Seq2Seq Imitation Learning for Tactile Feedback-based Manipulation