1. 28 10月, 2021 2 次提交
  2. 27 10月, 2021 8 次提交
  3. 26 10月, 2021 7 次提交
  4. 25 10月, 2021 6 次提交
    • A
      [NPU] modifications for model ernie-1.0 (#36642) · 19b02d95
      Aganlengzi 提交于
      * [NPU] modifications for model ernie-1.0
      
      * rollback 503003 and change cast to dtype
      19b02d95
    • Z
      add op: fused_feedforward(backward) (#35611) · 2dd0a46a
      zhangkaihuo 提交于
      这个PR是fused_feedforward反向的代码
      
      相关kernel实现:fused_dropout_act_bias, fused_residual_dropout_bias, fused_layernorm_residual_dropout_bias
      
      fused_feedforward是一个融合算子,该算子对transformer模型的feed forward层的算子进行融合和封装,使得前端只呈现一个接口,通过融合减少部分访存和kernel launch的时间,以此提升性能。
      2dd0a46a
    • S
      Add bincount op (#36317) · 39f19127
      smallv0221 提交于
      * Add bincount op
      
      * upload cpu version
      
      * fix unitest
      
      * fix unittest
      
      * fix unittest
      
      * fix en doc
      
      * add more test
      
      * fix en doc
      
      * add more test case
      
      * fix test
      
      * fix input vailidation
      
      * fix input check
      
      * fix unittest
      
      * fix test
      
      * fix en doc
      39f19127
    • T
      add some ops to train ssd on kunlun (#36407) · 50778ad6
      TTerror 提交于
      * add some ops to train ssd on kunlun
      
      * add some ops to train ssd on kunlun
      
      * add some ops to train ssd on kunlun
      
      * update cast op unittest
      
      * update cast op unittest
      
      * update cast op unittest
      
      * update xpu cmake
      
      * update cast unittest
      50778ad6
    • W
      Fix grid sampler while input size is [1] (#36183) · eff3ee5e
      whs 提交于
      eff3ee5e
    • Z
      add op: fused_feedforward(forward) (#35843) · b18cbfb2
      zhangkaihuo 提交于
      这个PR只包含fused_feedforward前向的代码。
      
      相关kernel实现:fused_dropout_act_bias, fused_residual_dropout_bias, fused_layernorm_residual_dropout_bias
      
      fused_feedforward是一个融合算子,该算子对transformer模型的feed forward层的算子进行融合和封装,使得前端只呈现一个接口,通过融合减少部分访存和kernel launch的时间,以此提升性能。
      b18cbfb2
  5. 23 10月, 2021 1 次提交
  6. 22 10月, 2021 3 次提交
    • Z
      add fp16 kernel for clip_op (#36577) · 1962d3af
      zhangbo9674 提交于
      1962d3af
    • L
      Fused attention op forward (#35905) · d4906214
      Li Min 提交于
      功能:本PR的目标是提高attention模块的计算性能。
      为了减少框架层对op的调度开销,本PR通过在C++层手动实现attention模块,对外提供attention 大op;
      为了减少防存开销,本PR采取了两种优化方法:
      (1)在q,k,v计算时通过共享输入X,将该处的gemm,transpose和bias add从三次调用减少为一次;
      (2)使用kernel融合优化技术,在不同cuda kernel之间通过寄存器传输数据;
      d4906214
    • W
      【Bug Fixes】Elementwise_add triple grad, fixed an input uninitialized problem (#36618) · 6580ad16
      Weilong Wu 提交于
      * Support elementwise_add triple grad Kernel
      
      * Change code-format to follow CI std
      
      * Removed unreasonable code, and fixed an input uninitialized issue
      
      * Support elementwise_add triple grad Kernel
      
      * Change code-format to follow CI std
      
      * Removed unreasonable code, and fixed an input uninitialized issue
      6580ad16
  7. 21 10月, 2021 7 次提交
    • Z
      [NPU] Add p_norm_grad (#36497) · ed478a3e
      zhulei 提交于
      ed478a3e
    • R
      add swish_op for npu (#36579) · 7eab0fa6
      ronnywang 提交于
      7eab0fa6
    • J
      Added matmul_v2+transpose+reshape fuse pass (#36481) · 856cb9c5
      jakpiase 提交于
      * added base changes for matmul_v2+trans+resh fuse pass
      
      * added full matmul_v2+transpose+reshape pass
      
      * removed a file added by mistake
      
      * added reviewers suggestions
      
      * Changed ops type in checking capatibility version
      
      * Deteled one statement
      856cb9c5
    • F
      [NPU] Add sync_batch_norm and sync_batch_norm_grad NPU Kernel (#36320) · 0ca2807c
      furnace 提交于
      * add sync_batch_norm (support train, infer, and fp32, fp16, and NCHW, NHWC)
      
      * [NPU] Delete debug codes
      
      * [NPU] Remove FP16
      0ca2807c
    • J
      Add viterbi decode (#35778) · 6072aecb
      Jack Zhou 提交于
      * add viterbi decode cpu kernel
      
      * add viterbi decoder api in paddle.text
      
      * add a data buffer once to avoid create many small pieces of data buffer frequently
      
      * fix viterbi max_seq_length bug
      
      * fix seq_len=1 bug
      
      * fix device context
      
      * move split out of for loop
      
      * remove INVERSE_SUB
      
      * remove 2 GET_CAST_MASK
      
      * remove 1 loop
      
      * remove Functor
      
      * add to_static deploy code
      
      * use MAX_FUNC instead of ELE_MAX
      
      * add MaxFunctor
      
      * impl max_func
      
      * remove MaxFunctor
      
      * remove cast op
      
      * use REGISTER_OP_WITHOUT_GRADIENT
      
      * add viterbi cuda kernel
      
      * add FIX_BLOCKDIM_CASE macro
      
      * add MKL add, mul; add get data mask
      
      * add arange mkl impl
      
      * add CPU Argmax
      
      * add cpu gather
      
      * use EXECUTE_MKL_ELEMENT_BINARY_OP instead of some ADD, MUL
      
      * use SameDimsBinaryOP instead of EXECUTE_MKL_ELEMENT_BINARY_OP
      
      * use SAME_DIMS_ELEMENT_BINARY_OP
      
      * add SimpleBroadcastBinaryOP
      
      * use int instead of int64_t to accelerate
      
      * optimize SimpleBroadcastBinaryOP
      
      * optimize SimpleBroadcastBinaryOP
      
      * optimize performance in both single thread and multithread situation
      
      * remove useless line
      
      * remove useless code
      
      * add CREATE_TENSOR_BUFFER macro
      
      * add INIT_REQUIRED_TENSOR macro
      
      * add comment
      
      * fix windows ci
      
      * add viterbi unittest
      
      * remove cuda add functor
      
      * remove cuda equal
      
      * remove a template function
      
      * fix windows ci
      
      * fix windows dtype
      
      * remove some template instance
      
      * remove useless header file
      
      * remove some blockdim
      
      * remove transpose impl
      
      * accelerate cpu performance on single thread situation
      
      * viterbi_decode->crf_decode
      
      * rename crf params name
      
      * add viterbi api test
      
      * remove useless import
      
      * add enable_static
      
      * use viterbi decoder
      
      * fix viterbi len=1
      
      * fix  viterbi unittest
      
      * remove useless comments
      
      * reconstruct viterbi decode
      
      * remove ADD,SUB,MUL structure
      
      * fix coverage
      
      * remove CREATE_TENSOR
      
      * add name args
      
      * crf.py->ops.py; with_start_stop_tag->include_start_end_tag
      
      * update crf_decode en docs
      
      * fix viterbi decode en docs
      
      * fix some review comments
      
      * add FIXED_BLOCK_DIM_CASE in cuda
      
      * push_back->emplace_back
      
      * crf_decode->viterbi_decode; include_start_end_tag->include_bos_eos_tag
      
      * paddle.text.ops.viterbi_decode->paddle.text.viterbi_decode
      
      * fix viterbi_decode en docs
      6072aecb
    • T
      add fill_any_like/flatten ops to train ssd on kunlun (#36550) · 7bf2aa38
      TTerror 提交于
      * add some ops to train ssd on kunlun
      
      * update test_fill_any_like_op_xpu.py
      7bf2aa38
    • N
      Fix a bug in ReadData, ReadDataBc and ReadDataReduce when NX != 1 (#36373) · 921c0917
      niuliling123 提交于
      * Update the implement of reduceAnyKernel according to kernel primitive api
      * Fix a bug in ReadData, ReadDataBc and ReadDataReduce when NX != 1
      921c0917
  8. 20 10月, 2021 5 次提交
  9. 19 10月, 2021 1 次提交