chore: Merge branch 'inter_document_masking_for_attention' into loss_filtering

Author: le1nux

src/modalities/config/config.py

@@ -461,6 +461,21 @@ class BatchSamplerConfig(BaseModel):
 class GPT2LLMCollateFnConfig(BaseModel):
     sample_key: str
     target_key: str
+    sub_seq_lengths_key: str | None = None
+    eos_token_id: int | None = None
+    padding_token_id: int | None = None
+
+    @model_validator(mode="after")
+    def check_sub_seq_lengths_and_eos_token(self) -> "GPT2LLMCollateFnConfig":
+        if (self.sub_seq_lengths_key is None) != (self.eos_token_id is None):
+            raise ValueError("Either both or neither of sub_seq_lengths_key and eos_token_id must be provided.")
+        return self
+
+    @model_validator(mode="after")
+    def check_padding_token_and_sub_seq_lengths(self) -> "GPT2LLMCollateFnConfig":
+        if self.padding_token_id is not None and self.sub_seq_lengths_key is None:
+            raise ValueError("If padding_token_id is provided, sub_seq_lengths_key must also be provided.")
+        return self
 
 
 class LLMDataLoaderConfig(BaseModel):

src/modalities/models/gpt2/collator.py

@@ -7,16 +7,31 @@
 class GPT2LLMCollateFn(CollateFnIF):
     """GPT2LLMCollateFn class to define a collate function for GPT2 language model."""
 
-    def __init__(self, sample_key: str, target_key: str):
+    def __init__(
+        self,
+        sample_key: str,
+        target_key: str,
+        sub_seq_lengths_key: str | None = None,
+        eos_token_id: int | None = None,
+        padding_token_id: int | None = None,
+    ):
         """
         Initializes the Collator object.
+        If the eos token ID and the sub_seq_lengths_key are provided,
+        a list[list[int]] representing the sub-sequence lengths will be created.
 
         Args:
             sample_key (str): The key for accessing the sample data.
             target_key (str): The key for accessing the target data.
+            sub_seq_lengths_key (str | None): The key for accessing the sub-sequence lengths.
+            eos_token_id (int | None): The end-of-sequence token ID.
+            padding_token_id (int | None): The padding token ID.
         """
         self.sample_key = sample_key
         self.target_key = target_key
+        self.sub_seq_lengths_key = sub_seq_lengths_key
+        self.eos_token_id = eos_token_id
+        self.padding_token_id = padding_token_id
 
     def __call__(self, batch: list[dict[str, torch.Tensor]]) -> DatasetBatch:
         """
@@ -33,4 +48,43 @@ def __call__(self, batch: list[dict[str, torch.Tensor]]) -> DatasetBatch:
         sample_tensor = torch.stack([torch.tensor(d[self.sample_key]) for d in batch])
         samples = {self.sample_key: sample_tensor[:, :-1]}
         targets = {self.target_key: sample_tensor[:, 1:]}
+        if self.sub_seq_lengths_key is not None:
+            # Determine sub sequence lengths by finding the eos tokens in each sequence in the batch.
+            sub_seq_lengths = self._compute_sub_sequence_lengths_for_each_sequence(samples[self.sample_key])
+            samples[self.sub_seq_lengths_key] = sub_seq_lengths
         return DatasetBatch(targets=targets, samples=samples)
+
+    def _compute_sub_sequence_lengths_for_each_sequence(self, sample_tensor: torch.Tensor) -> list[list[int]]:
+        sub_seq_lengths_in_batch = []
+        for batch_seq in sample_tensor:
+            eos_positions = (batch_seq == self.eos_token_id).nonzero(as_tuple=True)[0]
+            if len(eos_positions) == 0:
+                assert self.padding_token_id is None or (
+                    batch_seq[0] != self.padding_token_id and torch.all(batch_seq != self.padding_token_id)
+                ), "Whole batch sequence consists of padding tokens."
+                sub_seq_lengths_in_batch.append([len(batch_seq)])
+            else:
+                lens_in_seq = self._compute_subsequence_length_in_sequence(batch_seq, eos_positions)
+                sub_seq_lengths_in_batch.append(lens_in_seq)
+        return sub_seq_lengths_in_batch
+
+    def _compute_subsequence_length_in_sequence(self, seq: torch.Tensor, eos_positions: torch.Tensor) -> list[int]:
+        # If the last sequence is cut, i.e. does not end on an eos token,
+        # it should also be included unless the padding token is set and
+        # the last sequence is just padding.
+        last_eos_pos = eos_positions[-1].item()
+        if self._has_cutoff_final_sequence(seq, last_eos_pos):
+            eos_positions = torch.cat([eos_positions, eos_positions.new_tensor([len(seq) - 1])])
+        # Compute length of each subsequence and add to lengths list.
+        sub_seq_lengths = []
+        prev_pos = 0
+        for pos in eos_positions:
+            sub_seq_lengths.append(pos.item() - prev_pos + 1)
+            prev_pos = pos.item() + 1
+        return sub_seq_lengths
+
+    def _has_cutoff_final_sequence(self, seq: torch.Tensor, last_eos_pos: int) -> bool:
+        # Assumption: If the first token of the last sequence is padding, so is the rest.
+        return last_eos_pos < len(seq) - 1 and (
+            self.padding_token_id is None or seq[last_eos_pos + 1] != self.padding_token_id
+        )