- 16 7月, 2021 7 次提交
-
-
由 levi131 提交于
As the title, this PR converts all blocks in program into SSA sub graphs and it is guarded by flag
-
由 Leo Chen 提交于
* add clear_float_status op * refine infershape * fix typo * refine check_finite_and_scale * refine code
-
由 Aurelius84 提交于
* Add NoNeedBufferVarsInferer * fix code style
-
由 Yuang Liu 提交于
-
由 Haohongxiang 提交于
-
由 Fan Zhang 提交于
-
由 WangXi 提交于
-
- 15 7月, 2021 11 次提交
-
-
由 danleifeng 提交于
-
由 ceci3 提交于
* fix bn * fix * add unittest * fix cpu
-
由 王明冬 提交于
-
由 Ren Wei (任卫) 提交于
* print the signatures of the apis * some faults, and a test case * use the FullArgSpec as the ArgSpec, repr>str 为了延续脚本历史，把FullArgsSpec替换为ArgSpec；在逆转回对象时，记得加上inpect.FullArgSpec对象 * script form @zhiboniu * parsearg * read argspec from file * add comments * refactor check_compatible and its testcases * skip the empty ArgSepc() * update the function, get inspect.FullArgSpec instances directly. * typo * for the functions using inner-defined VarType or other Class Type not accessable, use the string comparements. * typo * logging added * 若变量和默认值都增加了，用负数索引需要再处理下这个增加的内容 * defaults itself may be NoneType * defaults added 2 items when args only added 1 item * same as the previous comment * enable the check_api_compatible.py * run test_check_api_compatible.py * sort the output of print_signatures.py
-
由 Qi Li 提交于
-
由 huangxu96 提交于
This PR creates a class to process the program at the C++ level. Currently, this class has one class method: GetInputsOutputsInBlock()
-
由 Zhanlue Yang 提交于
* Add DCU backend support for custom ops * Added checks for DeviceCopy and renamed some macros
-
由 Qi Li 提交于
-
由 王明冬 提交于
[pass enhance] make the attribute check only object to which defined in op proto. test=develop (#34146)
-
由 wanghuancoder 提交于
* cache core.ops, test=develop * refine, test=develop
-
由 Aurelius84 提交于
* Refine Constructor logic of ParallelExecutor * Replace executor into ParallelExecutor in run_program_op
-
- 14 7月, 2021 17 次提交
-
-
由 Feiyu Chan 提交于
* exclude lr scheduler's state from accumulators_holder * fix when there is no learning rate scheduler * make a copy of the loaded state dict to avoid modifying it
-
由 ShenLiang 提交于
* add scatter-gather opt * fix topo for pp * rename function
-
由 Leo Chen 提交于
* adam add input SkipUpdate * add unittest * add npu unittest * fix xpu compile * remove param stream
-
由 ceci3 提交于
-
由 Wilber 提交于
-
由 kuizhiqing 提交于
* kill process by group
-
由 zhouweiwei2014 提交于 -
由 WeiXin 提交于
* jit.save/load support method with parameters. * add unittest and warning * polish warning message.
-
由 ShenLiang 提交于
-
由 WangXi 提交于
-
由 Huihuang Zheng 提交于
As the title, add Python helper function to get block Inputs/Outputs. This function is helpful to distributed computing for sharding.
-
由 tianshuo78520a 提交于
* Support Mac M1 make * cmake version check
-
由 arlesniak 提交于
* Added OpTestTool for BF16 convenience * fixes after review, names changed to snake case. * fixes after review, naming reflects cpu.
-
由 zhouweiwei2014 提交于
* Support sccache to speed up compilation on Windows * Support sccache to speed up compilation on Windows
-
由 Aurelius84 提交于
* fix paddle.no_grad * fix paddle.no_grad
-
由 Haohongxiang 提交于
-
由 Yuang Liu 提交于
-
- 13 7月, 2021 5 次提交