Support T5 Distillation w/hidden state supervision (#7599)

examples/seq2seq/distillation.py

@@ -28,7 +28,7 @@ from lightning_base import generic_train  # noqa
 class BartSummarizationDistiller(SummarizationModule):
     """Supports Bart, Pegasus and other models that inherit from Bart."""
 
-    loss_names = ["loss", "ce_loss", "mlm_loss", "enc_mse_loss", "hid_loss_enc", "hid_loss_dec"]
+    loss_names = ["loss", "ce_loss", "mlm_loss", "hid_loss_enc", "hid_loss_dec"]
 
     def __init__(self, hparams):
         assert Path(hparams.data_dir).exists()
@@ -46,9 +46,19 @@ class BartSummarizationDistiller(SummarizationModule):
         if hparams.length_penalty != -1:
             student.config.length_penalty = hparams.length_penalty
         super().__init__(hparams, model=student, config=student.config)
+        model_type = student.config.model_type
         self.e_layer_ids, self.d_layer_ids = e_layer_ids, d_layer_ids  # type: List[int], List[int]
-        self.different_encoder = hparams.student_encoder_layers != teacher.config.encoder_layers
-        self.different_decoder = hparams.student_decoder_layers != teacher.config.decoder_layers
+
+        if model_type == "t5":
+            teacher_encoder_layers = len(teacher.get_encoder().block)
+            teacher_decoder_layers = len(teacher.get_decoder().block)
+        else:
+            teacher_encoder_layers = teacher.config.encoder_layers
+            teacher_decoder_layers = teacher.config.decoder_layers
+
+        self.different_encoder = hparams.student_encoder_layers != teacher_encoder_layers
+        self.different_decoder = hparams.student_decoder_layers != teacher_decoder_layers
+
         self.teacher = teacher
         freeze_params(self.teacher)
 
@@ -59,17 +69,17 @@ class BartSummarizationDistiller(SummarizationModule):
                 del self.teacher.encoder
         # Intermediate supervision: Decide which layers to supervise
         if hparams.supervise_forward:
-            self.d_matches = get_layers_to_supervise(
-                n_student=len(self.d_layer_ids), n_teacher=self.teacher.config.decoder_layers
-            )
-        else:
+            self.e_matches = get_layers_to_supervise(n_student=len(self.e_layer_ids), n_teacher=teacher_encoder_layers)
+            self.d_matches = get_layers_to_supervise(n_student=len(self.d_layer_ids), n_teacher=teacher_decoder_layers)
+        else:  # student layer should emulate hidden states of the teacher layer it was copied from
+            self.e_matches = self.e_layer_ids
             self.d_matches = self.d_layer_ids
+
         self.ce_loss_fct = nn.KLDivLoss(reduction="batchmean")
         self.temperature = 2.0
         self.alpha_mlm = hparams.alpha_mlm
         self.alpha_ce = hparams.alpha_ce
         self.alpha_hid = hparams.alpha_hid
-        self.alpha_encoder_loss = hparams.alpha_encoder_loss
         gc.collect()
         torch.cuda.empty_cache()
 
@@ -129,7 +139,7 @@ class BartSummarizationDistiller(SummarizationModule):
             output_hidden_states=True,
             output_attentions=False,
             use_cache=False,
-        )  # TODO(@sshleifer): return_dict=True cleanup
+        )
 
         # Same cross entropy vs. label smoothing logic as finetune.py
         assert lm_logits.shape[-1] == self.model.config.vocab_size
@@ -146,30 +156,32 @@ class BartSummarizationDistiller(SummarizationModule):
         def zero_tensor():
             return torch.tensor(0.0).type_as(student_lm_loss)
 
-        loss_encoder, hid_loss_enc, hid_loss_dec = zero_tensor(), zero_tensor(), zero_tensor()
-        if self.different_encoder:
+        hid_loss_enc, hid_loss_dec = zero_tensor(), zero_tensor()
+        if self.different_encoder:  # compute encoder hidden state loss
             with torch.no_grad():
-                teacher_enc_outputs, teacher_enc_hid, _ = self.teacher.get_encoder()(
-                    input_ids, attention_mask=src_mask, output_hidden_states=True
-                )
-            # DEPRECATE THIS
-            if self.hparams.alpha_encoder_loss > 0:
-                loss_encoder = self.calc_mse_loss(enc_outputs, teacher_enc_outputs, src_mask)
-
-            hid_loss_enc = self.calc_hidden_loss(src_mask, enc_hidden_state, teacher_enc_hid, self.e_layer_ids)
-
-        teacher_enc_outputs = (enc_outputs,)
-        assert isinstance(teacher_enc_outputs, tuple), type(teacher_enc_outputs)
+                teacher_enc_hid = self.teacher.get_encoder()(
+                    input_ids, attention_mask=src_mask, output_hidden_states=True, return_dict=True
+                ).hidden_states
+
+            hid_loss_enc = self.calc_hidden_loss(
+                src_mask,
+                enc_hidden_state,
+                teacher_enc_hid,
+                self.e_matches,
+                normalize_hidden=self.hparams.normalize_hidden,
+            )
 
         with torch.no_grad():
-            tloss, tlogits, tdec_hidden, _ = self.teacher(
+            outputs = self.teacher(
                 input_ids,
                 attention_mask=src_mask,
-                encoder_outputs=teacher_enc_outputs,
+                encoder_outputs=(enc_outputs,),
                 decoder_input_ids=decoder_input_ids,
                 lm_labels=labels,
                 output_hidden_states=True,
+                return_dict=True,
             )
+            tlogits, tdec_hidden = outputs.logits, outputs.decoder_hidden_states
         dec_mask = decoder_input_ids.ne(pad_token_id)
         loss_ce = self.calc_ce_loss(dec_mask, lm_logits, tlogits)
         if self.alpha_hid > 0:  # Intermediate supervision of decoder hidden states
@@ -180,10 +192,9 @@ class BartSummarizationDistiller(SummarizationModule):
         blended_loss = (
             self.alpha_ce * loss_ce
             + self.alpha_mlm * student_lm_loss
-            + self.hparams.alpha_encoder_loss * loss_encoder
             + self.hparams.alpha_hid * (hid_loss_enc + hid_loss_dec)
         )
-        return blended_loss, loss_ce, student_lm_loss, loss_encoder, hid_loss_enc, hid_loss_dec
+        return blended_loss, loss_ce, student_lm_loss, hid_loss_enc, hid_loss_dec
 
     @staticmethod
     def calc_hidden_loss(attention_mask, hidden_states, hidden_states_T, matches, normalize_hidden):
@@ -207,7 +218,6 @@ def add_distill_args(parser):
     parser.add_argument("--teacher", type=str)
     parser.add_argument("--alpha_ce", default=0.8, type=float)
     parser.add_argument("--alpha_mlm", default=0.2, type=float)
-    parser.add_argument("--alpha_encoder_loss", default=0.0, type=float)
     parser.add_argument("--alpha_hid", default=0.0, type=float, required=False)
     parser.add_argument("--student_decoder_layers", default=12, type=int, required=False)
     parser.add_argument("--student_encoder_layers", default=12, type=int, required=False)

examples/seq2seq/test_seq2seq_examples.py

@@ -86,7 +86,6 @@ CHEAP_ARGS = {
     "n_val": -1,
     "n_test": -1,
     "student_encoder_layers": 1,
-    "alpha_encoder_loss": 0.0,
     "freeze_encoder": False,
     "auto_scale_batch_size": False,
 }
@@ -230,7 +229,6 @@ class TestSummarizationDistiller(unittest.TestCase):
 
         evaluate_checkpoint(ckpts[0], dest_dir=Path(tempfile.mkdtemp()))
 
-    @unittest.skip("T5 distillation is broken at the moment")
     def test_distill_t5(self):
         updates = dict(
             student_encoder_layers=1,
@@ -255,7 +253,6 @@ class TestSummarizationDistiller(unittest.TestCase):
             model_name_or_path="sshleifer/tinier_bart",
             teacher=CHEAP_ARGS["model_name_or_path"],
             val_check_interval=0.5,
-            alpha_encoder_loss=0.4,
         )
         default_updates.update(updates)
         args_d: dict = CHEAP_ARGS.copy()