EXP: "exp1" DEVICE: "cpu" NO_UNK: 1 BEAM_SEARCH: 0 EVAL_PER_EPOCH : 10 BATCH_SIZE: 32 EPOCHS: 50 LEARNING_RATE: 0.0001 MAX_SEQ_LEN: 50 D_MODEL: 512 N_HEAD: 8 NUM_LAYERS: 6 DIM_FEEDFORWARD: 2048 DROPOUT: 0.1