Fine Tuning Universal Sentence Encoder Model

1770 views
Skip to first unread message

Hao Zhang

unread,
May 8, 2018, 4:23:33 AM5/8/18
to TensorFlow Hub
Hi, 

I want to fine tuning the USE model with my dataset. I have defined a cosine similarity layer after sentence encodings, and defined a logistic regression loss to optimize the model. But there is an error that no variables to optimize, I have set the trainable to True. There should be variables in the USE model. Could you tell me how to fix this issue?

Best regards,
Hao  

Vojtech Bardiovsky

unread,
May 16, 2018, 5:30:34 AM5/16/18
to zhangha...@gmail.com, TensorFlow Hub
Hi Hao,


--
You received this message because you are subscribed to the Google Groups "TensorFlow Hub" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hub+uns...@tensorflow.org.
Visit this group at https://groups.google.com/a/tensorflow.org/group/hub/.


--
Vojtech

dheeraj agrawal

unread,
Oct 16, 2019, 4:26:36 AM10/16/19
to TensorFlow Hub

Amogh Mishra

unread,
Oct 25, 2019, 2:10:25 AM10/25/19
to dheeraj agrawal, TensorFlow Hub
Dear Dheeraj,

This is great. Thanks for sharing.

Regards,

Amogh Mishra

MS in Data Science at Columbia University (Class of 2020)

--
You received this message because you are subscribed to the Google Groups "TensorFlow Hub" group.
To unsubscribe from this group and stop receiving emails from it, send an email to hub+uns...@tensorflow.org.
To view this discussion on the web visit https://groups.google.com/a/tensorflow.org/d/msgid/hub/3625e7d2-ad9b-4ef4-9b3c-153994580077%40tensorflow.org.


--

Amogh Mishra (id: am5323)

MS in Data Science at Columbia University (Class of 2020)
Reply all
Reply to author
Forward
0 new messages