File size: 1,789 Bytes
4557487 437a8ab 4557487 9258b0c 437a8ab 4557487 4f33d95 4557487 4f33d95 9258b0c 4f33d95 9258b0c 4f33d95 4557487 4f33d95 4557487 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 |
from __gin__ import dynamic_registration
import tasks
import __main__ as train_script
from t5.data import mixtures
from t5x import models
from t5x import partitioning
from t5x import utils
include "t5x/examples/t5/mt5/large.gin"
include "t5x/configs/runs/finetune.gin"
MIXTURE_OR_TASK_NAME = "angry_tweets"
TASK_FEATURE_LENGTHS = {"inputs": 512, "targets": 8}
TRAIN_STEPS = 1_005_000 # 1000000 pre-trained steps + 10000 fine-tuning steps.
USE_CACHED_TASKS = False
DROPOUT_RATE = 0.1
RANDOM_SEED = 0
# Pere: Only necessary if we load a t5 model. We can start with an t5x model here
# `LOSS_NORMALIZING_FACTOR`: When fine-tuning a model that was pre-trained
# using Mesh Tensorflow (e.g. the public T5 / mT5 / ByT5 models), this should be
# set to `pretraining batch_size` * `target_token_length`. For T5 and T5.1.1:
# `2048 * 114`. For mT5: `1024 * 229`. For ByT5: `1024 * 189`.
#LOSS_NORMALIZING_FACTOR = 234496
#INITIAL_CHECKPOINT_PATH = "gs://t5-data/pretrained_models/t5x/mt5_base/checkpoint_1000000"
#INITIAL_CHECKPOINT_PATH = "gs://t5-data/pretrained_models/t5x/mt5_large/checkpoint_1000000"
#INITIAL_CHECKPOINT_PATH = "gs://nb-t5x-us-central2/pk_nb_t5x_base_run1/checkpoint_1100000"
INITIAL_CHECKPOINT_PATH = "gs://nb-t5x-us-central2/norwegian_NCC_plus_English_pluss200k_scandinavian_t5x_large/checkpoint_1700000"
#train_script.train:
# eval_period = 500
# partitioner = @partitioning.ModelBasedPjitPartitioner()
# partitioning.PjitPartitioner.num_partitions = 1
# `num_decodes` is equivalent to a beam size in a beam search decoding.
# models.EncoderDecoderModel.predict_batch_with_aux.num_decodes = 1
#mesh_transformer.learning_rate_schedules.constant_learning_rate.learning_rate = 0.0005
#run.learning_rate_schedule = @learning_rate_schedules.constant_learning_rate
|