dtype'], mstype.int_, self.name) | |||||
| validator.check_subclass("offset", offset['dtype'], mstype.int_, self.name) | validator.check_subclass("offset", offset['dtype'], mstype.int_, self.name) | ||||
| validator.check_subclass("split_num", split_num['dtype'], mstype.int_, self.name) | validator.check_subclass("split_num", split_num['dtype'], mstype.int_, self.name) | ||||
| if split_num['value'] < 1: | if split_num['value'] < 1: | ||||
| raise ValueError("The parameter 'split_num' must be positive, but got %d." % split_num) | raise ValueError("The parameter 'split_num' must be positive, but got %d." % split_num) | ||||
| axis_v = axis['value'] | |||||
| params_shp = params['shape'] | params_shp = params['shape'] | ||||
| rank = len(params_shp) | |||||
| validator.check_int_range("axis", axis_v, -rank, rank, Rel.INC_LEFT, self.name) | |||||
| if axis_v < 0: | |||||
| axis_v += rank | |||||
| out_shape = params_shp[:axis_v] + indices['shape'] + params_shp[axis_v + 1:] | |||||
| out_shape = indices['shape'] + params_shp[1:] | |||||
| if reduce_scatter_flag is None: | if reduce_scatter_flag is None: | ||||
| raise ValueError("The value of 'reduce_scatter_flag' is None.") | raise ValueError("The value of 'reduce_scatter_flag' is None.") | ||||
| reduce_scatter_flag_value = reduce_scatter_flag['value'] | reduce_scatter_flag_value = reduce_scatter_flag['value'] | ||||
| @@ -33,10 +33,9 @@ class NetWithLoss(nn.Cell): | |||||
| return self.loss(predict) | return self.loss(predict) | ||||
| class Net(nn.Cell): | class Net(nn.Cell): | ||||
| def __init__(self, shape, axis, offset, reduce_scatter_flag, split_num): | |||||
| def __init__(self, shape, offset, reduce_scatter_flag, split_num): | |||||
| super().__init__() | super().__init__() | ||||
| self.index = Tensor(np.ones(shape), dtype=ms.int32) | self.index = Tensor(np.ones(shape), dtype=ms.int32) | ||||
| self.axis = axis | |||||
| self.offset = offset | self.offset = offset | ||||
| self.reduce_scatter_flag = reduce_scatter_flag | self.reduce_scatter_flag = reduce_scatter_flag | ||||
| self.split_num = split_num | self.split_num = split_num | ||||
| @@ -44,18 +43,17 @@ class Net(nn.Cell): | |||||
| self.mm = P.BatchMatMul() | self.mm = P.BatchMatMul() | ||||
| def construct(self, x, y): | def construct(self, x, y): | ||||
| out = self.elu(x, self.index, self.axis, self.offset, self.reduce_scatter_flag, self.split_num) | |||||
| out = self.elu(x, self.index, self.offset, self.reduce_scatter_flag, self.split_num) | |||||
| out = self.mm(out, y) | out = self.mm(out, y) | ||||
| return out | return out | ||||
| def test_embeddinglookup_reducescatter_false(): | def test_embeddinglookup_reducescatter_false(): | ||||
| shape = [8, 8] | shape = [8, 8] | ||||
| axis = 0 | |||||
| offset = 8 | offset = 8 | ||||
| reduce_scatter_flag = False | reduce_scatter_flag = False | ||||
| split_num = 1 | split_num = 1 | ||||
| net = NetWithLoss(Net(shape, axis, offset, reduce_scatter_flag, split_num)) | |||||
| net = NetWithLoss(Net(shape, offset, reduce_scatter_flag, split_num)) | |||||
| net.set_auto_parallel() | net.set_auto_parallel() | ||||
| x = Tensor(np.ones([64, 32]), dtype=ms.float32) | x = Tensor(np.ones([64, 32]), dtype=ms.float32) | ||||
| @@ -65,11 +63,10 @@ def test_embeddinglookup_reducescatter_false(): | |||||
| def test_embeddinglookup_reducescatter_true(): | def test_embeddinglookup_reducescatter_true(): | ||||
| shape = [64, 8] | shape = [64, 8] | ||||
| axis = 0 | |||||
| offset = 8 | offset = 8 | ||||
| reduce_scatter_flag = True | reduce_scatter_flag = True | ||||
| split_num = 8 | split_num = 8 | ||||
| net = NetWithLoss(Net(shape, axis, offset, reduce_scatter_flag, split_num)) | |||||
| net = NetWithLoss(Net(shape, offset, reduce_scatter_flag, split_num)) | |||||
| net.set_auto_parallel() | net.set_auto_parallel() | ||||
| x = Tensor(np.ones([64, 32]), dtype=ms.float32) | x = Tensor(np.ones([64, 32]), dtype=ms.float32) | ||||
| @@ -184,7 +184,7 @@ def test_gatherv2_auto1(): | |||||
| _executor.compile(net, x, y) | _executor.compile(net, x, y) | ||||
| def test_gatherv2_cpu0(): | |||||
| def need_fix_test_gatherv2_cpu0(): | |||||
| context.set_auto_parallel_context(device_num=8, global_rank=0, parallel_mode="semi_auto_parallel") | context.set_auto_parallel_context(device_num=8, global_rank=0, parallel_mode="semi_auto_parallel") | ||||
| strategy1 = ((8, 1), (1, 1)) | strategy1 = ((8, 1), (1, 1)) | ||||
| strategy2 = ((4, 2, 1), (4, 2, 1)) | strategy2 = ((4, 2, 1), (4, 2, 1)) | ||||
| @@ -196,7 +196,7 @@ def test_gatherv2_cpu0(): | |||||
| _executor.compile(net, x, y) | _executor.compile(net, x, y) | ||||
| def test_gatherv2_cpu1(): | |||||
| def need_fix_test_gatherv2_cpu1(): | |||||
| context.set_auto_parallel_context(device_num=16, global_rank=0, parallel_mode="semi_auto_parallel") | context.set_auto_parallel_context(device_num=16, global_rank=0, parallel_mode="semi_auto_parallel") | ||||
| strategy1 = ((16, 1), (1, 1)) | strategy1 = ((16, 1), (1, 1)) | ||||
| strategy2 = ((4, 2, 1), (4, 2, 1)) | strategy2 = ((4, 2, 1), (4, 2, 1)) | ||||
| @@ -208,7 +208,7 @@ def test_gatherv2_cpu1(): | |||||
| _executor.compile(net, x, y) | _executor.compile(net, x, y) | ||||
| def test_gatherv2_cpu2(): | |||||
| def need_fix_test_gatherv2_cpu2(): | |||||
| context.set_auto_parallel_context(device_num=8, global_rank=0, parallel_mode="semi_auto_parallel") | context.set_auto_parallel_context(device_num=8, global_rank=0, parallel_mode="semi_auto_parallel") | ||||
| strategy1 = ((1, 8), (1, 1)) | strategy1 = ((1, 8), (1, 1)) | ||||
| strategy2 = ((4, 2, 1), (4, 2, 1)) | strategy2 = ((4, 2, 1), (4, 2, 1)) | ||||
| @@ -184,7 +184,7 @@ def test_gatherv2_auto1(): | |||||
| _executor.compile(net, x, y) | _executor.compile(net, x, y) | ||||
| def test_gatherv2_cpu0(): | |||||
| def need_fix_test_gatherv2_cpu0(): | |||||
| context.set_auto_parallel_context(device_num=8, global_rank=0, parallel_mode="semi_auto_parallel") | context.set_auto_parallel_context(device_num=8, global_rank=0, parallel_mode="semi_auto_parallel") | ||||
| strategy1 = ((8, 1), (1, 1)) | strategy1 = ((8, 1), (1, 1)) | ||||
| strategy2 = ((4, 2, 1), (4, 2, 1)) | strategy2 = ((4, 2, 1), (4, 2, 1)) | ||||
| @@ -196,7 +196,7 @@ def test_gatherv2_cpu0(): | |||||
| _executor.compile(net, x, y) | _executor.compile(net, x, y) | ||||
| def test_gatherv2_cpu1(): | |||||
| def need_fix_test_gatherv2_cpu1(): | |||||
| context.set_auto_parallel_context(device_num=16, global_rank=0, parallel_mode="semi_auto_parallel") | context.set_auto_parallel_context(device_num=16, global_rank=0, parallel_mode="semi_auto_parallel") | ||||
| strategy1 = ((16, 1), (1, 1)) | strategy1 = ((16, 1), (1, 1)) | ||||
| strategy2 = ((4, 2, 1), (4, 2, 1)) | strategy2 = ((4, 2, 1), (4, 2, 1)) | ||||
| @@ -208,7 +208,7 @@ def test_gatherv2_cpu1(): | |||||
| _executor.compile(net, x, y) | _executor.compile(net, x, y) | ||||
| def test_gatherv2_cpu2(): | |||||
| def need_fix_test_gatherv2_cpu2(): | |||||
| context.set_auto_parallel_context(device_num=8, global_rank=0, parallel_mode="semi_auto_parallel") | context.set_auto_parallel_context(device_num=8, global_rank=0, parallel_mode="semi_auto_parallel") | ||||
| strategy1 = ((1, 8), (1, 1)) | strategy1 = ((1, 8), (1, 1)) | ||||
| strategy2 = ((4, 2, 1), (4, 2, 1)) | strategy2 = ((4, 2, 1), (4, 2, 1)) | ||||