I think the usage of this framework should more focus on embedding construction with other deep learning framework. such as non.Embedding in pytorch and so on.
With the help of Additional Featurization section, will add more useful info to embedding.
And try init input_embeds in huggingface bert model,
This may be a convenient way to add info into bert model in the embedding level.
I think the usage of this framework should more focus on embedding construction with other deep learning framework. such as non.Embedding in pytorch and so on.
With the help of Additional Featurization section, will add more useful info to embedding.
And try init input_embeds in huggingface bert model,
This may be a convenient way to add info into bert model in the embedding level.