This documentation is for a development version. Click here for the latest stable release (v1.2.0).
See the following examples for demonstrations
In most of the examples, we set seeds on the network and vocabulary objects. This is done for consistency; all the examples should work across a wide variety of seeds, but occasionally a “bad seed” will cause the results to be different from what we describe. We set the seeds to avoid this problem. In practice, such adverse results can typically be avoided by using more dimensions per semantic pointer (at least 64). We often use smaller numbers of dimensions in the examples so that they run quickly, but this makes bad seeds somewhat more common.
- Convolution network
- Associative Memory
- Simple question answering
- Controlled question answering
- Simple question answering with memory
- Routing through a sequence
- Routed sequencing
- Parsing simple commands
- Vocabulary casting