1. 09 12月, 2020 2 次提交
  2. 08 12月, 2020 7 次提交
  3. 07 12月, 2020 10 次提交
  4. 05 12月, 2020 1 次提交
    • myq406450149's avatar
      update unbind norm add CUDAPlace api doc information (#29322) · 7c508d86
      myq406450149 提交于
      * enhance array_to_lod_tensor_op lod_tensor_to_array_op errors information. test=develop
      
      * fix format. test=develop
      
      * format fix. test=develop
      
      * add lod_rank_table. test=develop
      
      * fix format. test=develop
      
      * fix doc info. test=develop
      
      * fix np error
      
      * add unbind dygraph api. test=develop
      
      * fix unbind doc.test=develop
      7c508d86
  5. 04 12月, 2020 6 次提交
  6. 03 12月, 2020 9 次提交
  7. 02 12月, 2020 5 次提交
    • W
      fix analysis_config bug. (#29304) · d68af02c
      Wilber 提交于
      d68af02c
    • S
      enforce the matmul_v2 error message (#29297) · 0fb18bc2
      ShenLiang 提交于
      0fb18bc2
    • Z
      Remove some useless log. (#29300) · 9b59a589
      Zhen Wang 提交于
      9b59a589
    • L
      fix shape of tile_grad op (#29289) · 13a22a37
      Leo Chen 提交于
      13a22a37
    • Z
      Add pure fp16 training with master weights. (#27712) · be3777a5
      Zhen Wang 提交于
      * add the weight decay func for the momentum op
      
      * Add the multi_precision function in Momentum Optimizer.
      
      * Make sure that the initial value of master weights are same with the fp16 weights.
      
      * add static loss scaling.
      
      * add the rescale_grad function in the pure fp16 training.
      
      * use the original momentum updating method.
      
      * Polish some codes, such as variable names.
      
      * add docstring for apis.
      
      * update the var creation details of _create_master_weight.
      
      * not modify codes about imperative momentum updating.
      
      * Fix the error of test_dist_sparse_tensor_load_momentum UT.
      
      * add unit test for multi precision fp16 training.
      
      * add more unit tests for CI.
      
      * Use lower threshold values for allclose comparing in test_multi_precision_fp16_train UT.
      
      * For CI Coverage Checking.
      be3777a5