src/hyperpod_nemo_adapter/collections/data/dummy_data_module.py [29:52]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
        return self._build_dataloader(self._train_ds, batch_size=self.cfg.model.train_batch_size)

    def val_dataloader(self):
        """We're not doing validation for synthetic data"""
        return None

    def get_vocab_size(self):
        """
        Respect predefined model vocab size from recipe. Otherwise use
        default vocab size unless hf config vocab size if provided.
        """
        vocab_size = _DEFAULT_VOCAB_SIZE
        if self.cfg.model.get("vocab_size", None) and self.trainer.model.predefined_model:
            return self.cfg.model.vocab_size

        hf_model_name_or_path = self.cfg.model.get("hf_model_name_or_path", None)
        if hf_model_name_or_path:
            hf_config = get_hf_config_from_name_or_path(self.cfg.model)
            if hf_config:
                vocab_size = hf_config.vocab_size

        return vocab_size

    def get_batch(self, data):
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -



src/hyperpod_nemo_adapter/collections/data/dummy_dpo_data_module.py [29:52]:
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
        return self._build_dataloader(self._train_ds, batch_size=self.cfg.model.train_batch_size)

    def val_dataloader(self):
        """We're not doing validation for synthetic DPO data"""
        return None

    def get_vocab_size(self):
        """
        Respect predefined model vocab size from recipe. Otherwise use
        default vocab size unless hf config vocab size if provided.
        """
        vocab_size = _DEFAULT_VOCAB_SIZE
        if self.cfg.model.get("vocab_size", None) and self.trainer.model.predefined_model:
            return self.cfg.model.vocab_size

        hf_model_name_or_path = self.cfg.model.get("hf_model_name_or_path", None)
        if hf_model_name_or_path:
            hf_config = get_hf_config_from_name_or_path(self.cfg.model)
            if hf_config:
                vocab_size = hf_config.vocab_size

        return vocab_size

    def get_batch(self, data):
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -