Yu Meng comments

Results 43 comments of


                                            Yu Meng

Python package (bindings)

Hi, Thanks for asking. I do plan to make a python package for the code but have been busy with some deadlines. I might work on this sometime after June...

Can this method be used in the scenario of multi-label text classification?

Hi @ScottishFold007 The method may require some nontrivial changes to be applied to multi-label classification tasks. The most challenging part to adapt will be the self-training stage. The current self-training...

Upgrade tensorflow version to support higher versions of CUDA

Hi @sundoufu , Thanks for the suggestion! I've updated the requirements file. Best, Yu

Upgrade tensorflow version to support higher versions of CUDA

Yes, it should work without changing any code.

The package version

Hi, You could refer to [this requirement file](https://github.com/yumeng5/WeSHClass/blob/master/requirements.txt) for the versions. If you are still encountering errors, please let me know. Thanks, Yu

The example does not run on M1 processors

Hi @angelesaldunate, The code was developed on a Linux system, and I'm not sure whether it will be able to run on Apple M1 at all. Could you please try...

Can I train on cpu and control layer freezing ?

Hi, While it's possible to freeze most layers to save computation and make it possible to train on CPUs, it's not recommended to do so -- the hyperparameters (e.g., learning...

AssertionError: Too few (0) documents with category indicative terms found for category 1; try to add more unlabeled documents to the training corpus (recommend) or reduce `--match_threshold` (not recommend)

Hi, The error is pretty much explained by the printouts -- for several categories (1, 2, 6) there are 0 documents with category indicative terms (as indicated by the dictionary...

AssertionError: Too few (0) documents with category indicative terms found for category 1; try to add more unlabeled documents to the training corpus (recommend) or reduce `--match_threshold` (not recommend)

The number of documents found with category indicative terms is derived based on the category vocabulary constructed in the first step and is not directly related to the actual number...

Could you please offer a script for gpt2 to generate the training data?

Hi, We have only used GPT-2 for generation on MNLI and SST-2 tasks, but the code adaptation for other tasks should be straightforward. You do need to change a few...