search

stanfordnlp / GloVe

Software in C and data files for the popular GloVe model for distributed word representations, a.k.a. word vectors or embeddings
Apache License 2.0
6.82k stars 1.51k forks source link

Finetune GloVe #100

Open gombru opened 6 years ago

gombru commented 6 years ago

I have trained a GloVe model and I want to finetune it to another smaller dataset. Is it possible? I don't see code to do that.

gombru commented 6 years ago

If I'm not mistaken that provides support to convert a GloVe model to word2vec and work with it in Gensim. But what I want to do is to finetune a trained GloVe model following the GloVe training procedure.

2017-10-26 18:08 GMT+02:00 Athif Shaffy notifications@github.com:

This might help you. https://machinelearningmastery.com/develop-word-embeddings-python-gensim/

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/stanfordnlp/GloVe/issues/100#issuecomment-339716775, or mute the thread https://github.com/notifications/unsubscribe-auth/AORkUQF0a0p1i_pQEW4wCs4hyJih4Y4rks5swK5ygaJpZM4QH0MD .

ImtiazKhanDS commented 6 years ago

Did you get the answer to this? , If yes can you direct us to the resources?

npeirson commented 6 years ago

If I'm accurately understanding what you want to do, the answer is counterintuitive. You can fine-tune at any time by resuming training with the new parameters and/or dataset. If you prefer, you can also start from the ground up, but in most cases it's unnecessary and time consuming.
In other words, set up your new data and parameters like you're going to start over, but instead resume training of your original model.