You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@singa.apache.org by "wangwei (JIRA)" <ji...@apache.org> on 2016/05/03 11:22:12 UTC

[jira] [Created] (SINGA-164) Add the base Tensor class

wangwei created SINGA-164:
-----------------------------

             Summary: Add the base Tensor class
                 Key: SINGA-164
                 URL: https://issues.apache.org/jira/browse/SINGA-164
             Project: Singa
          Issue Type: New Feature
            Reporter: wangwei
            Assignee: wangwei


Each Tensor instance is a multi-dimension array allocated on a device (e.g., GPU or CPU).
Operations for linear algebra and neural networks will be implemented against Tensor. Other abstractions, e.g., Layer/Model, are on top of Tensor. Users can implement new layers/model using tensor operations without knowing the details of memory management and execution scheduling.

Sparse Tensor would be considered later for special applications like Word2Vec model.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)