- 17 9月, 2020 5 次提交
- 16 9月, 2020 10 次提交
-
-
由 ShenLiang 提交于
* add adaptivelsgd * Todo fix the code to avoid the conflict.
-
由 Jack Zhou 提交于
Error description optimize for the math dir
-
由 Zhou Wei 提交于
fix judge cache file of inference api more accurate
-
由 Jacek Czaja 提交于
test=develop
-
由 ShenLiang 提交于
* fix error message
-
由 Chen Weihang 提交于
* polish framework error message part 7 * fix typo * polish by reviewes comment
-
由 wawltor 提交于
update the error message check for the some ops
-
由 wawltor 提交于
add the error message check for the some operator
-
由 Wilber 提交于
-
由 Leo Chen 提交于
-
- 15 9月, 2020 7 次提交
-
-
由 Shang Zhizhou 提交于
* optimize slice TRT plugin This patch removes unnecessary barrier for data transfer of needed offset, so data transfer can be overlap with GPU kernel execution. This patch also fixes incorrect name of slice plugin. That is, replaces "layernorm" with "slice" test=develop * add serialize/deserialize to slice plugin * add static shape slice trt plugin * fix slice trt op convertor dynamic shape bug * fix format by clang-format * fix pylint format error * fix problems commented by peiyang Co-authored-by: NRyan Jeng <rjeng@nvidia.com>
-
由 Wilber 提交于
-
由 石晓伟 提交于
-
由 Chen Weihang 提交于
* polish framework error msg part 6 * polish lossed item * fix failed unittest * polish by reviewer comments
-
由 Shang Zhizhou 提交于
* optimize errror report * add test case for pad op converter * fix some spelling mistake commented by peiyang
-
由 GaoWei8 提交于
* replace sequence length attr to input
-
由 Pei Yang 提交于
* fix trt_dynamic_shape_ernie_deserialize_test * support when opt cache dir does not exist
-
- 14 9月, 2020 10 次提交
-
-
由 joanna.wozna.intel 提交于
-
由 lilong12 提交于
* improve err report, test=develop
-
由 Zhong Hui 提交于
Enhance the error messages for files in operators/math
-
由 Chen Weihang 提交于
-
由 Pei Yang 提交于
-
由 Zhen Wang 提交于
Update amp_check_finite_and_scale_op and add an updating_loss_scaling op for static graph amp training. (#26240) * update amp_check_finite_and_scale_op for static_amp. * use amp_check_finite_and_scale in static graph amp. * update grads to zero when grads own infinite values(as for amp_checkout_finite_and_scale op). * add update_loss_scaling op in cpp. * add update_loss_scaling_op unit test. * update the doc of the check_finite_and_unscale op * Update the process of gradients updating skipping if the gradients have infinite values. * update the way to zero grads. * update test_update_loss_scaling_op.py * add log info when find infinite grads. * add the unit test for UpdateLossScaling Layer.
-
由 ShenLiang 提交于
* rm auto from localsgd
-
由 Adam 提交于
* Add int8 GRU kernel with UTs * Lint fixes * More lint fixes
-
由 石晓伟 提交于
-
由 Jack Zhou 提交于
Error description optimize for math dir
-
- 13 9月, 2020 1 次提交
-
-
由 Zhang Ting 提交于
-
- 12 9月, 2020 1 次提交
-
-
由 lidanqing 提交于
* Fix the lookup_table_v2 failed on GRU mkldnn kernel issue test=develop * fix according to reviews, removed x_num_col_dims test=develop * update gru model. change according to reviews test=develop * change according to reviews test=develop
-
- 11 9月, 2020 4 次提交
-
-
由 Chen Weihang 提交于
-
由 Wilber 提交于
-
由 Wilber 提交于
-
由 furnace 提交于
-
- 10 9月, 2020 2 次提交