1. 07 4月, 2020 1 次提交
  2. 04 4月, 2020 1 次提交
  3. 03 4月, 2020 3 次提交
    • Y
      Merge pull request #26 from willthefrog/fix_lr_scaling · 44ba39b7
      Yang Zhang 提交于
      Fix learning rate scaling bug
      44ba39b7
    • Y
      Fix learning rate scaling bug · 810ece8f
      Yang Zhang 提交于
      this bug is quite peculiar and hard to track down, when learning rate for a
      parameter is scaled via param_attr and learning rate schedulers are used,
      `append_optimizer_op` will error out complaining `LearningRate` input is null
      
      turns out learning rate scaling is done in `_create_param_lr`, which basically
      add a scale op, the problem is: it is appended to `orig_prog` (since
      `global_learning_rate()` variable is in it), therefore the resulting scaled
      learning rate variable can not be found in `train_prog`.
      
      the reason it works previously w/o lr scaling is this:
      `clone()` will create a variable with the same name as the
      `global_learning_rate()` variable, which will be used in `append_optimizer_op`
      810ece8f
    • K
      Merge pull request #24 from huangjun12/bmn-model · d8d5176d
      Kaipeng Deng 提交于
      Add bmn model
      d8d5176d
  4. 02 4月, 2020 6 次提交
  5. 01 4月, 2020 4 次提交
  6. 31 3月, 2020 6 次提交
  7. 30 3月, 2020 7 次提交
  8. 27 3月, 2020 2 次提交
  9. 26 3月, 2020 5 次提交
  10. 25 3月, 2020 5 次提交