Skip to content

Latest commit

 

History

History
6 lines (4 loc) · 339 Bytes

README.md

File metadata and controls

6 lines (4 loc) · 339 Bytes

cllm_rnn

first stab at character level language model rnn

Note that this is a simple RNN. It is NOT a seq2seq model. That is, it DOES not consume the entire input sequence before generating the entire output sequence.

In fact, the output is kind of interesting. We simply output the loss directly, without any prediction task.