diff --git a/multimolecule/models/__init__.py b/multimolecule/models/__init__.py
index 48c90187..1076773d 100644
--- a/multimolecule/models/__init__.py
+++ b/multimolecule/models/__init__.py
@@ -13,6 +13,13 @@
     RnaMsmForTokenClassification,
     RnaMsmModel,
 )
+from .splicebert import (
+    SpliceBertConfig,
+    SpliceBertForMaskedLM,
+    SpliceBertForSequenceClassification,
+    SpliceBertForTokenClassification,
+    SpliceBertModel,
+)
 
 __all__ = [
     "RnaBertConfig",
@@ -25,5 +32,10 @@
     "RnaMsmForMaskedLM",
     "RnaMsmForSequenceClassification",
     "RnaMsmForTokenClassification",
+    "SpliceBertConfig",
+    "SpliceBertModel",
+    "SpliceBertForMaskedLM",
+    "SpliceBertForSequenceClassification",
+    "SpliceBertForTokenClassification",
     "RnaTokenizer",
 ]
diff --git a/multimolecule/models/splicebert/__init__.py b/multimolecule/models/splicebert/__init__.py
new file mode 100644
index 00000000..ee9892cd
--- /dev/null
+++ b/multimolecule/models/splicebert/__init__.py
@@ -0,0 +1,36 @@
+from transformers import (
+    AutoConfig,
+    AutoModel,
+    AutoModelForMaskedLM,
+    AutoModelForSequenceClassification,
+    AutoModelForTokenClassification,
+    AutoModelWithLMHead,
+    AutoTokenizer,
+)
+
+from multimolecule.tokenizers.rna import RnaTokenizer
+
+from .configuration_splicebert import SpliceBertConfig
+from .modeling_splicebert import (
+    SpliceBertForMaskedLM,
+    SpliceBertForSequenceClassification,
+    SpliceBertForTokenClassification,
+    SpliceBertModel,
+)
+
+__all__ = [
+    "SpliceBertConfig",
+    "SpliceBertModel",
+    "SpliceTokenizer",
+    "SpliceBertForMaskedLM",
+    "SpliceBertForSequenceClassification",
+    "SpliceBertForTokenClassification",
+]
+
+AutoConfig.register("splicebert", SpliceBertConfig)
+AutoModel.register(SpliceBertConfig, SpliceBertModel)
+AutoModelForMaskedLM.register(SpliceBertConfig, SpliceBertForMaskedLM)
+AutoModelForSequenceClassification.register(SpliceBertConfig, SpliceBertForSequenceClassification)
+AutoModelForTokenClassification.register(SpliceBertConfig, SpliceBertForTokenClassification)
+AutoModelWithLMHead.register(SpliceBertConfig, SpliceBertForTokenClassification)
+AutoTokenizer.register(SpliceBertConfig, RnaTokenizer)
diff --git a/multimolecule/models/splicebert/configuration_splicebert.py b/multimolecule/models/splicebert/configuration_splicebert.py
new file mode 100644
index 00000000..d7b39aee
--- /dev/null
+++ b/multimolecule/models/splicebert/configuration_splicebert.py
@@ -0,0 +1,101 @@
+from transformers.configuration_utils import PretrainedConfig
+from transformers.utils import logging
+
+logger = logging.get_logger(__name__)
+
+
+class SpliceBertConfig(PretrainedConfig):
+    r"""
+    This is the configuration class to store the configuration of a [`SpliceBertModel`]. It is used to instantiate a
+    SpliceBert model according to the specified arguments, defining the model architecture. Instantiating a
+    configuration with the defaults will yield a similar configuration to that of the SpliceBert
+    [biomed-AI/SpliceBERT](https://github.com/biomed-AI/SpliceBERT) architecture.
+
+    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+    documentation from [`PretrainedConfig`] for more information.
+
+
+    Args:
+        vocab_size (`int`, *optional*):
+            Vocabulary size of the SpliceBert model. Defines the number of different tokens that can be represented by
+            the `inputs_ids` passed when calling [`SpliceBertModel`].
+        mask_token_id (`int`, *optional*):
+            The index of the mask token in the vocabulary. This must be included in the config because of the
+            "mask-dropout" scaling trick, which will scale the inputs depending on the number of masked tokens.
+        pad_token_id (`int`, *optional*):
+            The index of the padding token in the vocabulary. This must be included in the config because certain parts
+            of the SpliceBert code use this instead of the attention mask.
+        hidden_size (`int`, *optional*, defaults to 768):
+            Dimensionality of the encoder layers and the pooler layer.
+        num_hidden_layers (`int`, *optional*, defaults to 12):
+            Number of hidden layers in the Transformer encoder.
+        num_attention_heads (`int`, *optional*, defaults to 12):
+            Number of attention heads for each attention layer in the Transformer encoder.
+        intermediate_size (`int`, *optional*, defaults to 3072):
+            Dimensionality of the "intermediate" (often named feed-forward) layer in the Transformer encoder.
+        hidden_dropout (`float`, *optional*, defaults to 0.1):
+            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
+        attention_dropout (`float`, *optional*, defaults to 0.1):
+            The dropout ratio for the attention probabilities.
+        max_position_embeddings (`int`, *optional*, defaults to 1026):
+            The maximum sequence length that this model might ever be used with. Typically set this to something large
+            just in case (e.g., 512 or 1024 or 2048).
+        initializer_range (`float`, *optional*, defaults to 0.02):
+            The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
+        layer_norm_eps (`float`, *optional*, defaults to 1e-12):
+            The epsilon used by the layer normalization layers.
+
+    Examples:
+
+    ```python
+    >>> from multimolecule import SpliceBertModel, SpliceBertConfig
+
+    >>> # Initializing a SpliceBert style configuration >>> configuration = SpliceBertConfig()
+
+    >>> # Initializing a model from the configuration >>> model = SpliceBertModel(configuration)
+
+    >>> # Accessing the model configuration >>> configuration = model.config
+    ```
+    """
+
+    model_type = "splicebert"
+
+    def __init__(
+        self,
+        vocab_size=25,
+        hidden_size=None,
+        num_hidden_layers=6,
+        num_attention_heads=12,
+        intermediate_size=40,
+        hidden_act="gelu",
+        hidden_dropout=0.0,
+        attention_dropout=0.0,
+        max_position_embeddings=440,
+        initializer_range=0.02,
+        layer_norm_eps=1e-12,
+        pad_token_id=0,
+        position_embedding_type="absolute",
+        use_cache=True,
+        head_dropout=None,
+        head_transform=None,
+        head_act=None,
+        **kwargs,
+    ):
+        super().__init__(pad_token_id=pad_token_id, **kwargs)
+
+        self.vocab_size = vocab_size
+        self.hidden_size = hidden_size
+        self.num_hidden_layers = num_hidden_layers
+        self.num_attention_heads = num_attention_heads
+        self.intermediate_size = intermediate_size
+        self.hidden_act = hidden_act
+        self.hidden_dropout = hidden_dropout
+        self.attention_dropout = attention_dropout
+        self.max_position_embeddings = max_position_embeddings
+        self.initializer_range = initializer_range
+        self.layer_norm_eps = layer_norm_eps
+        self.position_embedding_type = position_embedding_type
+        self.use_cache = use_cache
+        self.head_dropout = head_dropout
+        self.head_transform = head_transform
+        self.head_act = head_act
diff --git a/multimolecule/models/splicebert/convert_checkpoint.py b/multimolecule/models/splicebert/convert_checkpoint.py
new file mode 100644
index 00000000..47491fb0
--- /dev/null
+++ b/multimolecule/models/splicebert/convert_checkpoint.py
@@ -0,0 +1,116 @@
+import os
+from typing import Optional
+
+import chanfig
+import torch
+from torch import nn
+
+from multimolecule.models import SpliceBertConfig as Config
+from multimolecule.models import SpliceBertForMaskedLM as Model
+from multimolecule.tokenizers.rna.utils import get_special_tokens_map, get_tokenizer_config, get_vocab_list
+
+try:
+    from huggingface_hub import HfApi
+except ImportError:
+    HfApi = None
+
+
+torch.manual_seed(1013)
+vocab_list = get_vocab_list()
+
+
+def _convert_checkpoint(config, original_state_dict, original_vocab_list):
+    state_dict = {}
+    for key, value in original_state_dict.items():
+        key = key.replace("LayerNorm", "layer_norm")
+        key = key.replace("gamma", "weight")
+        key = key.replace("beta", "bias")
+        if key.startswith("bert"):
+            state_dict["splice" + key] = value
+            continue
+        if key.startswith("cls"):
+            key = "lm_head." + key[4:]
+            state_dict[key] = value
+            continue
+
+    word_embed = nn.Embedding(config.vocab_size, config.hidden_size, padding_idx=config.pad_token_id)
+    word_embed_weight = word_embed.weight.data
+    predictions_bias = torch.zeros(config.vocab_size)
+    predictions_decoder_weight = torch.zeros((config.vocab_size, config.hidden_size))
+    # nn.init.normal_(pos_embed.weight, std=0.02)
+    for original_index, original_token in enumerate(original_vocab_list):
+        new_index = vocab_list.index(original_token)
+        word_embed_weight[new_index] = state_dict["splicebert.embeddings.word_embeddings.weight"][original_index]
+        predictions_bias[new_index] = state_dict["lm_head.predictions.bias"][original_index]
+        predictions_decoder_weight[new_index] = state_dict["lm_head.predictions.decoder.weight"][original_index]
+    state_dict["splicebert.embeddings.word_embeddings.weight"] = word_embed_weight
+    state_dict["lm_head.predictions.bias"] = predictions_bias
+    state_dict["lm_head.predictions.decoder.bias"] = predictions_bias
+    state_dict["lm_head.predictions.decoder.weight"] = predictions_decoder_weight
+    del state_dict["splicebert.embeddings.position_ids"]
+    return state_dict
+
+
+def convert_checkpoint(convert_config):
+    config = Config.from_dict(chanfig.load(os.path.join(convert_config.checkpoint_path, "config.json")))
+    del config._name_or_path
+    config.architectures = ["SpliceBertForMaskedLM"]
+    config.vocab_size = len(vocab_list)
+
+    model = Model(config)
+
+    ckpt = torch.load(
+        os.path.join(convert_config.checkpoint_path, "pytorch_model.bin"), map_location=torch.device("cpu")
+    )
+    vocab = []
+    for char in open(os.path.join(convert_config.checkpoint_path, "vocab.txt")).read().splitlines():  # noqa: SIM115
+        if char.startswith("["):
+            char = char.lower().replace("[", "<").replace("]", ">")
+        if char == "T":
+            char = "U"
+        if char == "<sep>":
+            char = "<eos>"
+        vocab.append(char)
+    state_dict = _convert_checkpoint(config, ckpt, vocab)
+
+    model.load_state_dict(state_dict)
+    model.save_pretrained(convert_config.output_path, safe_serialization=True)
+    model.save_pretrained(convert_config.output_path, safe_serialization=False)
+    chanfig.NestedDict(get_special_tokens_map()).json(
+        os.path.join(convert_config.output_path, "special_tokens_map.json")
+    )
+    chanfig.NestedDict(get_tokenizer_config()).json(os.path.join(convert_config.output_path, "tokenizer_config.json"))
+
+    if convert_config.push_to_hub:
+        if HfApi is None:
+            raise ImportError("Please install huggingface_hub to push to the hub.")
+        api = HfApi()
+        api.create_repo(
+            convert_config.repo_id,
+            token=convert_config.token,
+            exist_ok=True,
+        )
+        api.upload_folder(
+            repo_id=convert_config.repo_id, folder_path=convert_config.output_path, token=convert_config.token
+        )
+
+
+@chanfig.configclass
+class ConvertConfig:
+    checkpoint_path: str
+    output_path: Optional[str] = None
+    push_to_hub: bool = False
+    repo_id: Optional[str] = output_path
+    token: Optional[str] = None
+
+    def post(self):
+        if self.output_path is None:
+            self.output_path = self.checkpoint_path.split("/")[-1].lower()
+        if self.repo_id is None:
+            self.repo_id = f"multimolecule/{self.output_path}"
+
+
+if __name__ == "__main__":
+    config = ConvertConfig()
+    config.parse()  # type: ignore[attr-defined]
+    convert_checkpoint(config)
diff --git a/multimolecule/models/splicebert/modeling_splicebert.py b/multimolecule/models/splicebert/modeling_splicebert.py
new file mode 100644
index 00000000..0c8bd0da
--- /dev/null
+++ b/multimolecule/models/splicebert/modeling_splicebert.py
@@ -0,0 +1,1039 @@
+import math
+from typing import List, Optional, Tuple, Union
+
+import torch
+import torch.utils.checkpoint
+from torch import nn
+from torch.nn import functional as F
+from transformers.activations import ACT2FN
+from transformers.modeling_outputs import (
+    BaseModelOutputWithPastAndCrossAttentions,
+    BaseModelOutputWithPoolingAndCrossAttentions,
+    MaskedLMOutput,
+    SequenceClassifierOutput,
+    TokenClassifierOutput,
+)
+from transformers.modeling_utils import PreTrainedModel
+from transformers.pytorch_utils import apply_chunking_to_forward, find_pruneable_heads_and_indices, prune_linear_layer
+from transformers.utils import logging
+
+from ..modeling_utils import SequenceClassificationHead, TokenClassificationHead
+from .configuration_splicebert import SpliceBertConfig
+
+try:
+    from flash_attn import flash_attn_func
+except ImportError:
+    flash_attn_func = None
+
+
+logger = logging.get_logger(__name__)
+
+
+class SpliceBertPreTrainedModel(PreTrainedModel):
+    """
+    An abstract class to handle weights initialization and a simple interface for downloading and loading pretrained
+    models.
+    """
+
+    config_class = SpliceBertConfig
+    base_model_prefix = "splicebert"
+    supports_gradient_checkpointing = True
+    _no_split_modules = ["SpliceBertLayer", "SpliceBertEmbeddings"]
+
+    # Copied from transformers.models.bert.modeling_bert.BertPreTrainedModel._init_weights
+    def _init_weights(self, module: nn.Module):
+        """Initialize the weights"""
+        if isinstance(module, nn.Linear):
+            # Slightly different from the TF version which uses truncated_normal for initialization
+            # cf https://github.com/pytorch/pytorch/pull/5617
+            module.weight.data.normal_(mean=0.0, std=self.config.initializer_range)
+            if module.bias is not None:
+                module.bias.data.zero_()
+        elif isinstance(module, nn.Embedding):
+            module.weight.data.normal_(mean=0.0, std=self.config.initializer_range)
+            if module.padding_idx is not None:
+                module.weight.data[module.padding_idx].zero_()
+        elif isinstance(module, nn.LayerNorm):
+            module.bias.data.zero_()
+            module.weight.data.fill_(1.0)
+
+    def _set_gradient_checkpointing(self, module, value=False):
+        if isinstance(module, SpliceBertEncoder):
+            module.gradient_checkpointing = value
+
+
+class SpliceBertModel(SpliceBertPreTrainedModel):
+    """
+    The model can behave as an encoder (with only self-attention) as well as a decoder, in which case a layer of
+    cross-attention is added between the self-attention layers, following the architecture described in [Attention is
+    all you need](https://arxiv.org/abs/1706.03762) by Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit,
+    Llion Jones, Aidan N. Gomez, Lukasz Kaiser and Illia Polosukhin.
+
+    To behave as an decoder the model needs to be initialized with the `is_decoder` argument of the configuration set
+    to `True`. To be used in a Seq2Seq model, the model needs to initialized with both `is_decoder` argument and
+    `add_cross_attention` set to `True`; an `encoder_hidden_states` is then expected as an input to the forward pass.
+    """
+
+    def __init__(self, config, add_pooling_layer=True):
+        super().__init__(config)
+        self.config = config
+        self.embeddings = SpliceBertEmbeddings(config)
+        self.encoder = SpliceBertEncoder(config)
+        self.pooler = SpliceBertPooler(config) if add_pooling_layer else None
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def get_input_embeddings(self):
+        return self.embeddings.word_embeddings
+
+    def set_input_embeddings(self, value):
+        self.embeddings.word_embeddings = value
+
+    def _prune_heads(self, heads_to_prune):
+        """
+        Prunes heads of the model. heads_to_prune: dict of {layer_num: list of heads to prune in this layer} See base
+        class PreTrainedModel
+        """
+        for layer, heads in heads_to_prune.items():
+            self.encoder.layer[layer].attention.prune_heads(heads)
+
+    def forward(
+        self,
+        input_ids: Optional[torch.Tensor] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        token_type_ids: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.Tensor] = None,
+        head_mask: Optional[torch.Tensor] = None,
+        inputs_embeds: Optional[torch.Tensor] = None,
+        encoder_hidden_states: Optional[torch.Tensor] = None,
+        encoder_attention_mask: Optional[torch.Tensor] = None,
+        past_key_values: Optional[List[torch.FloatTensor]] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]:
+        r"""
+        encoder_hidden_states  (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*):
+            Sequence of hidden-states at the output of the last layer of the encoder. Used in the cross-attention if
+            the model is configured as a decoder.
+        encoder_attention_mask (`torch.FloatTensor` of shape `(batch_size, sequence_length)`, *optional*):
+            Mask to avoid performing attention on the padding token indices of the encoder input. This mask is used in
+            the cross-attention if the model is configured as a decoder. Mask values selected in `[0, 1]`:
+
+            - 1 for tokens that are **not masked**,
+            - 0 for tokens that are **masked**.
+        past_key_values (`tuple(tuple(torch.FloatTensor))` of length `config.n_layers` with each tuple having 4 tensors
+            of shape `(batch_size, num_heads, sequence_length - 1, embed_size_per_head)`):
+            Contains precomputed key and value hidden states of the attention blocks. Can be used to speed up decoding.
+
+            If `past_key_values` are used, the user can optionally input only the last `decoder_input_ids` (those that
+            don't have their past key value states given to this model) of shape `(batch_size, 1)` instead of all
+            `decoder_input_ids` of shape `(batch_size, sequence_length)`.
+        use_cache (`bool`, *optional*):
+            If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding (see
+            `past_key_values`).
+        """
+        output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions
+        output_hidden_states = (
+            output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states
+        )
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        if self.config.is_decoder:
+            use_cache = use_cache if use_cache is not None else self.config.use_cache
+        else:
+            use_cache = False
+
+        if input_ids is not None and inputs_embeds is not None:
+            raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time")
+        elif input_ids is not None:
+            # self.warn_if_padding_and_no_attention_mask(input_ids, attention_mask)
+            input_shape = input_ids.size()
+        elif inputs_embeds is not None:
+            input_shape = inputs_embeds.size()[:-1]
+        else:
+            raise ValueError("You have to specify either input_ids or inputs_embeds")
+
+        batch_size, seq_length = input_shape
+        device = input_ids.device if input_ids is not None else inputs_embeds.device  # type: ignore[union-attr]
+
+        # past_key_values_length
+        past_key_values_length = past_key_values[0][0].shape[2] if past_key_values is not None else 0
+
+        if attention_mask is None:
+            attention_mask = torch.ones(((batch_size, seq_length + past_key_values_length)), device=device)
+
+        if token_type_ids is None:
+            if hasattr(self.embeddings, "token_type_ids"):
+                buffered_token_type_ids = self.embeddings.token_type_ids[:, :seq_length]
+                buffered_token_type_ids_expanded = buffered_token_type_ids.expand(batch_size, seq_length)
+                token_type_ids = buffered_token_type_ids_expanded
+            else:
+                token_type_ids = torch.zeros(input_shape, dtype=torch.long, device=device)
+
+        # We can provide a self-attention mask of dimensions [batch_size, from_seq_length, to_seq_length]
+        # ourselves in which case we just need to make it broadcastable to all heads.
+        extended_attention_mask: torch.Tensor = self.get_extended_attention_mask(attention_mask, input_shape)
+
+        # If a 2D or 3D attention mask is provided for the cross-attention
+        # we need to make broadcastable to [batch_size, num_heads, seq_length, seq_length]
+        if self.config.is_decoder and encoder_hidden_states is not None:
+            encoder_batch_size, encoder_sequence_length, _ = encoder_hidden_states.size()
+            encoder_hidden_shape = (encoder_batch_size, encoder_sequence_length)
+            if encoder_attention_mask is None:
+                encoder_attention_mask = torch.ones(encoder_hidden_shape, device=device)
+            encoder_extended_attention_mask = self.invert_attention_mask(encoder_attention_mask)
+        else:
+            encoder_extended_attention_mask = None
+
+        # Prepare head mask if needed
+        # 1.0 in head_mask indicate we keep the head
+        # attention_probs has shape bsz x n_heads x N x N
+        # input head_mask has shape [num_heads] or [num_hidden_layers x num_heads]
+        # and head_mask is converted to shape [num_hidden_layers x batch x num_heads x seq_length x seq_length]
+        head_mask = self.get_head_mask(head_mask, self.config.num_hidden_layers)
+
+        embedding_output = self.embeddings(
+            input_ids=input_ids,
+            position_ids=position_ids,
+            token_type_ids=token_type_ids,
+            inputs_embeds=inputs_embeds,
+            past_key_values_length=past_key_values_length,
+        )
+        encoder_outputs = self.encoder(
+            embedding_output,
+            attention_mask=extended_attention_mask,
+            head_mask=head_mask,
+            encoder_hidden_states=encoder_hidden_states,
+            encoder_attention_mask=encoder_extended_attention_mask,
+            past_key_values=past_key_values,
+            use_cache=use_cache,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+        sequence_output = encoder_outputs[0]
+        pooled_output = self.pooler(sequence_output) if self.pooler is not None else None
+
+        if not return_dict:
+            return (sequence_output, pooled_output) + encoder_outputs[1:]
+
+        return BaseModelOutputWithPoolingAndCrossAttentions(
+            last_hidden_state=sequence_output,
+            pooler_output=pooled_output,
+            past_key_values=encoder_outputs.past_key_values,
+            hidden_states=encoder_outputs.hidden_states,
+            attentions=encoder_outputs.attentions,
+            cross_attentions=encoder_outputs.cross_attentions,
+        )
+
+
+class SpliceBertForMaskedLM(SpliceBertPreTrainedModel):
+    _tied_weights_keys = ["predictions.decoder.bias", "cls.predictions.decoder.weight"]
+
+    def __init__(self, config):
+        super().__init__(config)
+
+        if config.is_decoder:
+            logger.warning(
+                "If you want to use `SpliceBertForMaskedLM` make sure `config.is_decoder=False` for "
+                "bi-directional self-attention."
+            )
+
+        self.splicebert = SpliceBertModel(config, add_pooling_layer=False)
+        self.lm_head = SpliceBertOnlyMLMHead(config)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def get_output_embeddings(self):
+        return self.lm_head.predictions.decoder
+
+    def set_output_embeddings(self, new_embeddings):
+        self.lm_head.predictions.decoder = new_embeddings
+
+    def forward(
+        self,
+        input_ids: Optional[torch.Tensor] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        token_type_ids: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.Tensor] = None,
+        head_mask: Optional[torch.Tensor] = None,
+        inputs_embeds: Optional[torch.Tensor] = None,
+        encoder_hidden_states: Optional[torch.Tensor] = None,
+        encoder_attention_mask: Optional[torch.Tensor] = None,
+        labels: Optional[torch.Tensor] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple[torch.Tensor], MaskedLMOutput]:
+        r"""
+        labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
+            Labels for computing the masked language modeling loss. Indices should be in `[-100, 0, ...,
+            config.vocab_size]` (see `input_ids` docstring) Tokens with indices set to `-100` are ignored (masked), the
+            loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`
+        """
+
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        outputs = self.splicebert(
+            input_ids,
+            attention_mask=attention_mask,
+            token_type_ids=token_type_ids,
+            position_ids=position_ids,
+            head_mask=head_mask,
+            inputs_embeds=inputs_embeds,
+            encoder_hidden_states=encoder_hidden_states,
+            encoder_attention_mask=encoder_attention_mask,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+
+        sequence_output = outputs[0]
+        prediction_scores = self.lm_head(sequence_output)
+
+        masked_lm_loss = None
+        if labels is not None:
+            masked_lm_loss = F.cross_entropy(prediction_scores.view(-1, self.config.vocab_size), labels.view(-1))
+
+        if not return_dict:
+            output = (prediction_scores,) + outputs[2:]
+            return ((masked_lm_loss,) + output) if masked_lm_loss is not None else output
+
+        return MaskedLMOutput(
+            loss=masked_lm_loss,
+            logits=prediction_scores,
+            hidden_states=outputs.hidden_states,
+            attentions=outputs.attentions,
+        )
+
+    def prepare_inputs_for_generation(self, input_ids, attention_mask=None, **model_kwargs):
+        input_shape = input_ids.shape
+        effective_batch_size = input_shape[0]
+
+        #  add a dummy token
+        if self.config.pad_token_id is None:
+            raise ValueError("The PAD token should be defined for generation")
+
+        attention_mask = torch.cat([attention_mask, attention_mask.new_zeros((attention_mask.shape[0], 1))], dim=-1)
+        dummy_token = torch.full(
+            (effective_batch_size, 1), self.config.pad_token_id, dtype=torch.long, device=input_ids.device
+        )
+        input_ids = torch.cat([input_ids, dummy_token], dim=1)
+
+        return {"input_ids": input_ids, "attention_mask": attention_mask}
+
+
+class SpliceBertForSequenceClassification(SpliceBertPreTrainedModel):
+    def __init__(self, config):
+        super().__init__(config)
+        self.num_labels = config.num_labels
+        self.config = config
+
+        self.splicebert = SpliceBertModel(config)
+        head_dropout = config.head_dropout if config.head_dropout is not None else config.hidden_dropout
+        self.dropout = nn.Dropout(head_dropout)
+        self.classifier = SequenceClassificationHead(config)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def forward(
+        self,
+        input_ids: Optional[torch.Tensor] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        token_type_ids: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.Tensor] = None,
+        head_mask: Optional[torch.Tensor] = None,
+        inputs_embeds: Optional[torch.Tensor] = None,
+        labels: Optional[torch.Tensor] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple[torch.Tensor], SequenceClassifierOutput]:
+        r"""
+        labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*):
+            Labels for computing the sequence classification/regression loss. Indices should be in `[0, ...,
+            config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If
+            `config.num_labels > 1` a classification loss is computed (Cross-Entropy).
+        """
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        outputs = self.splicebert(
+            input_ids,
+            attention_mask=attention_mask,
+            token_type_ids=token_type_ids,
+            position_ids=position_ids,
+            head_mask=head_mask,
+            inputs_embeds=inputs_embeds,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+        logits = self.classifier(outputs)
+
+        loss = None
+        if labels is not None:
+            if self.config.problem_type is None:
+                if self.num_labels == 1:
+                    self.config.problem_type = "regression"
+                elif self.num_labels > 1 and (labels.dtype == torch.long or labels.dtype == torch.int):
+                    self.config.problem_type = "single_label_classification"
+                else:
+                    self.config.problem_type = "multi_label_classification"
+            if self.config.problem_type == "regression":
+                loss = (
+                    F.mse_loss(logits.squeeze(), labels.squeeze())
+                    if self.num_labels == 1
+                    else F.mse_loss(logits, labels)
+                )
+            elif self.config.problem_type == "single_label_classification":
+                loss = F.cross_entropy(logits.view(-1, self.num_labels), labels.view(-1))
+            elif self.config.problem_type == "multi_label_classification":
+                loss = F.binary_cross_entropy_with_logits(logits, labels)
+        if not return_dict:
+            output = (logits,) + outputs[2:]
+            return ((loss,) + output) if loss is not None else output
+
+        return SequenceClassifierOutput(
+            loss=loss,
+            logits=logits,
+            hidden_states=outputs.hidden_states,
+            attentions=outputs.attentions,
+        )
+
+
+class SpliceBertForTokenClassification(SpliceBertPreTrainedModel):
+    def __init__(self, config):
+        super().__init__(config)
+        self.num_labels = config.num_labels
+
+        self.splicebert = SpliceBertModel(config, add_pooling_layer=False)
+        head_dropout = config.head_dropout if config.head_dropout is not None else config.hidden_dropout
+        self.dropout = nn.Dropout(head_dropout)
+        self.classifier = TokenClassificationHead(config)
+
+        # Initialize weights and apply final processing
+        self.post_init()
+
+    def forward(
+        self,
+        input_ids: Optional[torch.Tensor] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        token_type_ids: Optional[torch.Tensor] = None,
+        position_ids: Optional[torch.Tensor] = None,
+        head_mask: Optional[torch.Tensor] = None,
+        inputs_embeds: Optional[torch.Tensor] = None,
+        labels: Optional[torch.Tensor] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+    ) -> Union[Tuple[torch.Tensor], TokenClassifierOutput]:
+        r"""
+        labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*):
+            Labels for computing the token classification loss. Indices should be in `[0, ..., config.num_labels - 1]`.
+        """
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        outputs = self.splicebert(
+            input_ids,
+            attention_mask=attention_mask,
+            token_type_ids=token_type_ids,
+            position_ids=position_ids,
+            head_mask=head_mask,
+            inputs_embeds=inputs_embeds,
+            output_attentions=output_attentions,
+            output_hidden_states=output_hidden_states,
+            return_dict=return_dict,
+        )
+        logits = self.classifier(outputs)
+
+        loss = None
+        if labels is not None:
+            if self.config.problem_type is None:
+                if self.num_labels == 1:
+                    self.config.problem_type = "regression"
+                elif self.num_labels > 1 and (labels.dtype == torch.long or labels.dtype == torch.int):
+                    self.config.problem_type = "single_label_classification"
+                else:
+                    self.config.problem_type = "multi_label_classification"
+            if self.config.problem_type == "regression":
+                loss = (
+                    F.mse_loss(logits.squeeze(), labels.squeeze())
+                    if self.num_labels == 1
+                    else F.mse_loss(logits, labels)
+                )
+            elif self.config.problem_type == "single_label_classification":
+                loss = F.cross_entropy(logits.view(-1, self.num_labels), labels.view(-1))
+            elif self.config.problem_type == "multi_label_classification":
+                loss = F.binary_cross_entropy_with_logits(logits, labels)
+        if not return_dict:
+            output = (logits,) + outputs[2:]
+            return ((loss,) + output) if loss is not None else output
+
+        return TokenClassifierOutput(
+            loss=loss,
+            logits=logits,
+            hidden_states=outputs.hidden_states,
+            attentions=outputs.attentions,
+        )
+
+
+class SpliceBertEmbeddings(nn.Module):
+    """Construct the embeddings from word, position and token_type embeddings."""
+
+    def __init__(self, config):
+        super().__init__()
+        self.word_embeddings = nn.Embedding(config.vocab_size, config.hidden_size, padding_idx=config.pad_token_id)
+        self.position_embeddings = nn.Embedding(config.max_position_embeddings, config.hidden_size)
+        self.token_type_embeddings = nn.Embedding(config.type_vocab_size, config.hidden_size)
+
+        # self.layer_norm is not snake-cased to stick with TensorFlow model variable name and be able to load
+        # any TensorFlow checkpoint file
+        self.layer_norm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps)
+        self.dropout = nn.Dropout(config.hidden_dropout)
+        # position_ids (1, len position emb) is contiguous in memory and exported when serialized
+        self.position_embedding_type = getattr(config, "position_embedding_type", "absolute")
+        self.register_buffer(
+            "position_ids", torch.arange(config.max_position_embeddings).expand((1, -1)), persistent=False
+        )
+        self.register_buffer(
+            "token_type_ids", torch.zeros(self.position_ids.size(), dtype=torch.long), persistent=False
+        )
+
+    def forward(
+        self,
+        input_ids: Optional[torch.LongTensor] = None,
+        token_type_ids: Optional[torch.LongTensor] = None,
+        position_ids: Optional[torch.LongTensor] = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        past_key_values_length: int = 0,
+    ) -> torch.Tensor:
+        if input_ids is not None:
+            input_shape = input_ids.size()
+        else:
+            input_shape = inputs_embeds.size()[:-1]  # type: ignore[union-attr]
+
+        seq_length = input_shape[1]
+
+        if position_ids is None:
+            position_ids = self.position_ids[:, past_key_values_length : seq_length + past_key_values_length]
+
+        # Setting the token_type_ids to the registered buffer in constructor where it is all zeros, which usually occurs
+        # when its auto-generated, registered buffer helps users when tracing the model without passing token_type_ids,
+        # solves issue #5664
+        if token_type_ids is None:
+            if hasattr(self, "token_type_ids"):
+                buffered_token_type_ids = self.token_type_ids[:, :seq_length]
+                buffered_token_type_ids_expanded = buffered_token_type_ids.expand(input_shape[0], seq_length)
+                token_type_ids = buffered_token_type_ids_expanded
+            else:
+                token_type_ids = torch.zeros(input_shape, dtype=torch.long, device=self.position_ids.device)
+
+        if inputs_embeds is None:
+            inputs_embeds = self.word_embeddings(input_ids)
+        token_type_embeddings = self.token_type_embeddings(token_type_ids)
+
+        embeddings = inputs_embeds + token_type_embeddings
+        if self.position_embedding_type == "absolute":
+            position_embeddings = self.position_embeddings(position_ids)
+            embeddings += position_embeddings
+        embeddings = self.layer_norm(embeddings)
+        embeddings = self.dropout(embeddings)
+        return embeddings
+
+
+class SpliceBertEncoder(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.config = config
+        self.layer = nn.ModuleList([SpliceBertLayer(config) for _ in range(config.num_hidden_layers)])
+        self.gradient_checkpointing = False
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.FloatTensor] = None,
+        head_mask: Optional[torch.FloatTensor] = None,
+        encoder_hidden_states: Optional[torch.FloatTensor] = None,
+        encoder_attention_mask: Optional[torch.FloatTensor] = None,
+        past_key_values: Optional[Tuple[Tuple[torch.FloatTensor]]] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = False,
+        output_hidden_states: Optional[bool] = False,
+        return_dict: Optional[bool] = True,
+    ) -> Union[Tuple[torch.Tensor], BaseModelOutputWithPastAndCrossAttentions]:
+        all_hidden_states = () if output_hidden_states else None
+        all_self_attentions = () if output_attentions else None
+        all_cross_attentions = () if output_attentions and self.config.add_cross_attention else None
+
+        if self.gradient_checkpointing and self.training and use_cache:
+            logger.warning_once(
+                "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..."
+            )
+            use_cache = False
+
+        next_decoder_cache = () if use_cache else None
+        for i, layer_module in enumerate(self.layer):
+            if output_hidden_states:
+                all_hidden_states = all_hidden_states + (hidden_states,)  # type: ignore[operator]
+
+            layer_head_mask = head_mask[i] if head_mask is not None else None
+            past_key_value = past_key_values[i] if past_key_values is not None else None
+
+            if self.gradient_checkpointing and self.training:
+
+                def create_custom_forward(module):
+                    def custom_forward(*inputs):
+                        return module(*inputs, past_key_value, output_attentions)  # noqa: B023
+
+                    return custom_forward
+
+                layer_outputs = torch.utils.checkpoint.checkpoint(
+                    create_custom_forward(layer_module),
+                    hidden_states,
+                    attention_mask,
+                    layer_head_mask,
+                    encoder_hidden_states,
+                    encoder_attention_mask,
+                )
+            else:
+                layer_outputs = layer_module(
+                    hidden_states,
+                    attention_mask,
+                    layer_head_mask,
+                    encoder_hidden_states,
+                    encoder_attention_mask,
+                    past_key_value,
+                    output_attentions,
+                )
+
+            hidden_states = layer_outputs[0]
+            if use_cache:
+                next_decoder_cache += (layer_outputs[-1],)  # type: ignore[operator]
+            if output_attentions:
+                all_self_attentions = all_self_attentions + (layer_outputs[1],)  # type: ignore[operator]
+                if self.config.add_cross_attention:
+                    all_cross_attentions = all_cross_attentions + (layer_outputs[2],)  # type: ignore[operator]
+
+        if output_hidden_states:
+            all_hidden_states = all_hidden_states + (hidden_states,)  # type: ignore[operator]
+
+        if not return_dict:
+            return tuple(
+                v
+                for v in [
+                    hidden_states,
+                    next_decoder_cache,
+                    all_hidden_states,
+                    all_self_attentions,
+                    all_cross_attentions,
+                ]
+                if v is not None
+            )
+        return BaseModelOutputWithPastAndCrossAttentions(
+            last_hidden_state=hidden_states,
+            past_key_values=next_decoder_cache,
+            hidden_states=all_hidden_states,
+            attentions=all_self_attentions,
+            cross_attentions=all_cross_attentions,
+        )
+
+
+class SpliceBertLayer(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.chunk_size_feed_forward = config.chunk_size_feed_forward
+        self.seq_len_dim = 1
+        self.attention = SpliceBertAttention(config)
+        self.is_decoder = config.is_decoder
+        self.add_cross_attention = config.add_cross_attention
+        if self.add_cross_attention:
+            if not self.is_decoder:
+                raise ValueError(f"{self} should be used as a decoder model if cross attention is added")
+            self.crossattention = SpliceBertAttention(config, position_embedding_type="absolute")
+        self.intermediate = SpliceBertIntermediate(config)
+        self.output = SpliceBertOutput(config)
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.FloatTensor] = None,
+        head_mask: Optional[torch.FloatTensor] = None,
+        encoder_hidden_states: Optional[torch.FloatTensor] = None,
+        encoder_attention_mask: Optional[torch.FloatTensor] = None,
+        past_key_value: Optional[Tuple[Tuple[torch.FloatTensor]]] = None,
+        output_attentions: Optional[bool] = False,
+    ) -> Tuple[torch.Tensor]:
+        # decoder uni-directional self-attention cached key/values tuple is at positions 1,2
+        self_attn_past_key_value = past_key_value[:2] if past_key_value is not None else None
+        self_attention_outputs = self.attention(
+            hidden_states,
+            attention_mask,
+            head_mask,
+            output_attentions=output_attentions,
+            past_key_value=self_attn_past_key_value,
+        )
+        attention_output = self_attention_outputs[0]
+
+        # if decoder, the last output is tuple of self-attn cache
+        if self.is_decoder:
+            outputs = self_attention_outputs[1:-1]
+            present_key_value = self_attention_outputs[-1]
+        else:
+            outputs = self_attention_outputs[1:]  # add self attentions if we output attention weights
+
+        cross_attn_present_key_value = None
+        if self.is_decoder and encoder_hidden_states is not None:
+            if not hasattr(self, "crossattention"):
+                raise ValueError(
+                    f"If `encoder_hidden_states` are passed, {self} has to be instantiated with cross-attention layers"
+                    " by setting `config.add_cross_attention=True`"
+                )
+
+            # cross_attn cached key/values tuple is at positions 3,4 of past_key_value tuple
+            cross_attn_past_key_value = past_key_value[-2:] if past_key_value is not None else None
+            cross_attention_outputs = self.crossattention(
+                attention_output,
+                attention_mask,
+                head_mask,
+                encoder_hidden_states,
+                encoder_attention_mask,
+                cross_attn_past_key_value,
+                output_attentions,
+            )
+            attention_output = cross_attention_outputs[0]
+            outputs = outputs + cross_attention_outputs[1:-1]  # add cross attentions if we output attention weights
+
+            # add cross-attn cache to positions 3,4 of present_key_value tuple
+            cross_attn_present_key_value = cross_attention_outputs[-1]
+            present_key_value = present_key_value + cross_attn_present_key_value
+
+        layer_output = apply_chunking_to_forward(
+            self.feed_forward_chunk, self.chunk_size_feed_forward, self.seq_len_dim, attention_output
+        )
+        outputs = (layer_output,) + outputs
+
+        # if decoder, return the attn key/values as the last output
+        if self.is_decoder:
+            outputs = outputs + (present_key_value,)
+
+        return outputs
+
+    def feed_forward_chunk(self, attention_output):
+        intermediate_output = self.intermediate(attention_output)
+        layer_output = self.output(intermediate_output, attention_output)
+        return layer_output
+
+
+class SpliceBertAttention(nn.Module):
+    def __init__(self, config, position_embedding_type=None):
+        super().__init__()
+        self.self = SpliceBertSelfAttention(config, position_embedding_type=position_embedding_type)
+        self.output = SpliceBertSelfOutput(config)
+        self.pruned_heads = set()
+
+    def prune_heads(self, heads):
+        if len(heads) == 0:
+            return
+        heads, index = find_pruneable_heads_and_indices(
+            heads, self.self.num_attention_heads, self.self.attention_head_size, self.pruned_heads
+        )
+
+        # Prune linear layers
+        self.self.query = prune_linear_layer(self.self.query, index)
+        self.self.key = prune_linear_layer(self.self.key, index)
+        self.self.value = prune_linear_layer(self.self.value, index)
+        self.output.dense = prune_linear_layer(self.output.dense, index, dim=1)
+
+        # Update hyper params and store pruned heads
+        self.self.num_attention_heads = self.self.num_attention_heads - len(heads)
+        self.self.all_head_size = self.self.attention_head_size * self.self.num_attention_heads
+        self.pruned_heads = self.pruned_heads.union(heads)
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.FloatTensor] = None,
+        head_mask: Optional[torch.FloatTensor] = None,
+        encoder_hidden_states: Optional[torch.FloatTensor] = None,
+        encoder_attention_mask: Optional[torch.FloatTensor] = None,
+        past_key_value: Optional[Tuple[Tuple[torch.FloatTensor]]] = None,
+        output_attentions: Optional[bool] = False,
+    ) -> Tuple[torch.Tensor]:
+        self_outputs = self.self(
+            hidden_states,
+            attention_mask,
+            head_mask,
+            encoder_hidden_states,
+            encoder_attention_mask,
+            past_key_value,
+            output_attentions,
+        )
+        attention_output = self.output(self_outputs[0], hidden_states)
+        outputs = (attention_output,) + self_outputs[1:]  # add attentions if we output them
+        return outputs
+
+
+class SpliceBertSelfAttention(nn.Module):
+    def __init__(self, config, position_embedding_type=None):
+        super().__init__()
+        if config.hidden_size % config.num_attention_heads != 0 and not hasattr(config, "embedding_size"):
+            raise ValueError(
+                f"The hidden size ({config.hidden_size}) is not a multiple of the number of attention "
+                f"heads ({config.num_attention_heads})"
+            )
+        self.flash = flash_attn_func is not None
+
+        self.num_attention_heads = config.num_attention_heads
+        self.attention_head_size = int(config.hidden_size / config.num_attention_heads)
+        self.all_head_size = self.num_attention_heads * self.attention_head_size
+
+        self.query = nn.Linear(config.hidden_size, self.all_head_size)
+        self.key = nn.Linear(config.hidden_size, self.all_head_size)
+        self.value = nn.Linear(config.hidden_size, self.all_head_size)
+
+        self.dropout = nn.Dropout(config.attention_dropout)
+        self.position_embedding_type = position_embedding_type or getattr(config, "position_embedding_type", "absolute")
+        if self.position_embedding_type == "relative_key" or self.position_embedding_type == "relative_key_query":
+            self.max_position_embeddings = config.max_position_embeddings
+            self.distance_embedding = nn.Embedding(2 * config.max_position_embeddings - 1, self.attention_head_size)
+
+        self.is_decoder = config.is_decoder
+
+    def transpose_for_scores(self, x: torch.Tensor) -> torch.Tensor:
+        new_x_shape = x.size()[:-1] + (self.num_attention_heads, self.attention_head_size)
+        x = x.view(new_x_shape)
+        return x.permute(0, 2, 1, 3)
+
+    def forward(
+        self,
+        hidden_states: torch.Tensor,
+        attention_mask: Optional[torch.FloatTensor] = None,
+        head_mask: Optional[torch.FloatTensor] = None,
+        encoder_hidden_states: Optional[torch.FloatTensor] = None,
+        encoder_attention_mask: Optional[torch.FloatTensor] = None,
+        past_key_value: Optional[Tuple[Tuple[torch.FloatTensor, ...], ...]] = None,
+        output_attentions: Optional[bool] = False,
+    ) -> Tuple[torch.Tensor]:
+        mixed_query_layer = self.query(hidden_states)
+
+        # If this is instantiated as a cross-attention module, the keys
+        # and values come from an encoder; the attention mask needs to be
+        # such that the encoder's padding tokens are not attended to.
+        is_cross_attention = encoder_hidden_states is not None
+
+        if is_cross_attention and past_key_value is not None:
+            # reuse k,v, cross_attentions
+            key_layer = past_key_value[0]
+            value_layer = past_key_value[1]
+            attention_mask = encoder_attention_mask
+        elif is_cross_attention:
+            key_layer = self.transpose_for_scores(self.key(encoder_hidden_states))
+            value_layer = self.transpose_for_scores(self.value(encoder_hidden_states))
+            attention_mask = encoder_attention_mask
+        elif past_key_value is not None:
+            key_layer = self.transpose_for_scores(self.key(hidden_states))
+            value_layer = self.transpose_for_scores(self.value(hidden_states))
+            key_layer = torch.cat([past_key_value[0], key_layer], dim=2)
+            value_layer = torch.cat([past_key_value[1], value_layer], dim=2)
+        else:
+            key_layer = self.transpose_for_scores(self.key(hidden_states))
+            value_layer = self.transpose_for_scores(self.value(hidden_states))
+
+        query_layer = self.transpose_for_scores(mixed_query_layer)
+
+        use_cache = past_key_value is not None
+        if self.is_decoder:
+            # if cross_attention save Tuple(torch.Tensor, torch.Tensor) of all cross attention key/value_states.
+            # Further calls to cross_attention layer can then reuse all cross-attention
+            # key/value_states (first "if" case)
+            # if uni-directional self-attention (decoder) save Tuple(torch.Tensor, torch.Tensor) of
+            # all previous decoder key/value_states. Further calls to uni-directional self-attention
+            # can concat previous decoder key/value_states to current projected key/value_states (third "elif" case)
+            # if encoder bi-directional self-attention `past_key_value` is always `None`
+            past_key_value = (key_layer, value_layer)
+
+        if self.flash:
+            # query_layer, key_layer shape: (batch_size, num_heads, seq_len, head_size)
+            # value_layer shape: (batch_size, num_heads, seq_len, head_size)
+            query_layer = query_layer.permute(0, 2, 1, 3).contiguous()  # (batch_size, seq_len, num_heads, head_size)
+            key_layer = key_layer.permute(0, 2, 1, 3).contiguous()  # type: ignore[attr-defined]
+            value_layer = value_layer.permute(0, 2, 1, 3).contiguous()  # type: ignore[attr-defined]
+            p = self.dropout.p if self.training else 0.0
+            context_layer = flash_attn_func(query_layer, key_layer, value_layer, dropout_p=p).transpose(1, 2)
+        else:
+            # Take the dot product between "query" and "key" to get the raw attention scores.
+            attention_scores = torch.matmul(query_layer, key_layer.transpose(-1, -2))  # type: ignore[attr-defined]
+
+            if self.position_embedding_type == "relative_key" or self.position_embedding_type == "relative_key_query":
+                raise NotImplementedError
+                query_length, key_length = query_layer.shape[2], key_layer.shape[2]
+                if use_cache:
+                    position_ids_l = torch.tensor(key_length - 1, dtype=torch.long, device=hidden_states.device).view(
+                        -1, 1
+                    )
+                else:
+                    position_ids_l = torch.arange(query_length, dtype=torch.long, device=hidden_states.device).view(
+                        -1, 1
+                    )
+                position_ids_r = torch.arange(key_length, dtype=torch.long, device=hidden_states.device).view(1, -1)
+                distance = position_ids_l - position_ids_r
+
+                positional_embedding = self.distance_embedding(distance + self.max_position_embeddings - 1)
+                positional_embedding = positional_embedding.to(dtype=query_layer.dtype)  # fp16 compatibility
+
+                if self.position_embedding_type == "relative_key":
+                    relative_position_scores = torch.einsum("bhld,lrd->bhlr", query_layer, positional_embedding)
+                    attention_scores = attention_scores + relative_position_scores
+                elif self.position_embedding_type == "relative_key_query":
+                    relative_position_scores_query = torch.einsum("bhld,lrd->bhlr", query_layer, positional_embedding)
+                    relative_position_scores_key = torch.einsum("bhrd,lrd->bhlr", key_layer, positional_embedding)
+                    attention_scores = attention_scores + relative_position_scores_query + relative_position_scores_key
+
+            attention_scores = attention_scores / math.sqrt(self.attention_head_size)
+            if attention_mask is not None:
+                # Apply the attention mask is (precomputed for all layers in SpliceBertModel forward() function)
+                attention_scores = attention_scores + attention_mask
+
+            # Normalize the attention scores to probabilities.
+            attention_probs = nn.functional.softmax(attention_scores, dim=-1)
+
+            # This is actually dropping out entire tokens to attend to, which might
+            # seem a bit unusual, but is taken from the original Transformer paper.
+            attention_probs = self.dropout(attention_probs)
+
+            # Mask heads if we want to
+            if head_mask is not None:
+                attention_probs = attention_probs * head_mask
+
+            context_layer = torch.matmul(attention_probs, value_layer)
+
+        context_layer = context_layer.permute(0, 2, 1, 3).contiguous()
+        new_context_layer_shape = context_layer.size()[:-2] + (self.all_head_size,)
+        context_layer = context_layer.view(new_context_layer_shape)
+
+        outputs = (context_layer, attention_probs) if output_attentions else (context_layer,)
+
+        if self.is_decoder:
+            outputs = outputs + (past_key_value,)
+        return outputs
+
+
+class SpliceBertSelfOutput(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.dense = nn.Linear(config.hidden_size, config.hidden_size)
+        self.layer_norm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps)
+        self.dropout = nn.Dropout(config.hidden_dropout)
+
+    def forward(self, hidden_states: torch.Tensor, input_tensor: torch.Tensor) -> torch.Tensor:
+        hidden_states = self.dense(hidden_states)
+        hidden_states = self.dropout(hidden_states)
+        hidden_states = self.layer_norm(hidden_states + input_tensor)
+        return hidden_states
+
+
+class SpliceBertIntermediate(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.dense = nn.Linear(config.hidden_size, config.intermediate_size)
+        if isinstance(config.hidden_act, str):
+            self.intermediate_act_fn = ACT2FN[config.hidden_act]
+        else:
+            self.intermediate_act_fn = config.hidden_act
+
+    def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
+        hidden_states = self.dense(hidden_states)
+        hidden_states = self.intermediate_act_fn(hidden_states)
+        return hidden_states
+
+
+class SpliceBertOutput(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.dense = nn.Linear(config.intermediate_size, config.hidden_size)
+        self.layer_norm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps)
+        self.dropout = nn.Dropout(config.hidden_dropout)
+
+    def forward(self, hidden_states: torch.Tensor, input_tensor: torch.Tensor) -> torch.Tensor:
+        hidden_states = self.dense(hidden_states)
+        hidden_states = self.dropout(hidden_states)
+        hidden_states = self.layer_norm(hidden_states + input_tensor)
+        return hidden_states
+
+
+class SpliceBertPooler(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.dense = nn.Linear(config.hidden_size, config.hidden_size)
+        self.activation = nn.Tanh()
+
+    def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
+        # We "pool" the model by simply taking the hidden state corresponding
+        # to the first token.
+        first_token_tensor = hidden_states[:, 0]
+        pooled_output = self.dense(first_token_tensor)
+        pooled_output = self.activation(pooled_output)
+        return pooled_output
+
+
+class SpliceBertOnlyMLMHead(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.predictions = SpliceBertLMPredictionHead(config)
+
+    def forward(self, sequence_output: torch.Tensor) -> torch.Tensor:
+        prediction_scores = self.predictions(sequence_output)
+        return prediction_scores
+
+
+class SpliceBertLMPredictionHead(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.transform = SpliceBertPredictionHeadTransform(config)
+
+        # The output weights are the same as the input embeddings, but there is
+        # an output-only bias for each token.
+        self.decoder = nn.Linear(config.hidden_size, config.vocab_size, bias=False)
+
+        self.bias = nn.Parameter(torch.zeros(config.vocab_size))
+
+        # Need a link between the two variables so that the bias is correctly resized with `resize_token_embeddings`
+        self.decoder.bias = self.bias
+
+    def forward(self, hidden_states):
+        hidden_states = self.transform(hidden_states)
+        hidden_states = self.decoder(hidden_states)
+        return hidden_states
+
+
+class SpliceBertPredictionHeadTransform(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.dense = nn.Linear(config.hidden_size, config.hidden_size)
+        if isinstance(config.hidden_act, str):
+            self.transform_act_fn = ACT2FN[config.hidden_act]
+        else:
+            self.transform_act_fn = config.hidden_act
+        self.layer_norm = nn.LayerNorm(config.hidden_size, eps=config.layer_norm_eps)
+
+    def forward(self, hidden_states: torch.Tensor) -> torch.Tensor:
+        hidden_states = self.dense(hidden_states)
+        hidden_states = self.transform_act_fn(hidden_states)
+        hidden_states = self.layer_norm(hidden_states)
+        return hidden_states
+
+
+class SpliceBertPreTrainingHeads(nn.Module):
+    def __init__(self, config):
+        super().__init__()
+        self.predictions = SpliceBertLMPredictionHead(config)
+        self.seq_relationship = nn.Linear(config.hidden_size, 2)
+
+    def forward(self, sequence_output, pooled_output):
+        prediction_scores = self.predictions(sequence_output)
+        seq_relationship_score = self.seq_relationship(pooled_output)
+        return prediction_scores, seq_relationship_score