28 Pull requests merged by 21 people

• [CINN] Accelerate recompute ips

  #70386 merged Dec 23, 2024

• fix compile [[maybe_unused]]

  #70384 merged Dec 23, 2024

• 【GPUPS】add env for gpups and fix cache table

  #70301 merged Dec 23, 2024

• [CINN]support backend PassManager run on block

  #70361 merged Dec 23, 2024

• Dis dcu paddlex

  #70401 merged Dec 23, 2024

• [CINN] fix slice op sym infer with input(with data) and symbol starts or ends

  #70282 merged Dec 23, 2024

• [CINN] skip CacheGradOpSymbolicShape check for fused_attention

  #70294 merged Dec 23, 2024

• [CI]update scripts of Auto-Parallel CI

  #70041 merged Dec 23, 2024

• [Cherr-pick] Fix take_along_axis_grad and p_norm_decomp

  #70340 merged Dec 23, 2024

• [CINN] Improve subgraph topo index update algorithm

  #70380 merged Dec 23, 2024

• [Prim] Add index_add_double_grad in dygraph composite

  #70367 merged Dec 23, 2024

• [Inference]Support plugin in python converter

  #70251 merged Dec 23, 2024

• [CINN][Add Backend Pass Comment No.12] Add comment for TransformGpuForloop

  #70289 merged Dec 23, 2024

• [CINN][Backend Pass Update No.9] Update RemoveScheduleBlock pass

  #70334 merged Dec 23, 2024

• [CodeStyle][Typos][M-[1-22]] Fix typo(macor, mantain, mak, manualy, mannualy, maping, maxmium, Maxinum, menas, momory, Meatadata,miliseconds, Minium, minumum, minimun, modication, modifed, modifing, modifty, moduels, Modul, Morevoer, mutiple, Mutiply, multipy, Multiplie, mutiply, Muti, muti, mutexs)

  #70392 merged Dec 23, 2024

• [SCU][Docathon][Add API Legend No.7] Add legend for expand API

  #70143 merged Dec 23, 2024

• [SCU][Docathon][Add API Legend No.44] Add legend for repeat_interleave API

  #70128 merged Dec 23, 2024

• [Inference] FIX OpenVINO COMPILE BUG

  #70372 merged Dec 23, 2024

• [Kernel] Fix IndexAddGradKernel

  #70353 merged Dec 23, 2024

• [CINN]fix DimExpr sort

  #70336 merged Dec 23, 2024

• 【api】support “dcu” for set_device and device_guard api

  #70308 merged Dec 23, 2024

• [SOT] Visualize subgraph relation info by info collector

  #70387 merged Dec 21, 2024

• [CodeStyle][Typos][K-[2-4]] Fix typo (keyward, Keyward)

  #70356 merged Dec 21, 2024

• Disable WITH_OPENVINO

  #70365 merged Dec 20, 2024

• [Fix] Reduce unnecessary GpuMemGetInfo in Tensor.to

  #70359 merged Dec 20, 2024

• [AutoParallel] fix shard stage2 memory error

  #70363 merged Dec 20, 2024

• [CINN]fix fused gemm grad infer meta bug

  #70256 merged Dec 20, 2024

• [CINN] Disable transpose for TileBroadcastTactic

  #70350 merged Dec 20, 2024

22 Pull requests opened by 18 people

• 【cherry-pick】Fixed pir fitness when exporting model and comm init when model inferring

  #70374 opened Dec 20, 2024

• support pylayer for moe

  #70375 opened Dec 20, 2024

• Expr div

  #70376 opened Dec 20, 2024

• [PIR] Add state_dict to program_patch to enable default paramater

  #70377 opened Dec 20, 2024

• add xpu local compilation

  #70378 opened Dec 20, 2024

• 【SCU】【Paddle Tensor 第二期 API 支持 0-size TensorNo.26、50】paddle.linalg.vecdot 支持 0-size Tensor

  #70379 opened Dec 20, 2024

• [CINN] Fallback FuionOp with single reshape to PHI kernel

  #70381 opened Dec 20, 2024

• optimize cinn compile tools

  #70382 opened Dec 20, 2024

• Add general function for PIR Custom Pass

  #70383 opened Dec 20, 2024

• CI测试不review[fluid_ops] c_allreduce_max 2

  #70388 opened Dec 20, 2024

• Update `F.multi_label_soft_margin_loss/nn.MultiLabelSoftMarginLoss` docs 易用性提升

  #70389 opened Dec 21, 2024

• [Windows CI] Record the exact size of dll and whl files and save them to disk D

  #70390 opened Dec 21, 2024

• Transform to trt program

  #70394 opened Dec 22, 2024

• [CI] Add clone pipeline in action

  #70395 opened Dec 23, 2024

• clear sym shape str print

  #70396 opened Dec 23, 2024

• [Warning] Refine `to_tensor` warning

  #70398 opened Dec 23, 2024

• cuda_error_check

  #70399 opened Dec 23, 2024

• [Paddle TensorRT] add pd_op.logical_not converter

  #70400 opened Dec 23, 2024

• [Prim] Optimize `where_double_grad`

  #70404 opened Dec 23, 2024

• [SOT] Support string slice

  #70405 opened Dec 23, 2024

• [CINN][Backend Pass Update No.7] Update merge_block_utils

  #70406 opened Dec 23, 2024

• [Cherr-pick] PR70288、PR69995、PR70325

  #70407 opened Dec 23, 2024

1 Issue closed by 1 person

• 电脑更新GPU后出现ImportError: DLL load failed while importing libpaddle: %1 不是有效的 Win32 应用程序。

  #70157 closed Dec 23, 2024

3 Issues opened by 3 people

• 安装报错

  #70393 opened Dec 22, 2024

• 调试代码错误

  #70391 opened Dec 21, 2024

• TransformerEncoderLayer中的activation激活参数需求。内容很简短烦请请大佬赐教。

  #70385 opened Dec 20, 2024

36 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• [PIR] translate old ir to pir_program and trt program

  #70364 commented on Dec 23, 2024 • 8 new comments

• [CINN]Add support for TensorList input in the Pad3d infer symbolic interface

  #70318 commented on Dec 23, 2024 • 6 new comments

• 【 Paddle Tensor 规范化二期 】解决隐式广播不能够处理0-size的问题

  #70369 commented on Dec 23, 2024 • 4 new comments

• [Docathon][Add API Legend No.55、22] add paddle.index_fill and paddle.shard_index API Paddle

  #70215 commented on Dec 20, 2024 • 3 new comments

• [SOT][Faster Guard] add `StringCompareGuard` and support `RangeVariable`

  #70362 commented on Dec 22, 2024 • 2 new comments

• [Docathon][Add API Legend No.40、51] add figures for unfold and unsqueeze API

  #70048 commented on Dec 20, 2024 • 2 new comments

• [CINN] Process 0D tensor in ReduceInferDim

  #70343 commented on Dec 23, 2024 • 2 new comments

• [SCU][Docathon][Add API Legend No.53] Add legend for masked_scatter API

  #70145 commented on Dec 23, 2024 • 1 new comment

• [SCU][Docathon][Add API Legend No.48]add illustration for as_strided API

  #69874 commented on Dec 22, 2024 • 1 new comment

• Strided_slice 官网文档中未描述stride 不能整除的情况，给出的示例也未说明出现这中情况时算子的行为

  #70203 commented on Dec 21, 2024 • 0 new comments

• [CINN] Add Div shape op

  #70224 commented on Dec 20, 2024 • 0 new comments

• [Docathon][Add API Legend No.16] Add legend and update doc for atleast_2d -part

  #70242 commented on Dec 23, 2024 • 0 new comments

• [CINN][optimize warning message] optimize `SetSymbolForValueByStaticShape` && `CheckInferSymWithInferMeta`

  #70326 commented on Dec 22, 2024 • 0 new comments

• [CINN] Fix no_need_buffer bug

  #70349 commented on Dec 22, 2024 • 0 new comments

• 【Comm】Comm init in dynamic mode

  #70352 commented on Dec 23, 2024 • 0 new comments

• CI测试不review[fluid_ops] allreduce_type

  #70358 commented on Dec 23, 2024 • 0 new comments

• fix

  #70360 commented on Dec 20, 2024 • 0 new comments

• 【custom】add loadCustomEnginelib method

  #70366 commented on Dec 23, 2024 • 0 new comments

• Fix DCU doc

  #70368 commented on Dec 23, 2024 • 0 new comments

• Directly use `paddle.full` to create complex tensor in `paddle.equal`.

  #70373 commented on Dec 23, 2024 • 0 new comments

• 3.0bata版本的lstm参数输出有误

  #70245 commented on Dec 21, 2024 • 0 new comments

• 安装报错。

  #70069 commented on Dec 23, 2024 • 0 new comments

• 【开源任务】CINN编译器后端Pass注释添加

  #70113 commented on Dec 23, 2024 • 0 new comments

• paddle.utils.run_check()报错

  #69096 commented on Dec 23, 2024 • 0 new comments

• [CodeStyle][Typos][A-N] ✏️ Typos 工具引入计划

  #69441 commented on Dec 23, 2024 • 0 new comments

• 模型剪枝后的速度问题

  #70339 commented on Dec 23, 2024 • 0 new comments

• 【开源任务】PIR-TensorRT converter推全升级

  #69178 commented on Dec 23, 2024 • 0 new comments

• 【开源任务】CINN编译器后端Pass改造

  #69639 commented on Dec 23, 2024 • 0 new comments

• [Paddle TensorRT No.42、45] Add (`pd_op.isnan`、`pd_op.embedding`) converter

  #69435 commented on Dec 23, 2024 • 0 new comments

• [CINN] apply vectorize Primitive in IRSchedule

  #69732 commented on Dec 20, 2024 • 0 new comments

• [CINN] open data check for sym infer

  #69891 commented on Dec 23, 2024 • 0 new comments

• [hackathon 7th] PaddleSpeech line: fix when shape=() can not load

  #69907 commented on Dec 23, 2024 • 0 new comments

• [PIR-Auto-Parallel]refactor refined recompute pass in PIR mode

  #70064 commented on Dec 23, 2024 • 0 new comments

• [CINN]optimize infer shape warning

  #70104 commented on Dec 23, 2024 • 0 new comments

• [SCU][Docathon][Add API Legend No.39]add the picture of unique_consecutive

  #70163 commented on Dec 23, 2024 • 0 new comments

• [AutoParallel] add while op test demo and fix has_elements bug

  #70207 commented on Dec 23, 2024 • 0 new comments