include 't5x/examples/t5/mt5/xl.gin'
include 'pretrain_cont.gin'
#include 't5x/configs/runs/pretrain.gin'
#iinclude 't5x/configs/runs/finetune.gin'


# Register necessary SeqIO Tasks/Mixtures.
import t5.data.mixtures
import tasks

MIXTURE_OR_TASK_NAME = "ncc_english_span_corruption_stream"
TASK_FEATURE_LENGTHS = {"inputs": 512, "targets": 512}
TRAIN_STEPS = 1_500_000
DROPOUT_RATE = 0.0 # Changed from the default since T5-1.1 recomments this.
INITIAL_CHECKPOINT_PATH = "gs://t5-data/pretrained_models/t5x/mt5_xl/checkpoint_1000000"
# For training on v4-32
PjitPartitioner.num_partitions = 2
utils.SaveCheckpointConfig.period = 20000
utils.SaveCheckpointConfig.keep = 3