Semantic enhancement

[LumosN]

hello, where is the code of the part of Semantic enhancement ?

thanks for your reply.

[zouguojian]

In the run_train.py file, please check from lines 150 to 170.

Best,
Guojian

[LumosN]

Thank you very much! If the tensor shape is (B,T,N,D), how should I deal with this...?  Conv1d just can be used for 3D data. Obliviate ***@***.***  

…

------------------ 原始邮件 ------------------ 发件人: "zouguojian/STGIN" ***@***.***>; 发送时间: 2024年10月24日(星期四) 下午4:27 ***@***.***>; ***@***.******@***.***>; 主题: Re: [zouguojian/STGIN] Semantic enhancement (Issue #7) In the run_train.py file, please check from lines 150 to 170. Best, Guojian — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: ***@***.***>

[zouguojian]

you can use like Pytorch function to transpose the shape position.

I mean, you can check the similar function to use in tensorflow。

Best,
Guojian