[PHI]Add new Tensor type and migrate save_combine kernel #47856

YuanRisheng · 2022-11-10T10:05:22Z

PR types

Others

PR changes

Others

Describe

背景：
当前PHI算子库对于功能性算子的开发支持不足，需要完善功能机制，支持全量场景算子研发
本PR工作：
数据结构类扩展:
a. 添加了核心Tensor类型：ExtendedTensor
b. 改写原Fluid下Vocab数据结构
c. 添加了RawTensor数据结构，兼容Fluid下RAW type类型输出接入到Phi Kernel
添加后的PHI Tensor结构如下：

说明：

ExtendedTensor是用户实现非Tensor输入输出的一个接口，用户可以通过继承ExtendedTensor，实现在PHI中加入Kernel所需要的自定义输入输出类型。
将原先Fluid下的Vocab类型(本质是std::unordered_map<std::wstring, std::int32_t>实现)进行改写，通过继承ExtendedTensor重新改写实现，以便于接入到PHI执行体系，改写后PHI Kernel可以在Kernel里获取Vocab
实现了RawTensor，主要是为了兼容原Fluid下的RawType数据类型，这样Fluid功能性算子里的RawType类型数据在PHI下也有对应的实现了

Kernel注册功能扩展：
d，扩展注册功能，支持ExtendedTensor类型注册，接入Kernel执行体系
e, 支持vector<Vocab*>类型作为Phi Kernel的输入，并兼容Fluid体系

改写Fluid为PHI Kernel:
f，改写save_combine算子为PHI kernel的形式

paddle-bot · 2022-11-10T10:05:27Z

你的PR提交成功，感谢你对开源项目的贡献!
请关注后续CI自动化测试结果，详情请参考Paddle-CI手册。
Your PR has been submitted. Thanks for your contribution!
Please wait for the result of CI firstly. See Paddle CI Manual for details.

…registry

zyfncg · 2022-11-28T07:09:40Z

paddle/phi/core/extended_tensor.cc

+int64_t ExtendedTensor::numel() const {
+  PADDLE_THROW(phi::errors::Unavailable(
+      "ExtendedTensor does not support `numel` method."));
+}
+
+const DDim& ExtendedTensor::dims() const {
+  PADDLE_THROW(phi::errors::Unavailable(
+      "ExtendedTensor does not support `dims` method."));
+}


这些函数可以声明为纯虚函数吗？

这里是故意设计成这样的，减少自定义输入输出类型继承后的不合理代码

zyfncg · 2022-11-28T07:16:21Z

paddle/phi/core/extended_tensor.h

+/// \brief The ExtendedTensor is a interface for custom designed class.
+/// If you want to pass some self-designed data as input/output to kernels,
+/// you can inherit from this class to store your self-designed data.
+class ExtendedTensor : public TensorBase,


ExtendedTensor看上去没有不同于TensorBase的特殊逻辑，对于新增的自定义Tensor类型直接继承TensorBase和继承ExtendedTensor有什么差异？

主要原因有如下俩点：
1，自定义类型不是Tensor，语义上的差异并不适合继承自TensorBase，使用ExtendedTensor可以对外暴露一致的接口，所有不具备Tensor性质的输入输出都可以继承这个类型，这样就可以防止出现直接继承TensorBase造成TensorBase体系概念混乱的问题。
2，如上所说，直接继承TensorBase需要实现虚函数，但是其实自定义类型并不需要，减少开发者在开发自定义类型中实现不合理代码

如果自定义类型不是Tensor的话，ExtendedTensor可以不继承TensorBase吗？

不继承TensorBase接入不到Phi Kernel执行体系，期间也考虑过将其非Tensor当做属性来处理的方案，但是实践后发现问题很多而且很不合理，目前能想到的比较合适的做法就是ExtendedTensor继承TensorBase

chenwhql · 2022-11-11T08:01:30Z

paddle/phi/core/kernel_factory.h

@@ -142,6 +142,7 @@ enum class AttributeType {
  DATA_TYPE,
  DATA_LAYOUT,
  PLACE,
+  STRING_PTR,


这个属性本身就比较trick，属性类型要有一定通用性，如果后续还有map_ptr之类的怎么处理呢？

评论的这个代码貌似不是最新代码，最新代码已经修改

chenwhql · 2022-11-11T08:01:54Z

paddle/phi/core/kernel_registry.h

@@ -27,6 +27,7 @@
 #include "paddle/phi/core/kernel_utils.h"
 #include "paddle/phi/core/macros.h"
 #include "paddle/phi/core/type_defs.h"
+#include "paddle/phi/core/vocab.h"


vocab是废弃概念，不建议继续迁移过来，这种应该在fluid内保留，将来是要删除的

同上，不是最新代码

chenwhql · 2022-11-28T09:34:24Z

cmake/operators.cmake

@@ -26,6 +26,36 @@ function(find_register FILENAME PATTERN OUTPUT)
      PARENT_SCOPE)
 endfunction()

+function(find_phi_register FILENAME ADD_PATH)
+  # find the op_name of REGISTER_OPERATOR(op_name, ...), REGISTER_OP_CPU_KERNEL(op_name, ...) , etc.


下面find的是PD_REGISTER_KERNEL？注释是不是不太match

chenwhql · 2022-11-28T09:41:19Z

paddle/phi/core/cplus_string.h

+
+namespace phi {
+
+class CPlusString : public ExtendedTensor,


这个也放到fluid下？phi不需要这个，且已经有了StringTensor，如果只有save_combine使用的话，建议放到save_combine下，这种数据结构比较trick，建议同时在加一些注释说明下背景，避免扩散使用

参照楼下刚刚的回复，由于处理Raw类型，必须要在kernel_registry里对CPlusString进行处理，如果把该结构挪到Fluid，将造成对Fluid的依赖，所以解决这个问题的根本做法是找到更好的处理Raw类型的方案

chenwhql · 2022-11-28T09:43:46Z

paddle/phi/core/kernel_registry.h

@@ -191,6 +198,11 @@ struct KernelArgsParseFunctor<Return_ (*)(Args_...)> {
                               default_tensor_layout,
                               default_key.dtype(),
                               arg_type);
+      } else if (arg_type == std::type_index(typeid(CPlusString*))) {


加这个分支是必要的吗，对外出口仅使用ExtendedTensor

这个是处理Raw类型的，目前没想到更好的解决方式，Raw类型必须要在Kernel参数列表里指定数据类型，不能使用ExtendedTensor，不然在BuildKernelContext中，无法确定variable是什么类型拿不到里边的数据，所以这里就加分支了

chenwhql · 2022-12-02T03:30:49Z

paddle/phi/core/raw_tensor.h

+
+namespace phi {
+
+class RawTensor : public ExtendedTensor,


建议加注释说明下为什么需要这个tensor？什么背景？

如果phi中的逻辑不需要适配这个tensor，建议还是放到fluid/framework中，phi中仅保留ExtendedTensor作为唯一扩展出口

chenwhql · 2022-12-05T02:22:15Z

paddle/fluid/framework/raw_tensor.h

+  }
+
+  template <typename T>
+  T* Get() {


Get方法返回值一般是const？Get是指返回一个不可修改的只读值，GetMutable至返回一个可修改的读写值

chenwhql · 2022-12-05T02:24:34Z

paddle/fluid/framework/string_array.h

+
+  std::string::iterator begin() { return data_.begin(); }
+
+  std::string* Get() { return &data_; }


同上，这个属于GetMutable？

chenwhql · 2022-12-05T02:25:18Z

paddle/fluid/operators/save_combine_op.cu

-WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-See the License for the specific language governing permissions and
-limitations under the License. */
+// Copyright (c) 2022 PaddlePaddle Authors. All Rights Reserved.


这种不需要改文件日期，以文件创建日期为基准

chenwhql · 2022-12-05T02:26:23Z

paddle/phi/core/kernel_context.cc

@@ -13,6 +13,7 @@
 // limitations under the License.

 #include "paddle/phi/core/kernel_context.h"
+#include <glog/logging.h>


不需要的glog依赖建议移除

chenwhql · 2022-12-06T08:33:30Z

paddle/fluid/framework/raw_tensor.h

+    return created_data;
+  }
+
+  // template <typename T>


这里注释还有用吗？需要移除吗？

谢谢，下个PR进行移除

…registry

jiahy0825 · 2022-12-09T03:15:52Z

paddle/fluid/framework/raw_tensor.h

+/// \brief Fluid Kernel and PHI Kernel will be unified in the future.
+/// So, we need a class in PHI that can represent the RAW type in Fluid.
+/// The RawTensor is for PHI Kernel that has RAW type arguments.
+class RawTensor : public phi::ExtendedTensor,


fluid 下的 RawType 是一个数据类型，然而在 phi 下变成了 "Tensor"，这个概念是有差异的：

后续是否考虑在 phi 下对 enum DataType 做扩展？（phi 下如何表示 RawTensor 的数据类型？）

如果 phi 下需要做数据类型转换，RawType 如何对应到 phi 下的数据类型？（是否会扩展 TransToProtoVarType函数？）

RawType in the fluid is datatype. However, it corresponds to "Tensor" in phi. The concepts between RawType and RawTensor have differences:

Will phi extend enum DataType in the future? (How to represent the datatype of RawTensor? )

If we need to transform the datatype between fluid and phi, how does RawType matches the datatype under phi?

这里RawTensor命名后边可以更改为Raw，可能理解上就更准确一点。
准确意义上来说，RawType在fluid下不是一个具体的数据类型，我的理解把它作为一个标志更为合理一点，类似于DataType里的Undefined。对于RawType的具体类型（可以是任意类型）是在执行Kernel的时候确定，Fluid由于是结构体类型的Kernel，其对于RawType的具体类型处理可以通过传入的参数「ExecutionContext」来展开进行。但是对于PHI Kernel，由于是函数式的，其函数参数类型必须明确，所以这里必须存在一个数据结构能表示不确定的数据类型。所以对于以上提出的俩点：
1，DataType里不需要扩展，从上边可以看出，Raw类型的在PHI下必须是一个类类型，DataType里存放的都是基础数据类型，所以显然是不合适的。
2，PHI下也不需要类型转换，因为RawType在Fluid下就不是一个具体的类型

XiaoguangHu01

LGTM

…e#47856) * add new tensor * fix windows compile bugs * fix ci bugs * fix ci bugs * fix ci bugs * perfect according comment * fix ci compile bugs * add raw tensor * fix ci bugs * modify code by comment * delete String

* merge gpugraph to develop, fix code style * update for untrainable params for stage3. (#48577) * merge gpugraph to develop, trigger ci * [CodeStyle][isort][Dy2St] sort imports in test_error (#48746) * [CodeStyle][isort][Dy2St] sort imports in test_error * update lineno * Clear extra input (Bias, ResidualData) in OpMaker of conv2d (#47579) * delete Bias and ResidualData in OpMaker of conv2d * delete extra input of conv3d * refactor pass of conv_bias_fusion * fix mkldnn dependency * fix mkldnn compile * fix test_conv_bias_mkldnn_fuse_pass * police some code * remove useless log * fix analyzer_vit_ocr_tester * fix conv_activation_mkldnn_fuse_pass * fix test_analyzer_ocr * add fused_conv_sig * fix performence regression * fix performance regression * make bilinear interpolate stable. (#48644) * make bilinear interpolate stable. * fix code * clear tmp var in ptq (#48660) * merge gpugraph to develop, fix py-api comment * merge gpugraph to develop, fix mac-python3 * merge gpugraph to develop, fix mac-python3 * [Dy2St] replace deprecated `load_module` with `exec_module` (#48679) * merge gpugraph to develop, fix mac-python3 * modify d2d copy to xpu::copy in xpu kernel, test=kunlun (#48710) * rm _test_eager_guard (#48767) * delete sampling_id api (#48543) * [NPU] add FLAGS_npu_storage_format env to enable npu storage format, test=develop (#48774) * optimize nchw<->nhwc kernel in fp16 model (#48692) * fix: oss just support sm>=75 (#48731) * update kl1 op list and optimize matmul unitest for kunlun (#48775) *test=kunlun * Fix accuracy fp16 kernel return fp32 tensor error (#48803) * [phi::DenseTensor] Replace Tensor with phi::DenseTensor (#48682) * [Zero-Dim] Support 0D for paddle.diagflat (#48735) * [Zero-Dim] Support 0D for paddle.diagflat * 【fluid api clear】Move batch norm1 (#47965) * modify slice infershape * code style * modify slice_unittest * temp fix * batch_norm api move * code_style * codestyle * ci_static * add __init__ * reset other change * revert .cc * add import batchnorm * conflict and revert * fix bug * fix third conflict one day * fix conflict * fix conflict bug * fix conflict bug * modify api * code_style * modify doc * add lost doc stable * fix conflict bug * ci lack of gpu * [remove fluid] PRelu BilinearTensorProduct Conv2DTranspose SequenceConv RowConv (#48654) * [remove fluid] PRelu BilinearTensorProduct * [remove fluid] PRelu BilinearTensorProduct Conv2DTranspose SequenceConv RowConv * [remove fluid] PRelu BilinearTensorProduct Conv2DTranspose SequenceConv RowConv * [remove fluid] PRelu BilinearTensorProduct Conv2DTranspose SequenceConv RowConv * [remove fluid] PRelu BilinearTensorProduct Conv2DTranspose SequenceConv RowConv * [remove fluid] PRelu BilinearTensorProduct Conv2DTranspose SequenceConv RowConv * [remove fluid] PRelu BilinearTensorProduct Conv2DTranspose SequenceConv RowConv * [remove fluid] PRelu BilinearTensorProduct Conv2DTranspose SequenceConv RowConv * merge gpugraph to develop, rollback graph_send_recv * fix ci (#48730) * Remove reduntant numpy output in Example code (1/3), test=document_fix (#48678) * 修改了英文API文档 (#48219) * 修改paddle.nn.dynamic_decode，paddle.nn.functional.diag_embed 示例 * mma qk tensor_core (#48087) * use mma for QK dot computing in fused_multi_transformer. * Update fused_multi_transformer_op.cu.h * remove lrn which is not used in paddle 2.0 (#47945) * replace scatter_nd and scatter_nd_add with paddle.scatter_nd and (#47960) paddle.scatter_nd_add * [PHI] Migrate mul_grad kernel (#48061) * cleanup unused code * unify is_int8 is_bfloat16 * Simplify matmul_v2 FWD kernel * remove RunKernel methods * remove import namespace * remove headers * clean fluid/phi cross imports * remove fluid axpy_handler * delete fluid methods * activations * OneDNNMemDesc * MKLDNNFormatForSize * MatchShapeToLayout * MKLDNNMemoryFormat * MKLDNNFormat * ReorderMKLDNNHandler * to_void_cast * review suggestions * interpolate * remove fluid depedency * init * ExecuteMatMulV2 * rm fluid kernel * matmul_grad * remove mutable_data * mul_grad * delete unnecessary shape and slice op (#48112) * 修改英文文档。 * 修改segment operator等英文文档。 * 重新修改了paddle.einsum，paddle.unique_consecutive， paddle.disable_signal_handler的英文文档格式。 * 重新修改了英文文档格式。;test=docs_preview * Update extension.py * 重新修改了英文文档格式。;test=docs_preview * 重新修改了英文文档格式。待验收： - paddle.linalg.svd - paddle.nn.functional.diag_embed - paddle.set_grad_enabled - paddle.disable_signal_handler - paddle.cumprod - paddle.devaice.cuda.stream_guard 待修改： - paddle.nn.dynamic_decode - paddle.einsum - paddle.unique_consecutive - paddle.linalg.svd - paddle.uncubate.segment_min - paddle.uncubate.segment_max - paddle.uncubate.segment_sum - paddle.uncubate.segment_mean ;test=docs_preview * 重新修改了英文文档格式。待验收： - paddle.linalg.svd - paddle.nn.functional.diag_embed - paddle.set_grad_enabled - paddle.disable_signal_handler - paddle.cumprod - paddle.devaice.cuda.stream_guard - paddle.nn.dynamic_decode - paddle.unique_consecutive - paddle.linalg.svd 待修改： - paddle.einsum - paddle.incubate.segment_min - paddle.incubate.segment_max - paddle.incubate.segment_sum - paddle.incubate.segment_mean ;test=docs_preview * 重新修改了英文文档格式。待验收： - paddle.linalg.svd - paddle.nn.functional.diag_embed - paddle.set_grad_enabled - paddle.disable_signal_handler - paddle.cumprod - paddle.devaice.cuda.stream_guard - paddle.nn.dynamic_decode - paddle.unique_consecutive - paddle.linalg.svd 待修改： - paddle.einsum - paddle.incubate.segment_min - paddle.incubate.segment_max - paddle.incubate.segment_sum - paddle.incubate.segment_mean ;test=docs_preview * update * test=docs_preview * update formula; test=docs_preview * update formula; test=docs_preview * remove this operator; test=docs_preview * add hyper link; test=docs_preview * add default value; test=docs_preview * update format; test=docs_preview * empty commit; test=docs_preview * fix codestyle issues; test=docs_preview * empty commit; test=docs_preview Co-authored-by: lzy <569782149@qq.com> Co-authored-by: Vvsmile <450864116@qq.com> Co-authored-by: Sławomir Siwek <slawomir.siwek@intel.com> Co-authored-by: RichardWooSJTU <37864677+RichardWooSJTU@users.noreply.github.com> Co-authored-by: Ligoml <39876205+Ligoml@users.noreply.github.com> Co-authored-by: Nyakku Shigure <sigure.qaq@gmail.com> * [PHI] Migrate squeeze and squeeze_grad kernels (#48634) * squeeze kernel * squeze fwd * whitespace * 修复paddle.nn.functinal包和paddle.nn包下API文档 (#48581) * assign cve number to pdsa, test=document_fix (#48846) * [fluid remove]: remove paddle.fluid.layers.yolo_box and paddle.fluid.layers.yolov3_loss (#48722) * remove paddle.fluid.layers.nn.temporal_shift * code check * rm unittest * remove fluid.yolo_box * remove fluid.yolov3_loss * change the comments of yolov3_loss to yolo_loss * merge gpugraph to develop, fix windows compile * merge gpugraph to develop, fix windows compile * merge gpugraph to develop, fix windows compile * Try add eval() to speedup the eigen performance. (#48855) * [Fluid Clean]move inplace_apis_indygraph_only from paddle.flud.dygraph.inplace_utils to paddle.utils (#48744) * move inplace_apis_indygraph_only from paddle.flud.dygraph.inplace_utils to paddle.utils * modify conflict * modify conflict * modify conflict * modify conflict * modify conflict * modify conflict * modify conflict * modify static-check ci error * fix conflict * modify failed tests * fix conflict * fix conflict * fix pool2d examples * modify conflict * fix failed tests * fix conflict * fix failed tests * modfiy problem of deleting pool2d * merge gpugraph to develop, fix windows compile * clean fluid task: transfer gaussian random api (#48529) * Delete duplicate quant nodes in QAT (#48751) * rm autograd func dynamic eager tests (#48788) * Setuptools optimization (#48770) * optimize setup.py * modify setup.py * modify setup.py * modify setup.py * modify setup.py after zhangbo reviewed * [CodeStyle][F811] fix some test cases shadowed by the same name (#48745) * [CodeStyle][F811] fix some unittests * fix setup.py * remove ignore from flake8 config * remove repeat TestAbsDoubleGradCheck * fix rrelu test * fix fft ut * add noqa in fluid.lstm ut * add rtol and atol in test_matmul_v2_op * update rtol * empty commit * empty commit * revert changes in matmul ut and add noqa * rename test case name * set free_when_no_cache_hit default value to true (#48815) * [Clean Fluid] Rm and mv some fluid dygrah apis (#48576) Remove fluid dygrah apis GroupNorm TreeConv Move fluid dygraph apis Flatten SpectralNorm * [Inference] inference add cinn interface (#48741) * Clean and migrate fluid APIs of paddle.fluid.layers.control_flow (#48233) * Merge branch 'reduce_sum' of https://github.com/GhostScreaming/Paddle into mine_fluid_clean_common. * Fix some bugs. * Clean APIs in python/paddle/fluid/layers/control_flow.py * Polish code style. * Change API. * Fix some bugs. * Fix some bugs. * remove gpu_info.h from phi dependencies (#48811) * [Paddle Inference] Add add onehot trt converter (#48655) * add onehot trt converter * add unitest * fix bug * opt code * fix bug * fix depth_tensor * fix unitest * fix bug * fix unitest * fix bug * fix bug * fix bug * fix bug * [PHI decoupling] remove bbox_util.h from phi dependencies (#48761) * remove bbox_util.h from phi * add file bbox_util.h * reframe bbox_util.h * Optimize Paddle diagonal (#47904) * [API Clean]Clean __all__ to avoid exposing usless API (#48713) * [API Clean]Clean __all__ to avoid exposing usless API * fix import * fix typo * remove tracedLayer unittest * Clean fluid APIs in distributed and fleet files (#48851) * Fix bug of reduce_sum op. When input.numel() > INT32_MAX, its result is wrong. * Remove climits. * Clean fluid API in paddle/distributed and paddle/fleetx folders. Include following files: python/paddle/distributed/__init__.py python/paddle/distributed/collective.py python/paddle/distributed/fleet/utils/fs.py python/paddle/distributed/fleet/utils/hybrid_parallel_inference.py python/paddle/distributed/fleet/utils/hybrid_parallel_util.py python/paddle/distributed/fleet/utils/internal_storage.py python/paddle/distributed/launch/context/device.py python/paddle/distributed/parallel.py python/paddle/distributed/parallel_with_gloo.py python/paddle/distributed/spawn.py python/paddle/framework/__init__.py To be mentioned, 'paddle.fluid.dygraph.parallel.ParallelEnv' and 'fluid.framework.core' keeps unchanged in those files. ParallelEnv is used by paddle.fluid.dygraph.parallel.DataParallel. However, APIs in paddle.fluid.dygraph.parallel can't be migrated to paddle.distributed, as there exists cyclic import dependencies in modules like paddle.static, paddle.tensor. And 'fluid.framework.core' will be changed to import framework.core after fluid.core is transmitted. * Change TODO authors. * rm kunlun xpu2_op_list (#48826) *test=kunlun * remove detection_output, iou_similarity and bipartite_match (#48773) * Set WaiterType of kGpuSync to kCPU (#48758) * [Migrate Fluid] Migrate Decoder, BeamSearchDecoder (#48754) * [Inference] Enable infer shape cache. (#48312) * [Fluid Clean] remove unfold, deformable_roi_pooling, shard_index, hard_swish, mish, uniform_random, unbind (#48451) * fix-gpups setup.py (#48888) * fix-gpups * test=document_fix * [PHI decoupling] move cuda_graph from fluid to phi (#48686) * move cuda_graph from fluid to phi * move device_memory_aligment from fluid to phi * Revert "move device_memory_aligment from fluid to phi" This reverts commit b92fcd3. * update xpu cmake * fix english docs typo errors (#48599) * fix english docs typo errors the errors in docs as same as chinese pr 5468 * update docs; test=docs_preview Co-authored-by: Ligoml <39876205+Ligoml@users.noreply.github.com> * [XPU] add load op into oplist. (#48860) * [XPU] add load op into oplist. * remove test_sampling_id_op_xpu.py * 【fluid clean】remove fluid.dygraph.rnn.lstmcell and fluid.dygraph.rnn.grucell (#48719) * refine bsd doc (#48882) * [Paddle Inference] General optimization for no_varlen embedding layernorm (#48580) * general optimization no_varlen embedding layernorm * fix tmp directories (#48863) * rm dygraph_to_static eager guard tests part2 minst2ptb_lm (#48793) * rm dygraph_to_static eager guard tests part2 minst2ptb_lm * merge gpugraph to develop, fix the_one_ps.py for gpups * [remove fluid] under unittesets of linear api (#48564) * [remove fluid] under unittesets of linear api * [remove fluid] under unittesets of linear api * [remove fluid] under unittesets of linear api * [remove fluid] under unittesets of linear api * [remove fluid] under unittesets of linear api * [remove fluid] under unittesets of linear api * [remove fluid] fluid dygrapn linear api * [remove fluid] fluid dygrapn linear api * [remove fluid] fluid dygrapn linear api * [remove fluid.layers.cross_entropy] remove unit tests (part 1) (#48726) * replace layers.cross_entropy with paddle.entropy * fix args * fix codestyle * proper fix (#48360) Reenabled ext_reorder recording for TransDataLayoutFromOneDNN * [remove fluid.layers.matmul] remove fluid.layers.matmul in example code (#48818) * replace fluid.layers.matmul in fluid/io.py * fix doc error in fluid.layers.nn.sampling_id * remove test_auto_search_dist_matmul_op.py (#48794) * delete mean api (#48764) * clean test_op_name_conflict (#48704) * opt kernel_selection error msg (#48864) * rewrite delete_weight_dequant_linear_op_encoder/decoder pass (#48650) * rewrite delete_weight_deqquant_linear_op_encoder/decoder pass * [XPU] add set_value and set_value_grad (#48845) * merge gpugraph to develop, fix gpups ut * Add QuantizedMatmul in QAT (#47997) * fix 'BlasAXPBY unimplemented' error with custom device (#48762) * fix 'BlasAXPBY unimplemented' error with custom device * fix utils CmakeLists bug * first commit (#38143) * [Auto Parallel] Add cluster partition and dm to pm (#48320) * add cluster_partition and device_meshes to process_meshes funcs * add unitest * fix paddle2cinn float16 type support bug (#48249) * remove pool2d from fluid (#48512) * remove pool2d * [fluid remove]: remove paddle.fluid.layers.detection_map, paddle.fluid.metrics.DetectionMAP and paddle.fluid.evaluator.DetectionMAP (#48674) * remove paddle.fluid.layers.nn.temporal_shift * code check * rm unittest * remove paddle.fluid.layers.detection_map and the class:DetectionMAP * [PHI decoupling] move "flags.h" from fluid to phi (#48696) * add set_lr & get_lr for stage2 optimizer. (#48857) * move share_buffer kernel to phi (#48858) * move share_buffer kernel to phi * fix ut * add source file * fix window links * [Kernel Selection] Simplify kernel selection process in phi, reduce search number to half (#47771) * simplify SelectKernelOrThrowError function in phi * opt kernel_selection process * polish code, fix backend error * Support static graph code-gen for scalar and int_array (#48792) * add suppport_tensor for code_gen to static graph * support code-gen for int_array * polish code * fix bug of data_type * clean unittest test_model_cast_to_bf16 (#48705) * rm dy2static eager tests part1 bert2loop (#48790) * rm dygraph_to_static eager guard tests part3 reinforce2yolo (#48795) * rm distribution uniform eager guard test (#48768) * rm distribution uniform eager guard test * review * replace cross_entropy in python/paddle/fluid/tests/unittests/test_[a-n]*.py except test_dist_transpiler.py (#48913) * replace cross_entropy except in python/paddle/fluid/tests/unittests/*.py && unittests/*/*.py (#48922) * [Paddle Inference]add cutlass act set in conv_elementwise_add_act_fuse_pass (#48838) * add cutlass act set in conv_elementwise_add_act_fuse_pass * move fluid.layers.create_global_var to static.create_global_var (#48777) * Modified the Kernel policy. When the compute is NHWC (#48563) * temporally disable set_value (#48942) * xpu support inplace flatten (#48909) This is a PR to catch up with latest xpu white list strategy (#48606) , since original list only include 'fluid' fashion names, but new list must include 'phi' fashion as well. Refer to paddle/phi/core/kernel_factory.cc for more details. * fix:vit_attention ut (#48884) * mv fused_bias_dropout_residual_ln to fluid manual dir (#48824) * mv fused_bias_dropout_residual_ln to fluid manual dir * rm useless comments * bug fix (#48829) * move ops_extra_info_gen.py from phi to fluid (#48926) * fix scale type in alpha and beta (#48887) * [inference][trt] upgrade prelu op (#48528) * add prelu * 对多个文档按照要求修改对应中文的#5453 (#48886) * fix doc * test=document_fix Co-authored-by: Ligoml <39876205+Ligoml@users.noreply.github.com> * replace cross_entropy in python/paddle/fluid/tests/unittests/*.py except test*.py (#48919) * [remove fluid] Remove fluid APIs (#48641) * [CodeStyle] fix renamed files not being monitored by Codestyle Check (#48892) * [fluid remove]: remove paddle.fluid.layers.box_coder and paddle.fluid.layers.polygon_box_transform (#48896) * remove fluid_box_coder and polygon_box_transform * code check * [Custom XPU Support] Custom extension support xpu backend (#48733) * support custom_xpu * update cmake to test xpu * support custom_xpu, verify mechanism * fix test_custom_relu_op_xpu_setup.py, test=kunlun * fix FLAGS_init_allocated_mem * cancel TIMEOUT property * reset FLAGS_init_allocated_mem property * rm mlu ops eager guard tests (#48769) * rm npu instance_np op for eager guard tests (#48785) * remove xpu eager guard tests (#48786) * [remove fluid.layers.cross_entropy] remove unit tests (part 3) (#48918) * replace cross_entropy in python/paddle/fluid/tests/unittests/test_[o-z]*.py plus test_dist_transpiler.py * fix test_prune * [Inference] optimize some code and fix some bug (#48780) * clean ir_pass_manager and fix map_depthwise_conv_to_conv_pass * fix unitest timeout * [PHI] Migrate reshape kernel (#48749) * reshape * typo * remove header * support py3 in setup.py (#48905) * support py3 in setup.py * support setup.py bdist_wheel in py3 * support py3 in setup.py * modify run_setup * [Paddle-TRT] add cast between int64 tensor and Paddle-TRT (#45547) * Add cast between int64 tensor and Paddle-TRT * Add Unit testing. * fix sharding_stage1 amp O2 decorate bug (#48960) * [remove fluid] fluid dygraph Embedding (#48806) * [remove fluid] fluid dygraph Embedding * [remove fluid] fluid dygraph Embedding * [remove fluid] fluid dygraph Embedding * [remove fluid] fluid dygraph Embedding * [remove fluid] fluid dygraph Embedding * [remove fluid] fluid dygraph Embedding * fix for mkldnn (#48852) * H2D data transfer optimization with usage of structure type for stack kernel (#48899) * first commit. * refine performance with fast_divmod * refine performance with fast_divmod * rm accuracy and auc in extra __all__ (#48986) * Add dynamic checks for collective communication on NCCL (#48915) * chore: unify `SingleTensor` * feat: dynamic check * support sharding in fp16 on xpu, (#48897) * support sharding in fp16 on xpu, change reduce_max to reduce_sum for found nan or inf * update * Support cross-step stream synchronization for standalone executor (#48809) * Add UT * Support cross-step stream synchronization for standalone executor * Fix typos * Fix typos * Update UTs * Generate static graph code of some ops by yaml (#48771) * generate static graph code of some ops by yaml, test = develop * fix 'take_along_axis' yaml style * reset scatter/scatter_nd_add * delete the comments of put_along_axis * fix a bug in GetTrtWeight (#48993) * add static_ops.yaml for static op (#48991) * [PHI decoupling] move norm_utils.cu.h from fluid to phi and remove norm_utils.h in fluid (#48930) * move norm_utils.cu.h from fluid to phi * remove norm_utils.h in fluid * fix bugs and replace mutable_data with Alloc * replace mutable_data with Alloc * forbid conv op whose weight is not a persistable weight into Paddle-TRT (#48763) * fix: Move the pass location to the appropriate location (#48951) * Enhance check_nan_inf implementation for CPU. (#48591) * Enable to print device info. * Enhance the nan and inf checking for cpu. * Implement a common print function. * Unify the check of complex numbers. * Rewrite the omp method. * Count and print the number of nan and inf. * Change the print content. * Add unittest. * [PHI] OneDNN version of Copy (#48539) * OneDNN version of Copy, tranpose kernels adjusted * style fixes in tranpose_grad * redundant headers deleted * fix: there are some bugs with trt 8.0 (#48921) * fix: there are some bugs with trt 8.0 * fix:windows CI trt is too old * Optimization of Eigh op with ssyevj_batched runtime api (#48560) * fix codestyle * add double complex<float> complex<double> dtype support for syevj_batched * fix use_syevj flag for precision loss when input dtype of syevj_batch is complex128 in some case * optimize eigh in different case * fix missing ; bug * fix use_syevj bug * fix use_cusolver_syevj_batched flag * replace cross_entropy in python/paddle/fluid/tests/unittests/*/*.py except unittests/*.py (#48920) * [PHI decoupling] replace dependency of inclusive_scan.h from phi (#48980) * replace dependency of inclusive_scan.h from phi * format code * fluid API magration : Assert, increment, cond (#48885) * [Clean fluid] Add inner function _elementwise_op_with_axis (#48748) * add inner function _elementwise_op_with_axis * fix transformer_model * polish API code * remove elementwise_div/mul api * delete API in __all__ * delete elementwise_mul completely * polish elementwise_mul call * polish internal api * resolve conflict, fix rnn.py * use non-inplace call * delete elementwise_mul api test * delete elementwise_mul api test * clean elementwise_add/sub * restore _elementwise_op_in_dygraph in nn.py * test_convert_to_mixed_precision.py use tempfile for temporary models/params (#48819) * Tighten the Interception strategy (#48947) * test approve ,test=document_fix * test approve ,test=document_fix * test approve ,test=document_fix * [CodeStyle][isort][F401] fix some regression issues (#48936) * [CodeStyle][isort][F401] fix some regression issues * add import paddle to fix eval call * rm multinode eager guard tests (#48766) * rm multinode eager guard tests * remove unwanted tests * reset process_mpi test * rm unittests eager guard tests part5 dataloader2dygraph_mnist (#48816) * [PHI]Add new Tensor type and migrate save_combine kernel (#47856) * add new tensor * fix windows compile bugs * fix ci bugs * fix ci bugs * fix ci bugs * perfect according comment * fix ci compile bugs * add raw tensor * fix ci bugs * modify code by comment * delete String * [Fluid Clean]move BatchNorm from flud.dygraph.nn to paddle.nn.layer.norm (#48734) * move BatchNorm from flud.dygraph.nn to paddle.nn.layer.norm * modfiy conflict * modify pre-commit error * modify static-check ci error * fix failed tests * modify conflict * modify conflict * delete import modelu GRUUnit * fix falied test * fix failed testes * fix failed tests * fix failed tests * fix failed test * fix error in test_fused_resenet_basic_block_op_xpu.py * modify after xiaoguang reviewed * [Setup] Ignore @PADDLE_BINARY_DIR@ files (#49002) * [Setup] Ignore @PADDLE_BINARY_DIR@ files * test=document_fix * reshape onednn test reimplemented (#48850) * - UT reshape onednn - Fix test test2 - test4 - test5 - test6 test7 - test8 - Ut reinvented - cosmetic * - fix * - fix * - fix * - fix * - Fix * - fix * - fix * - fix * - Fix * lint * update fused_multi_transformer_encoder_pass support GPT new matmul API (#48953) * fit paddle.matmul in fleetx.gpt * Revert "set free_when_no_cache_hit default value to true (#48815)" (#48968) This reverts commit 592ed40. * [Paddle Inference]fix some transformer unitest (#48929) * fix some transformer unitest * Enable Generic-Plugin support FP16 (#48807) * support conv1d quant & skip calibrate zero-size tensor (#48912) * enable custom device save model on device memory && fix conflict (#48221) * [api move] cvm (#48989) * [api move] cvm * [api move] cvm * [api move] cvm * [api move] cvm * [api move] cvm * [api move] cvm * [api move] cvm * [api move] ci test * [api move] ci test * [api move] ci test * Bugfix: xpu now only support single node multi-card, bkcl_comm_num should always set to 1 (#48961) * rm unittests eager guard tests part23 where2zeros (#48895) * rm unittests eager guard tests part17 number2pool1d (#48840) * [NPU] fix FLAGS_npu_storage_format flag in python, test=develop (#48976) * remove fleet eager guard tests (#48765) * rm unittests eager guard tests part6 eager_run2expand_v2 (#48817) * rm unittests eager guard tests part12 imperative_optimizer2resnet (#48833) * [fluid clean] remove 4 fluid.layers api and imigrate 2 fluid.layer api (#48972) * fluid clean layer * docs * remove reset reference in unittest for `fluid.layers.cross_entropy` (#49012) * replace cross_entropy in test*.py except python/paddle/fluid/tests/unittests/*.py (#48978) * remove linear_chain_crf and crf_decoding from fluid (#48996) * remove linear_chain_crf and crf_decoding * Generate static graph code of some ops by yaml (#48977) * generate static graph code of some ops by yaml * fix the code-style of yaml * fix the framework_ci for triangular_solve * change the 'data_type' of scatter * add the 'out: Out' of scatter_nd_add * [tools] Update summary env (#48627) * [tools] remove deprecated api , fix macOS get version error * [tools] Rename the value that returns null * [tools] add gcc, clang, cmak, libc version * [tools] fix cudnn read error * [tools] add gpu devices list, drive based * [issue] update 3_build-installation-issue.yml * [tools] fix get gpu list AttributeError * [Dy2St] transforms.RandomVerticalFlip Support static mode (#49024) * add static RandomVerticalFlip * object => unittest.TestCase * Save fused_attention op memory when dropout_rate = 0.0 (#48902) * save fused_attention memory when dropout_rate = 0.0 * add ut * fix ut bug * fix fused_layernorm_residual_dropout_bias_test.cu * Correct multiple inputs and outputs (#48872) * [CodeStyle][isort][Dy2St] sort imports for paddle.jit (#48637) * isort jit * refine comment * remove non-public apis from __all__ (#48952) * remove non-public apis from __all__ * fix code style * fix rmsprop_ yaml bug (#49026) * fix rmsprop_ yaml bug * Fixed the dead link bug in the API documentation (#48969) * first pr * Revise nn.py * Revise nn.py 2.0 * Revise rnn.py;test=document_fix * test=document_fix Co-authored-by: Ligoml <39876205+Ligoml@users.noreply.github.com> * Change mutable_data to ctx.Alloc. (#49001) * [inference][trt] add more unary op and square (#48534) * add more unary op and square * Support ninja (#48932) * move inplace_apis_indygraph_only from paddle.flud.dygraph.inplace_utils to paddle.utils * modify conflict * modify conflict * modify conflict * modify conflict * modify conflict * modify conflict * modify conflict * modify static-check ci error * fix conflict * modify failed tests * fix conflict * fix conflict * fix pool2d examples * modify conflict * fix failed tests * fix conflict * fix failed tests * modfiy problem of deleting pool2d * support Ninja in setup.py * support different cmake_generators * modify after reviewed * delete unused denotes * Deleted mkldnn_inplace_pass code (#47818) * Deleted mkldnn_inplace_pass code * Fixed error with cmake * Resolve conflicts * hide log (#49045) * test=doucment_fix * test=document_fix * [Sparse]Optimize performance of sparse conv on T4 (#49009) * modify cmake file for cuda11.8 compile (#49020) * modify cmake file for cuda11.8 compile * add op_library(fused_embedding_eltwise_layernorm_op DEPS bert_encoder_functor) * remove dropout from fluid (#48319) * remove dropout * nullptr bugfix for XPU pg mode (#49043) * nullptr bugfix for XPU pg mode Also a few kernels is added to xpu whitelist * increase error msg length * Divide elementwise case from BroadcastKernel and refine transpose autotune (#33051) * First Commit. * add some codes * add elementwise loader * fix code styles * merge with develop * add some changes both in elementwise and transpose * add init operation in broadcast kernel. * change codes according to pr suggestions about transpose file * fix error for op-benchmark ci * fix according to ci * add condition of skipif (#48791) * add condition of skipif * fix code format error * Update test_fused_gate_attention_op.py update * rm unittests eager guard tests part9 histogram2imperative_dataloader (#48825) * rm unittests eager guard tests part9 histogram2imperative_dataloader * rm basic * rm unittests eager guard test part14 initializer2layer_norm (#48835) * rm unittests eager guard test part14 initializer2layer_norm * monior change * [Bugfix] recompute dep filter param (#49010) * recompute dep filter param * recompute dep for reshard * [Paddle Inference] rewrite convert_to_mixed_precision (#48853) * [CodeStyle] fix c++17-extensions warning on macos (#49017) * fix c++17-extensions warning on macos * fix type fix c++17-extensions warning on macos fix c++17-extensions warning on macos * Add custom CUDNN finding paths for 64bit Windows (#49066) * remove prior_box (#49006) * remove prior_box * modify the sequence of paras of prior_box in multi_box_head api * InstanceNorm1D、InstanceNorm2D、InstanceNorm3D (#48940) * modified: python/paddle/nn/layer/norm.py * modified: python/paddle/nn/layer/norm.py * modified: python/paddle/nn/layer/norm.py * modified: python/paddle/nn/layer/norm.py * modified: python/paddle/nn/layer/norm.py * modified: python/paddle/nn/layer/norm.py * test=docs_preview * InstanceNorm2D中文档格式修改 * test=docs_preview * modified: python/paddle/nn/functional/loss.py modified: python/paddle/nn/functional/norm.py modified: python/paddle/nn/layer/loss.py modified: python/paddle/nn/layer/norm.py * test=docs_preview * test=docs_preview * [AutoParallel] recompute tuning (#48608) * [AutoParallel] recompute tuning * fix conflict * update comment * bug fix * update rc algo * tiny fix * fix clear process_group * remove comment * update segment print * fix import OpRole * adapt amp pass and grad_clip pass for opt_tuner * update tuning config * fix import * annotate recompute info on ops and upgrade recompute pass * add op_namescope for seed op * record reserved vars * fix recompute var's dist_attr * fix strategy unittest * adapt for fp16 * update unittest * revert copy opt * update unittest * rename set_recompute_segments * fix unittest * fluid API magration : array_read, array_write (#49022) * del array_write & array_read * fix import err * fix import err * fix example codes * Keep double-buffer reader for static mode (#49068) * Fix nullptr to TestFuseGemmEpilogueReluBWDFP* (#48997) * support fp16 index sample (#47897) * add index sample fp16 support * remove fluid APIs in distributed_strategy.py and role_maker.py * Revert "remove fluid APIs in distributed_strategy.py and role_maker.py" This reverts commit 223bbee. * fix instantiated more than once * clean codes * rm unittest eager guard tests part20 sparse_mv2split (#48879) * rm unittests eager guard tests part11 imperative_layer2ocr (#48828) * rm unittests eager guard tests part11 imperative_layer2ocr * review * rm eager guard tests part3_1 (#49059) * fix: gloo compatible (#49084) * rm eager guard tests part3_3 (#49061) * fix bug (#49081) * [Inference] memory_optimize and mkdlnn problem (#49054) * memory_optimize and mkdlnn problem * update * update * update * Remove/move 16 fluid APIs (#48377) * remove density_prior_box * remove anchor_generator * remove roi_perspective_transform * remove generate_proposal_labels * remove generate_mask_labels * remove generate_proposals * remove box_clip * remove retinanet_detection_output * remove multiclass_nms * remove locality_aware_nms * remove matrix_nms * remove distribute_fpn_proposals * remove box_decoder_and_assign * remove collect_fpn_proposals * remove 2 trt files * move prior_box to static/nn/common.py * move multi_box_head to static/nn/common.py * fix for CI/CE * remove retinanet_detection_output * restore compile_vs_runtime_white_list.py * restore test_retinanet_detection_output to white list * replace nn.flatten by paddle.flatten, and fix doc for retinanet_target_assign * add enable_static in demo and fix bug * remove roi_perspective_transform in test_layers * remove multi_box_head * change self.multiclass_nms to _legacy_C_ops.multiclass_nms * empty commit * empty commit * check code style * fix prior_box * fix CI * remove redundant prior_box in detection.py * fix docs * remove detection * fix prior_box en doc * delete prior_box in common * remote proir_box from __init__.py * fix embedding multihead (#49085) * SetDeviceId in StreamSafeCUDAAllocation (#49080) * [PHI decoupling] Remove fluid imports from MKLDNN code (#48981) * fix wrong handler name * mkldnn_engine -> onednn_engine * remove fluid/errors.h imports * remove fluid/enforce.h imports * remove note and unnecessary import * remove fluid/pretty_log.h imports * remove fluid/place.h imports * remove fluid/data_layout_transform.h imports * remove fluid/device_context.h imports * remove mkldnn_helper code * remove fluid/mkldnn_reuse.h imports * pretty_log import * replace cross_entropy in python/paddle/fluid/tests/unittests/*.py (#48975) * 修复paddle.amp.decorate等API的文档 (#48983) * 涉及到的api有 paddle.amp.decorate paddle.static.npu_places paddle.signal.istft paddle.signal.stft paddle.linalg.eigvalsh paddle.randint_like * change signal.stft * randint_like的low增加optional * ; test=docs_preview * 修改了注解格式; test=docs_preview * 修改了公式格式 * 修改了decorate的models等 * test=document_fix Co-authored-by: Ligoml <39876205+Ligoml@users.noreply.github.com> * 按在线文档需求 61~70 更新了部分文档 (#49014) * Update docstring: 1. 去除 python/paddle/tensor/manipulation.py 中 cast 函数描述中的 This OP； 2. 调整 python/paddle/fluid/layers/control_flow.py 中 Print 函数中参数描述的顺序，添加 optional 描述； 3. 为 python/paddle/tensor/logic.py 中 logical_and 函数添加 optional 描述； 4. 为 python/paddle/fluid/reader.py 中 DataLoader 类中 from_generator、from_dataset 函数添加 optional 描述； 5. 在 python/paddle/fluid/layers/nn.py 中 crf_decoding 函数的 param_attr 在使用中确实可视为存在默认值 None，故添加 optional 描述； 6. 修复 python/paddle/static/nn/common.py 中 data_norm 函数描述里 tex 语法错误的问题，并一并修复同一文件中的相同问题。 * 根据 review 意见修改部分内容。 * 将谓语动词去掉第三人称单数形式。 * 同步中文文档变更。 * string-->str; test=document_fix Co-authored-by: Ligoml <39876205+Ligoml@users.noreply.github.com> * merge gpugraph to develop, fix gloo wrapper * merge gpugraph to develop, fix ci * merge gpugraph to develop, fix gloo wrapper * merge gpugraph to develop, fix ci * merge gpugraph to develop, fix fleet.py * merge gpugraph to develop, fix merge error * merge gpugraph to develop, fix merge error * merge gpugraph to develop, add python ut * merge gpugraph to develop, fix code style * merge gpugraph to develop, add c++ ut * merge gpugraph to develop, fix code style * merge gpugraph to develop, fix data_feed.h * merge gpugraph to develop, fix code style * merge gpugraph to develop, fix code style * merge gpugraph to develop, fix code style * merge gpugraph to develop, fix code style Co-authored-by: wuhuachaocoding <77733235+wuhuachaocoding@users.noreply.github.com> Co-authored-by: Nyakku Shigure <sigure.qaq@gmail.com> Co-authored-by: zyfncg <zhangyunfei07@baidu.com> Co-authored-by: xiongkun <xiongkun03@baidu.com> Co-authored-by: ceci3 <ceci3@users.noreply.github.com> Co-authored-by: zhangyikun02 <48021248+zhangyk0314@users.noreply.github.com> Co-authored-by: Weilong Wu <veyron_wu@163.com> Co-authored-by: 201716010711 <87008376+201716010711@users.noreply.github.com> Co-authored-by: Qi Li <qili93@qq.com> Co-authored-by: zhoutianzi666 <39978853+zhoutianzi666@users.noreply.github.com> Co-authored-by: feng_shuai <fengshuai03@baidu.com> Co-authored-by: QingshuChen <chenqingshu@baidu.com> Co-authored-by: WangZhen <23097963+0x45f@users.noreply.github.com> Co-authored-by: 张春乔 <83450930+Liyulingyue@users.noreply.github.com> Co-authored-by: 傅剑寒 <Xs1580802568@gmail.com> Co-authored-by: xiaoguoguo626807 <100397923+xiaoguoguo626807@users.noreply.github.com> Co-authored-by: wangzhen38 <41941775+wangzhen38@users.noreply.github.com> Co-authored-by: Zhou Wei <1183042833@qq.com> Co-authored-by: Kevin吴嘉文 <417333277@qq.com> Co-authored-by: Zman <35071129+Atlantisming@users.noreply.github.com> Co-authored-by: lzy <569782149@qq.com> Co-authored-by: Vvsmile <450864116@qq.com> Co-authored-by: Sławomir Siwek <slawomir.siwek@intel.com> Co-authored-by: RichardWooSJTU <37864677+RichardWooSJTU@users.noreply.github.com> Co-authored-by: Ligoml <39876205+Ligoml@users.noreply.github.com> Co-authored-by: hjyp <53164956+Tomoko-hjf@users.noreply.github.com> Co-authored-by: Vigi Zhang <VigiZhang@users.noreply.github.com> Co-authored-by: zqw_1997 <118182234+zhengqiwen1997@users.noreply.github.com> Co-authored-by: Yiqun Liu <Xreki@users.noreply.github.com> Co-authored-by: risemeup1 <62429225+risemeup1@users.noreply.github.com> Co-authored-by: Guanghua Yu <742925032@qq.com> Co-authored-by: 姜永久 <34344716+yjjiang11@users.noreply.github.com> Co-authored-by: wanghuancoder <wanghuan29@baidu.com> Co-authored-by: Roc <30228238+sljlp@users.noreply.github.com> Co-authored-by: Wilber <jiweibo@baidu.com> Co-authored-by: Ghost Screaming <mofengshenjieII@163.com> Co-authored-by: Netpunk <69072522+Patrick-Star125@users.noreply.github.com> Co-authored-by: 六个骨头 <46243324+zrr1999@users.noreply.github.com> Co-authored-by: Aurelius84 <zhangliujie@baidu.com> Co-authored-by: Ruibiao Chen <chenruibiao@baidu.com> Co-authored-by: liu zhengxi <380185688@qq.com> Co-authored-by: heyanru <81976792+heyanru01@users.noreply.github.com> Co-authored-by: tianshuo78520a <707759223@qq.com> Co-authored-by: huangjiyi <43315610+huangjiyi@users.noreply.github.com> Co-authored-by: Infinity_lee <luhputu0815@gmail.com> Co-authored-by: houj04 <35131887+houj04@users.noreply.github.com> Co-authored-by: lugimzzz <63761690+lugimzzz@users.noreply.github.com> Co-authored-by: Wangzheee <634486483@qq.com> Co-authored-by: sneaxiy <32832641+sneaxiy@users.noreply.github.com> Co-authored-by: kangguangli <kangguangli@hotmail.com> Co-authored-by: jakpiase <jakpia21@gmail.com> Co-authored-by: HongyuJia <jiahongyu@baidu.com> Co-authored-by: haosicheng <47998305+HarperCy@users.noreply.github.com> Co-authored-by: Chang Xu <molixu7@gmail.com> Co-authored-by: Kai Song <50285351+USTCKAY@users.noreply.github.com> Co-authored-by: limingshu <61349199+JamesLim-sy@users.noreply.github.com> Co-authored-by: Jianghai <72591262+CjhHa1@users.noreply.github.com> Co-authored-by: jiangcheng <thisjiang@qq.com> Co-authored-by: ccrrong <101700995+ccrrong@users.noreply.github.com> Co-authored-by: PuQing <me@puqing.work> Co-authored-by: Leo Chen <chenqiuliang@baidu.com> Co-authored-by: cyber-pioneer <116002591+cyber-pioneer@users.noreply.github.com> Co-authored-by: niuliling123 <51102941+niuliling123@users.noreply.github.com> Co-authored-by: james <zhangxiaoci@baidu.com> Co-authored-by: wenbin <wang3323032@qq.com> Co-authored-by: ZZK <359521840@qq.com> Co-authored-by: Zhang Jun <ewalker@live.cn> Co-authored-by: yjphhw <43883055+yjphhw@users.noreply.github.com> Co-authored-by: Yuanle Liu <yuanlehome@163.com> Co-authored-by: Wen Sun <35923278+HermitSun@users.noreply.github.com> Co-authored-by: lzydev <1528794076@qq.com> Co-authored-by: Paulina Gacek <paulina.gacek@intel.com> Co-authored-by: feifei-111 <2364819892@qq.com> Co-authored-by: YuanRisheng <yuanrisheng@baidu.com> Co-authored-by: Jacek Czaja <jacek.czaja@intel.com> Co-authored-by: weishengying <63448337+weishengying@users.noreply.github.com> Co-authored-by: engineer1109 <jialiang.wang@xdxct.com> Co-authored-by: gouzil <66515297+gouzil@users.noreply.github.com> Co-authored-by: Ryan <44900829+DrRyanHuang@users.noreply.github.com> Co-authored-by: joanna.wozna.intel <joanna.wozna@intel.com> Co-authored-by: JYChen <zoooo0820@qq.com> Co-authored-by: jjyaoao <88936287+jjyaoao@users.noreply.github.com> Co-authored-by: Hulek <jakub.hulek@intel.com> Co-authored-by: zhangkaihuo <zhangkaihuo@baidu.com> Co-authored-by: YUNSHEN XIE <1084314248@qq.com> Co-authored-by: JZ-LIANG <jianzhongliang10@gmail.com> Co-authored-by: Tinson Lai <laitingsheng@hotmail.com> Co-authored-by: Ayuan <79981115+Ayuan2021@users.noreply.github.com> Co-authored-by: zhaoyingli <86812880+zhaoyinglia@users.noreply.github.com> Co-authored-by: Ming-Xu Huang <mingh@nvidia.com> Co-authored-by: wangxiaoning <71813629+wangxn12138@users.noreply.github.com> Co-authored-by: Haohongxiang <86215757+haohongxiang@users.noreply.github.com> Co-authored-by: HydrogenSulfate <490868991@qq.com> Co-authored-by: mjxs <52824616+kk-2000@users.noreply.github.com> Co-authored-by: 学渣戊 <x19403@163.com>

add new tensor

60f8c2a

YuanRisheng added 5 commits November 10, 2022 12:43

fix windows compile bugs

60e3717

Merge branch 'develop' of github.com:YuanRisheng/Paddle into perfect_…

ebbee9a

…registry

fix ci bugs

a87999d

fix ci bugs

9929a28

fix ci bugs

c4a66e6

jiahy0825 previously approved these changes Nov 28, 2022

View reviewed changes

zyfncg reviewed Nov 28, 2022

View reviewed changes

chenwhql reviewed Nov 28, 2022

View reviewed changes

YuanRisheng added 2 commits November 28, 2022 12:01

deal with conflict

ee11ac6

perfect according comment

01e976c

YuanRisheng dismissed jiahy0825’s stale review via 01e976c November 28, 2022 12:03

YuanRisheng added 3 commits November 29, 2022 13:41

deal with conflict

bec4990

fix ci compile bugs

4010a6f

add raw tensor

4ec7569

chenwhql reviewed Dec 2, 2022

View reviewed changes

fix ci bugs

c4cca54

chenwhql reviewed Dec 5, 2022

View reviewed changes

modify code by comment

9bbd512

chenwhql previously approved these changes Dec 6, 2022

View reviewed changes

jiahy0825 previously approved these changes Dec 6, 2022

View reviewed changes

zyfncg previously approved these changes Dec 6, 2022

View reviewed changes

YuanRisheng added 2 commits December 8, 2022 07:37

delete String

e552774

Merge branch 'develop' of github.com:YuanRisheng/Paddle into perfect_…

059e85f

…registry

YuanRisheng dismissed stale reviews from zyfncg, jiahy0825, and chenwhql via 059e85f December 8, 2022 07:38

chenwhql approved these changes Dec 9, 2022

View reviewed changes

zyfncg approved these changes Dec 9, 2022

View reviewed changes

jiahy0825 reviewed Dec 9, 2022

View reviewed changes

jiahy0825 approved these changes Dec 9, 2022

View reviewed changes

XiaoguangHu01 approved these changes Dec 12, 2022

View reviewed changes

YuanRisheng merged commit ecf892f into PaddlePaddle:develop Dec 12, 2022

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

[PHI]Add new Tensor type and migrate save_combine kernel #47856

[PHI]Add new Tensor type and migrate save_combine kernel #47856

YuanRisheng commented Nov 10, 2022 •

edited

Loading

paddle-bot bot commented Nov 10, 2022

zyfncg Nov 28, 2022

YuanRisheng Nov 28, 2022

zyfncg Nov 28, 2022

YuanRisheng Nov 28, 2022

zyfncg Nov 28, 2022

YuanRisheng Nov 28, 2022

chenwhql Nov 11, 2022

YuanRisheng Nov 28, 2022

chenwhql Nov 11, 2022

YuanRisheng Nov 28, 2022

chenwhql Nov 28, 2022

YuanRisheng Nov 28, 2022

chenwhql Nov 28, 2022

YuanRisheng Nov 28, 2022 •

edited

Loading

chenwhql Nov 28, 2022

YuanRisheng Nov 28, 2022

chenwhql Dec 2, 2022

YuanRisheng Dec 2, 2022

chenwhql Dec 5, 2022

YuanRisheng Dec 5, 2022

chenwhql Dec 5, 2022

YuanRisheng Dec 5, 2022

chenwhql Dec 5, 2022

YuanRisheng Dec 5, 2022

chenwhql Dec 5, 2022

YuanRisheng Dec 5, 2022

chenwhql Dec 6, 2022

YuanRisheng Dec 6, 2022

YuanRisheng Dec 8, 2022

jiahy0825 Dec 9, 2022

YuanRisheng Dec 9, 2022

XiaoguangHu01 left a comment


		std::string::iterator begin() { return data_.begin(); }

		std::string* Get() { return &data_; }

[PHI]Add new Tensor type and migrate save_combine kernel #47856

[PHI]Add new Tensor type and migrate save_combine kernel #47856

Conversation

YuanRisheng commented Nov 10, 2022 • edited Loading

PR types

PR changes

Describe

paddle-bot bot commented Nov 10, 2022

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

YuanRisheng Nov 28, 2022 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

XiaoguangHu01 left a comment

Choose a reason for hiding this comment

YuanRisheng commented Nov 10, 2022 •

edited

Loading

YuanRisheng Nov 28, 2022 •

edited

Loading