GitLab

Some tests expect a clean model parallel slate and complain if a previous test left something behind; this change clears more variables that the tests complain about.

hidden_size % attention_heads == 0 is handled above when dealing with kv_channels.

Adding check for decoder sequence length.