[Examples] Allow EncoderDecoderModels to be trained with Seq2Seq (#7809)

* Make Seq2Seq Trainer more similar to Trainer

* fix typo

* fix seq2seq trainer

* remove from tests

* remove lock

* remove train files

* delete test files

* correct typo

* check at init

* make sure trainer is not slowed down on TPU

* correct isort

* remove use cache

* fix use cache

* add last use chache = false

examples/seq2seq/finetune_trainer.py

@@ -16,7 +16,6 @@ from transformers import (
 )
 from transformers.trainer_utils import EvaluationStrategy
 from utils import (
-    LegacySeq2SeqDataset,
     Seq2SeqDataCollator,
     Seq2SeqDataset,
     assert_all_frozen,
@@ -138,6 +137,10 @@ class DataTrainingArguments:
     src_lang: Optional[str] = field(default=None, metadata={"help": "Source language id for translation."})
     tgt_lang: Optional[str] = field(default=None, metadata={"help": "Target language id for translation."})
     eval_beams: Optional[int] = field(default=None, metadata={"help": "# num_beams to use for evaluation."})
+    ignore_pad_token_for_loss: bool = field(
+        default=True,
+        metadata={"help": "If only pad tokens should be ignored. This assumes that `config.pad_token_id` is defined."},
+    )
 
 
 def main():
@@ -223,7 +226,7 @@ def main():
         freeze_params(model.get_encoder())
         assert_all_frozen(model.get_encoder())
 
-    dataset_class = Seq2SeqDataset if hasattr(tokenizer, "prepare_seq2seq_batch") else LegacySeq2SeqDataset
+    dataset_class = Seq2SeqDataset
 
     # Get datasets
     train_dataset = (

examples/seq2seq/seq2seq_trainer.py

-import logging
+import copy
 from typing import Any, Dict, Optional, Tuple, Union
 
 import torch
 from torch import nn
 from torch.utils.data import DistributedSampler, RandomSampler
 
-from transformers import Trainer
+from transformers import PreTrainedModel, Trainer, logging
 from transformers.configuration_fsmt import FSMTConfig
 from transformers.file_utils import is_torch_tpu_available
 from transformers.optimization import (
@@ -27,7 +27,7 @@ except ImportError:
     from utils import label_smoothed_nll_loss
 
 
-logger = logging.getLogger(__name__)
+logger = logging.get_logger(__name__)
 
 arg_to_scheduler = {
     "linear": get_linear_schedule_with_warmup,
@@ -41,13 +41,25 @@ arg_to_scheduler_choices = sorted(arg_to_scheduler.keys())
 
 
 class Seq2SeqTrainer(Trainer):
-    def __init__(self, config, data_args, *args, **kwargs):
+    def __init__(self, config=None, data_args=None, *args, **kwargs):
         super().__init__(*args, **kwargs)
-        self.config = config
+
+        if config is None:
+            assert isinstance(
+                self.model, PreTrainedModel
+            ), f"If no `config` is passed the model to be trained has to be of type `PreTrainedModel`, but is {self.model.__class__}"
+            self.config = self._actual_model(self.model).config
+        else:
+            self.config = config
+
         self.data_args = data_args
-        self.max_gen_length = data_args.val_max_target_length
         self.vocab_size = self.config.tgt_vocab_size if isinstance(self.config, FSMTConfig) else self.config.vocab_size
 
+        if self.args.label_smoothing != 0 or (self.data_args is not None and self.data_args.ignore_pad_token_for_loss):
+            assert (
+                self.config.pad_token_id is not None
+            ), "Make sure that `config.pad_token_id` is correcly defined when ignoring `pad_token` for loss calculation or doing label smoothing."
+
     def create_optimizer_and_scheduler(self, num_training_steps: int):
         """
         Setup the optimizer and the learning rate scheduler.
@@ -114,23 +126,31 @@ class Seq2SeqTrainer(Trainer):
                 else DistributedSampler(self.train_dataset)
             )
 
-    def compute_loss(self, model, inputs):
-        labels = inputs.pop("labels")
-        outputs = model(**inputs, use_cache=False)
-        logits = outputs[0]
-        return self._compute_loss(logits, labels)
-
-    def _compute_loss(self, logits, labels):
+    def _compute_loss(self, model, inputs):
+        inputs = copy.deepcopy(inputs)
         if self.args.label_smoothing == 0:
-            # Same behavior as modeling_bart.py
-            loss_fct = torch.nn.CrossEntropyLoss(ignore_index=self.config.pad_token_id)
-            assert logits.shape[-1] == self.vocab_size
-            loss = loss_fct(logits.view(-1, logits.shape[-1]), labels.view(-1))
+            if self.data_args is not None and self.data_args.ignore_pad_token_for_loss:
+                # force training to ignore pad token
+                labels = inputs.pop("labels")
+                logits = model(**inputs, use_cache=False)[0]
+
+                loss_fct = torch.nn.CrossEntropyLoss(ignore_index=self.config.pad_token_id)
+                loss = loss_fct(logits.view(-1, logits.shape[-1]), labels.view(-1))
+            else:
+                # compute usual loss via models
+                loss, logits = model(**inputs, use_cache=False)[:2]
         else:
+            # compute label smoothed loss
+            labels = inputs.pop("labels")
+            logits = model(**inputs, use_cache=False)[0]
             lprobs = torch.nn.functional.log_softmax(logits, dim=-1)
-            loss, nll_loss = label_smoothed_nll_loss(
+            loss, _ = label_smoothed_nll_loss(
                 lprobs, labels, self.args.label_smoothing, ignore_index=self.config.pad_token_id
             )
+        return loss, logits
+
+    def compute_loss(self, model, inputs):
+        loss, _ = self._compute_loss(model, inputs)
         return loss
 
     def prediction_step(
@@ -158,31 +178,37 @@ class Seq2SeqTrainer(Trainer):
         """
         inputs = self._prepare_inputs(inputs)
 
+        if self.args.predict_with_generate and not self.args.prediction_loss_only:
+            gen_kwargs = {
+                "max_length": self.data_args.val_max_target_length
+                if self.data_args is not None
+                else self.config.max_length,
+                "num_beams": self.data_args.eval_beams if self.data_args is not None else self.config.num_beams,
+            }
+            generated_tokens = model.generate(
+                inputs["input_ids"],
+                attention_mask=inputs["attention_mask"],
+                **gen_kwargs,
+            )
+            # in case the batch is shorter than max length, the output should be padded
+            if self.config.pad_token_id is not None:
+                generated_tokens = self._pad_tensors_to_max_len(generated_tokens, gen_kwargs["max_length"])
+
+            # compute loss on predict data
         with torch.no_grad():
-            if self.args.predict_with_generate and not self.args.prediction_loss_only:
-                generated_tokens = model.generate(
-                    inputs["input_ids"],
-                    attention_mask=inputs["attention_mask"],
-                    use_cache=True,
-                    num_beams=self.data_args.eval_beams,
-                    max_length=self.max_gen_length,
-                )
-                # in case the batch is shorter than max length, the output should be padded
-                generated_tokens = self._pad_tensors_to_max_len(generated_tokens, self.max_gen_length)
-
-            labels_out = inputs.get("labels")
-            # Call forward again to get loss # TODO: avoidable?
-            outputs = model(**inputs, use_cache=False)
-            loss = self._compute_loss(outputs[1], labels_out)
-            loss = loss.mean().detach()
-            if self.args.prediction_loss_only:
-                return (loss, None, None)
-
-            logits = generated_tokens if self.args.predict_with_generate else outputs[1]
-
-        labels_out = labels_out.detach()
-        labels = self._pad_tensors_to_max_len(labels_out, self.max_gen_length)
-        return (loss, logits.detach(), labels)
+            loss, logits = self._compute_loss(model, inputs)
+
+        loss = loss.mean().detach()
+        if self.args.prediction_loss_only:
+            return (loss, None, None)
+
+        logits = generated_tokens if self.args.predict_with_generate else logits
+
+        labels = inputs["labels"]
+        if self.config.pad_token_id is not None:
+            labels = self._pad_tensors_to_max_len(labels, self.config.max_length)
+
+        return (loss, logits, labels)
 
     def _pad_tensors_to_max_len(self, tensor, max_length):
         padded_tensor = self.config.pad_token_id * torch.ones(

examples/seq2seq/test_finetune_trainer.py

@@ -5,12 +5,14 @@ from unittest.mock import patch
 
 import pytest
 
-from transformers import is_torch_available
+from transformers import BertTokenizer, EncoderDecoderModel, is_torch_available
+from transformers.file_utils import is_datasets_available
 from transformers.testing_utils import TestCasePlus, slow
 from transformers.trainer_callback import TrainerState
 from transformers.trainer_utils import set_seed
 
-from .finetune_trainer import main
+from .finetune_trainer import Seq2SeqTrainingArguments, main
+from .seq2seq_trainer import Seq2SeqTrainer
 from .test_seq2seq_examples import MBART_TINY
 from .utils import execute_async_std
 
@@ -50,6 +52,117 @@ class TestFinetuneTrainer(TestCasePlus):
         assert "test_generations.txt" in contents
         assert "test_results.json" in contents
 
+    @slow
+    def test_finetune_bert2bert(self):
+        if not is_datasets_available():
+            return
+
+        import datasets
+
+        bert2bert = EncoderDecoderModel.from_encoder_decoder_pretrained("prajjwal1/bert-tiny", "prajjwal1/bert-tiny")
+        tokenizer = BertTokenizer.from_pretrained("bert-base-uncased")
+
+        bert2bert.config.vocab_size = bert2bert.config.encoder.vocab_size
+        bert2bert.config.decoder_start_token_id = tokenizer.cls_token_id
+
+        train_dataset = datasets.load_dataset("cnn_dailymail", "3.0.0", split="train[:1%]")
+        val_dataset = datasets.load_dataset("cnn_dailymail", "3.0.0", split="validation[:1%]")
+
+        train_dataset = train_dataset.select(range(32))
+        val_dataset = val_dataset.select(range(16))
+
+        rouge = datasets.load_metric("rouge")
+
+        batch_size = 4
+
+        def _map_to_encoder_decoder_inputs(batch):
+            # Tokenizer will automatically set [BOS] <text> [EOS]
+            inputs = tokenizer(batch["article"], padding="max_length", truncation=True, max_length=512)
+            outputs = tokenizer(batch["highlights"], padding="max_length", truncation=True, max_length=128)
+            batch["input_ids"] = inputs.input_ids
+            batch["attention_mask"] = inputs.attention_mask
+
+            batch["decoder_input_ids"] = outputs.input_ids
+            batch["labels"] = outputs.input_ids.copy()
+            batch["labels"] = [
+                [-100 if token == tokenizer.pad_token_id else token for token in labels] for labels in batch["labels"]
+            ]
+            batch["decoder_attention_mask"] = outputs.attention_mask
+
+            assert all([len(x) == 512 for x in inputs.input_ids])
+            assert all([len(x) == 128 for x in outputs.input_ids])
+
+            return batch
+
+        def _compute_metrics(pred):
+            labels_ids = pred.label_ids
+            pred_ids = pred.predictions
+
+            # all unnecessary tokens are removed
+            pred_str = tokenizer.batch_decode(pred_ids, skip_special_tokens=True)
+            label_str = tokenizer.batch_decode(labels_ids, skip_special_tokens=True)
+
+            rouge_output = rouge.compute(predictions=pred_str, references=label_str, rouge_types=["rouge2"])[
+                "rouge2"
+            ].mid
+
+            return {
+                "rouge2_precision": round(rouge_output.precision, 4),
+                "rouge2_recall": round(rouge_output.recall, 4),
+                "rouge2_fmeasure": round(rouge_output.fmeasure, 4),
+            }
+
+        # map train dataset
+        train_dataset = train_dataset.map(
+            _map_to_encoder_decoder_inputs,
+            batched=True,
+            batch_size=batch_size,
+            remove_columns=["article", "highlights"],
+        )
+        train_dataset.set_format(
+            type="torch",
+            columns=["input_ids", "attention_mask", "decoder_input_ids", "decoder_attention_mask", "labels"],
+        )
+
+        # same for validation dataset
+        val_dataset = val_dataset.map(
+            _map_to_encoder_decoder_inputs,
+            batched=True,
+            batch_size=batch_size,
+            remove_columns=["article", "highlights"],
+        )
+        val_dataset.set_format(
+            type="torch",
+            columns=["input_ids", "attention_mask", "decoder_input_ids", "decoder_attention_mask", "labels"],
+        )
+
+        output_dir = self.get_auto_remove_tmp_dir()
+
+        training_args = Seq2SeqTrainingArguments(
+            output_dir=output_dir,
+            per_device_train_batch_size=batch_size,
+            per_device_eval_batch_size=batch_size,
+            predict_with_generate=True,
+            evaluate_during_training=True,
+            do_train=True,
+            do_eval=True,
+            warmup_steps=0,
+            eval_steps=2,
+            logging_steps=2,
+        )
+
+        # instantiate trainer
+        trainer = Seq2SeqTrainer(
+            model=bert2bert,
+            args=training_args,
+            compute_metrics=_compute_metrics,
+            train_dataset=train_dataset,
+            eval_dataset=val_dataset,
+        )
+
+        # start training
+        trainer.train()
+
     def run_trainer(self, eval_steps: int, max_len: str, model_name: str, num_train_epochs: int):
 
         # XXX: remove hardcoded path