The reformer training for the English-Polish pair (paracrawl).

PiperOrigin-RevId: 321256624

trax/supervised/configs/reformer_pc_enpl.gin

@@ -0,0 +1,108 @@
+# Copyright 2020 The Trax Authors.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+import trax.models
+import trax.optimizers
+import trax.supervised.tf_inputs
+import trax.supervised.trainer_lib
+
+import t5.data.preprocessors
+import t5.data.sentencepiece_vocabulary
+
+max_length = 512
+sequence_length = {'inputs': 512, 'targets': 512}
+mean_noise_span_length = 3.0
+noise_density = 0.15
+
+# Parameters for batcher:
+# ==============================================================================
+batcher.data_streams = @tf_inputs.data_streams
+batcher.batch_size_per_device = 256
+batcher.eval_batch_size = 64
+batcher.max_eval_length = 512
+batcher.bucket_length = 32
+batcher.buckets_include_inputs_in_length=True
+batcher.id_to_mask = 0
+
+# Parameters for data_streams:
+# ==============================================================================
+data_streams.data_dir = None
+data_streams.eval_holdout_size = 0.1
+data_streams.dataset_name = 'para_crawl/enpl_plain_text'
+data_streams.bare_preprocess_fn = @trax.supervised.tf_inputs.generic_text_dataset_preprocess_fn
+data_streams.input_name = 'inputs'
+data_streams.target_name = 'targets'
+
+# Parameters for filter_dataset_on_len:
+# ==============================================================================
+filter_dataset_on_len.len_map = {'inputs': (1, 512), 'targets': (1, 512)}
+filter_dataset_on_len.filter_on_eval = True
+
+# Parameters for truncate_dataset_on_len:
+# ==============================================================================
+truncate_dataset_on_len.len_map = {'inputs': 512, 'targets': 512}
+truncate_dataset_on_len.truncate_on_eval = True
+
+# Parameters for generic_text_dataset_preprocess_fn:
+# ==============================================================================
+
+generic_text_dataset_preprocess_fn.text_preprocess_fns = [
+    @rekey/get_t5_preprocessor_by_name()
+]
+generic_text_dataset_preprocess_fn.token_preprocess_fns = [
+    @trax.supervised.tf_inputs.truncate_dataset_on_len,
+    @trax.supervised.tf_inputs.filter_dataset_on_len,
+]
+
+# Parameters for get_t5_preprocessor_by_name:
+# ==============================================================================
+rekey/get_t5_preprocessor_by_name.name = 'rekey'
+rekey/get_t5_preprocessor_by_name.fn_kwargs = {'key_map': {'inputs': 'en', 'targets': 'pl'}}
+
+# Parameters for multifactor:
+# ==============================================================================
+# 0.044 ~= 512^-0.5 = d_model^-0.5
+multifactor.constant = 0.088
+multifactor.factors = 'constant * linear_warmup * rsqrt_decay'
+multifactor.warmup_steps = 8000
+
+# Parameters for Adam:
+# ==============================================================================
+Adam.b1 = 0.9
+Adam.b2 = 0.98
+Adam.eps = 1e-9
+
+# Parameters for train:
+# ==============================================================================
+train.eval_frequency = 1000
+train.eval_steps = 10
+train.model = @trax.models.Reformer
+train.optimizer = @trax.optimizers.Adam
+train.steps = 500000
+train.save_graphs = False
+train.checkpoints_at = [100000, 200000, 300000, 400000, 500000]
+
+# Parameters for Reformer:
+# ==============================================================================
+Reformer.d_model= 512
+Reformer.d_ff = 2048
+Reformer.dropout = 0.1
+Reformer.ff_activation = @trax.layers.Relu
+Reformer.ff_dropout = 0.1
+Reformer.max_len = 2048
+Reformer.mode = 'train'
+Reformer.n_heads = 8
+Reformer.n_encoder_layers = 6
+Reformer.n_decoder_layers = 6
+Reformer.input_vocab_size = 8064