[Seq2Seq Trainer] Make sure padding is implemented for models without pad_token (#8043)

* make sure padding is implemented for non-padding tokens models as well

* add better error message

* add better warning

* remove results files

* Update examples/seq2seq/seq2seq_trainer.py

* remove unnecessary copy line

* correct usage of labels

* delete test files

examples/seq2seq/seq2seq_trainer.py

@@ -1,4 +1,3 @@
-import copy
 from typing import Any, Dict, Optional, Tuple, Union
 
 import torch
@@ -60,6 +59,11 @@ class Seq2SeqTrainer(Trainer):
                 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."
 
+        if self.config.pad_token_id is None and self.config.eos_token_id is not None:
+            logger.warn(
+                f"The `config.pad_token_id` is `None`. Using `config.eos_token_id` = {self.config.eos_token_id} for padding.."
+            )
+
     def create_optimizer_and_scheduler(self, num_training_steps: int):
         """
         Setup the optimizer and the learning rate scheduler.
@@ -126,22 +130,19 @@ class Seq2SeqTrainer(Trainer):
                 else DistributedSampler(self.train_dataset)
             )
 
-    def _compute_loss(self, model, inputs):
+    def _compute_loss(self, model, inputs, labels):
-        inputs = copy.deepcopy(inputs)
         if self.args.label_smoothing == 0:
             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]
+                loss, logits = model(**inputs, labels=labels, 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, _ = label_smoothed_nll_loss(
@@ -150,7 +151,8 @@ class Seq2SeqTrainer(Trainer):
         return loss, logits
 
     def compute_loss(self, model, inputs):
-        loss, _ = self._compute_loss(model, inputs)
+        labels = inputs.pop("labels")
+        loss, _ = self._compute_loss(model, inputs, labels)
         return loss
 
     def prediction_step(
@@ -178,25 +180,27 @@ class Seq2SeqTrainer(Trainer):
         """
         inputs = self._prepare_inputs(inputs)
 
+        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,
+        }
+
         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:
+            if generated_tokens.shape[-1] < gen_kwargs["max_length"]:
                 generated_tokens = self._pad_tensors_to_max_len(generated_tokens, gen_kwargs["max_length"])
 
-            # compute loss on predict data
+        labels = inputs.pop("labels")
         with torch.no_grad():
-            loss, logits = self._compute_loss(model, inputs)
+            # compute loss on predict data
+            loss, logits = self._compute_loss(model, inputs, labels)
 
         loss = loss.mean().detach()
         if self.args.prediction_loss_only:
@@ -204,14 +208,21 @@ class Seq2SeqTrainer(Trainer):
 
         logits = generated_tokens if self.args.predict_with_generate else logits
 
-        labels = inputs["labels"]
+        if labels.shape[-1] < gen_kwargs["max_length"]:
-        if self.config.pad_token_id is not None:
+            labels = self._pad_tensors_to_max_len(labels, gen_kwargs["max_length"])
-            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(
+        # If PAD token is not defined at least EOS token has to be defined
+        pad_token_id = self.config.pad_token_id if self.config.pad_token_id is not None else self.config.eos_token_id
+
+        if pad_token_id is None:
+            raise ValueError(
+                f"Make sure that either `config.pad_token_id` or `config.eos_token_id` is defined if tensor has to be padded to `max_length`={max_length}"
+            )
+
+        padded_tensor = pad_token_id * torch.ones(
             (tensor.shape[0], max_length), dtype=tensor.dtype, device=tensor.device
         )
         padded_tensor[:, : tensor.shape[-1]] = tensor

examples/seq2seq/test_finetune_trainer.py

@@ -63,7 +63,9 @@ class TestFinetuneTrainer(TestCasePlus):
         tokenizer = BertTokenizer.from_pretrained("bert-base-uncased")
 
         bert2bert.config.vocab_size = bert2bert.config.encoder.vocab_size
+        bert2bert.config.eos_token_id = tokenizer.sep_token_id
         bert2bert.config.decoder_start_token_id = tokenizer.cls_token_id
+        bert2bert.config.max_length = 128
 
         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%]")