upload code model

.gitattributes

@@ -0,0 +1 @@
+*.safetensors filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,5 @@
+---
+license: apache-2.0
+---
+
+Post-Training Full models on code task based on LLaDA-8B-Instruct for the paper Principled RL for Diffusion LLMs Emerges from a Sequence-Level Perspective

config.json

@@ -0,0 +1,55 @@
+{
+  "activation_type": "silu",
+  "alibi": false,
+  "alibi_bias_max": 8.0,
+  "architectures": [
+    "LLaDAModelLM"
+  ],
+  "attention_dropout": 0.0,
+  "attention_layer_norm": false,
+  "attention_layer_norm_with_affine": true,
+  "auto_map": {
+    "AutoConfig": "configuration_llada.LLaDAConfig",
+    "AutoModel": "modeling_llada.LLaDAModelLM",
+    "AutoModelForCausalLM": "modeling_llada.LLaDAModelLM"
+  },
+  "bias_for_layer_norm": false,
+  "block_group_size": 1,
+  "block_type": "llama",
+  "d_model": 4096,
+  "embedding_dropout": 0.0,
+  "embedding_size": 126464,
+  "eos_token_id": 126081,
+  "flash_attention": false,
+  "include_bias": false,
+  "include_qkv_bias": false,
+  "init_cutoff_factor": null,
+  "init_device": "meta",
+  "init_fn": "mitchell",
+  "init_std": 0.02,
+  "input_emb_norm": false,
+  "layer_norm_type": "rms",
+  "layer_norm_with_affine": true,
+  "mask_token_id": 126336,
+  "max_sequence_length": 4096,
+  "mlp_hidden_size": 12288,
+  "mlp_ratio": 4,
+  "model_type": "llada",
+  "multi_query_attention": null,
+  "n_heads": 32,
+  "n_kv_heads": 32,
+  "n_layers": 32,
+  "pad_token_id": 126081,
+  "precision": "amp_bf16",
+  "residual_dropout": 0.0,
+  "rms_norm_eps": 1e-05,
+  "rope": true,
+  "rope_full_precision": true,
+  "rope_theta": 500000.0,
+  "scale_logits": false,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.52.4",
+  "use_cache": false,
+  "vocab_size": 126464,
+  "weight_tying": false
+}

configuration_llada.py

@@ -0,0 +1,463 @@
+"""
+LLaDA configuration
+"""
+from transformers import AutoConfig, PretrainedConfig
+
+from enum import Enum
+from os import PathLike
+from typing import Union
+from dataclasses import asdict, dataclass, field
+from glob import glob
+from pathlib import Path
+from typing import (
+    Any,
+    Dict,
+    Iterable,
+    List,
+    Optional,
+    Tuple,
+    Type,
+    TypeVar,
+    Union,
+    cast,
+)
+
+
+__all__ = [
+    "ActivationType",
+    "ActivationCheckpointingStrategy",
+    "BlockType",
+    "LayerNormType",
+    "InitFnType",
+    "ModelConfig",
+]
+
+PathOrStr = Union[str, PathLike]
+
+
+class StrEnum(str, Enum):
+    """
+    This is equivalent to Python's :class:`enum.StrEnum` since version 3.11.
+    We include this here for compatibility with older version of Python.
+    """
+
+    def __str__(self) -> str:
+        return self.value
+
+    def __repr__(self) -> str:
+        return f"'{str(self)}'"
+
+
+class LayerNormType(StrEnum):
+    default = "default"
+    """
+    The default LayerNorm implementation, equivalent to PyTorch's built-in version.
+    """
+
+    low_precision = "low_precision"
+    """
+    A low-precision version of the default LayerNorm.
+    """
+
+    rms = "rms"
+    """
+    An RMSNorm implementation. When using ``torch.compile`` this is
+    probably the fastest implementation.
+    """
+
+    gemma_rms = "gemma_rms"
+    """
+    An RMSNorm implementation by gemmma. When using ``torch.compile`` this is
+    probably the fastest implementation.
+    """
+
+    amd_compatible = "amd_compatible"
+    """
+    LayerNorm implemented manually to work around an issue with ROCm.
+    """
+
+
+class ActivationType(StrEnum):
+    gelu = "gelu"
+    relu = "relu"
+    silu = "silu"
+    swiglu = "swiglu"
+
+
+class BlockType(StrEnum):
+    sequential = "sequential"
+    parallel = "parallel"
+
+    llama = "llama"
+    """
+    A block similar to the sequential block with slightly different
+    implementations of operations like attention to imitate the behavior of Llama.
+    """
+
+
+class InitFnType(StrEnum):
+    mitchell = "mitchell"
+    """
+    The strategy suggested to us by Mitchell Wortsman from UW.
+    This uses a truncated normal distribution with an adaptive standard deviation that depends
+    on the size of the weights as well as the depth of the layer.
+    """
+
+    normal = "normal"
+    """
+    All weights are initialized from the same normal distribution.
+    """
+
+    kaiming_normal = "kaiming_normal"
+    """
+    All weights are initialized with the Kaiming method from a normal distribution.
+    Note this currently won't work with FSDP.
+    """
+
+    fan_in = "fan_in"
+    """
+    "Fan-in variance scaling", i.e. normal with a standard deviation of ``1/sqrt(d_in)`` where ``d_in``
+    is the input dimensionality of the kernel.
+    """
+
+    full_megatron = "full_megatron"
+    """
+    This is what metaseq calls "full megatron init". It is the init used for Llama 2.
+    """
+
+
+@dataclass
+class ModelConfig():
+    """
+    LLaDA (model) configuration.
+    """
+
+    # Note that the defaults for these attributes are equivalent to the base GPT2 model.
+
+    d_model: int = 768
+    """
+    The hidden size of the model.
+    """
+
+    n_heads: int = 12
+    """
+    The number of self-attention heads.
+    """
+
+    n_kv_heads: Optional[int] = None
+    """
+    The number of heads to use for keys and values. Defaults to `n_heads`.
+    Set this to ``None`` or ``n_heads`` for normal multi-head attention.
+    Set this to 1 for multi-query attention.
+    Set it to some in-between value for Llama2-style grouped query attention.
+    """
+
+    n_layers: int = 12
+    """
+    The number of layers/blocks.
+    """
+
+    mlp_ratio: int = 4
+    """
+    The ratio of the inner MLP dimensionality to ``d_model``.
+    This is only used when ``mlp_hidden_size`` is not set.
+    """
+
+    mlp_hidden_size: Optional[int] = None
+    """
+    Set the exact hidden size for the MLP. Otherwise the inner MLP hidden size will be set to `mlp_ratio * d_model`.
+    """
+
+    activation_type: ActivationType = ActivationType.swiglu
+    """
+    The activation function to use within the MLP layers.
+    """
+
+    block_type: BlockType = BlockType.sequential
+    """
+    The transformer block implementation.
+    """
+
+    block_group_size: int = 1
+    """
+    The number of blocks to group together into a single parent block.
+    This has no affect on the number of parameters in the model and is only used to wrap groups
+    of blocks together with a single FSDP wrapper during training.
+    """
+
+    alibi: bool = False
+    """
+    If ``True``, use ALiBi embeddings. Mutually exclusive with ``rope``.
+    """
+
+    alibi_bias_max: float = 8.0
+    """
+    Maximum absolute value of ALiBi bias.
+    """
+
+    rope: bool = False
+    """
+    Use rotary positional embeddings (RoPE). Mutually exclusive with ``alibi``.
+    """
+
+    rope_full_precision: bool = True
+    """
+    If ``True``, apply RoPE embeddings at full precision regardless of the input type. Otherwise,
+    apply RoPE at the precision of the input.
+    """
+
+    flash_attention: bool = False
+    """
+    If ``True``, use ``FlashAttention``.
+    """
+
+    attention_dropout: float = 0.1
+    """
+    The dropout probability within the attention modules.
+    """
+
+    multi_query_attention: Optional[bool] = None
+    """
+    Use the Multi-Query formulation of attention used in PaLM. This reduces the number of parameters
+    and is more efficient during inference.
+    """
+
+    attention_layer_norm: bool = False
+    """
+    Apply layer norm to the keys and queries within the attention mechanism.
+    This can help stabilize training.
+    """
+
+    residual_dropout: float = 0.1
+    """
+    The dropout probability for the MLP and attention output within each block.
+    """
+
+    embedding_dropout: float = 0.1
+    """
+    The dropout probability for embeddings.
+    """
+
+    input_emb_norm: bool = False
+    """
+    An input hidden_states norm implementation by gemmma.
+    """
+
+    layer_norm_type: LayerNormType = LayerNormType.default
+    """
+    The layernorm implementation to use.
+    """
+
+    layer_norm_with_affine: bool = True
+    """
+    Whether to include bias and weight parameters for the layer norms.
+    This only affects layer norms that are immediately followed by a linear layer in the forward pass,
+    so everything except QK-norms. To turn off affines for QK norms as well, set :attr:`attention_layer_norm_with_affine`
+    to ``False``.
+    """
+
+    rms_norm_eps: float = 1e-05
+    """
+    The rms layernorm eps param.
+    """
+
+    attention_layer_norm_with_affine: bool = True
+    """
+    Toggle affine transform for the QK norms.
+    """
+
+    max_sequence_length: int = 1024
+    """
+    The maximum input sequence length supported by the model.
+    """
+
+    rope_theta: float = 10000.0
+    """
+    The rope base param.
+    """
+
+    include_qkv_bias: Optional[bool] = False
+    """
+    Whether or not to include bias parameters in qkv linear layers.
+    """
+
+    include_bias: bool = False
+    """
+    Whether or not to include bias parameters in linear layers.
+    In PaLM, they got rid of all bias terms because they found that large
+    models tend to have near 0 bias terms anyway.
+    """
+
+    bias_for_layer_norm: Optional[bool] = None
+    """
+    Whether or not to include bias parameters in layer norm.
+    This is separate from the include_bias parameter, because of a ROCm crash when biases are disabled in
+    layer norm.
+    When this is None (the default), it inherits the setting from include_bias.
+    """
+
+    scale_logits: bool = False
+    """
+    If ``True``, scale the output logits by ``1 / sqrt(d_model)``.
+    """
+
+    vocab_size: int = 50257
+    """
+    Vocabulary size of the model.
+    """
+
+    embedding_size: Optional[int] = 50304
+    """
+    The number of embeddings, i.e. the number of tokens. If set to ``None`` it will default
+    to ``vocab_size``. If ``vocab_size`` is not a multiple of 128, setting this to the
+    next multiple of 128 that's greater than ``vocab_size`` can improve throughput
+    substantially.
+    """
+
+    weight_tying: bool = True
+    """
+    Whether to tie output linear weights to the input embedding.
+    """
+
+    eos_token_id: int = 50256
+    """
+    The ID of the end-of-sentence special token.
+    """
+
+    pad_token_id: int = 50256
+    """
+    The ID of the token to use for padding. Defaults to the ID of the EOS token.
+    """
+
+    mask_token_id: Optional[int] = 50256
+    """
+    The ID of the token to use for mask token. Defaults to the ID of the EOS token.
+    """
+
+    init_device: Optional[str] = None
+    """
+    The torch device to use when initializing the model parameters, e.g. "cpu", "cuda:0", "meta".
+    """
+
+    init_fn: InitFnType = InitFnType.normal
+    """
+    The weight initialization strategy.
+    """
+
+    init_std: float = 0.02
+    """
+    The standard deviation to use when initializing weights with a "fixed distribution" ``init_fn``, such
+    as "normal".
+    """
+
+    init_cutoff_factor: Optional[float] = None
+    """
+    A positive factor used to scale the cutoff values when initializing weights with a "fixed distribution" ``init_fn``, such
+    as "normal". Setting this to None means values are not cutoff.
+    """
+
+    precision: Optional[str] = None
+    """
+    Precision used to train/evaluate with. You shouldn't set this directly.
+    See :data:`TrainConfig.precision` instead.
+    """
+
+    @property
+    def effective_n_kv_heads(self) -> int:
+        if self.n_kv_heads is None:
+            if self.multi_query_attention is True:
+                return 1
+            else:
+                return self.n_heads
+        else:
+            if self.multi_query_attention is None:
+                return self.n_kv_heads
+            if self.multi_query_attention:
+                n_kv_heads_should_be = 1
+            else:
+                n_kv_heads_should_be = self.n_heads
+            if self.n_kv_heads == n_kv_heads_should_be:
+                return n_kv_heads_should_be
+            else:
+                raise Exception(
+                    "You can't set `multi_query_attention` and `n_kv_heads` at the same time."
+                )
+
+class ActivationCheckpointingStrategy(StrEnum):
+    whole_layer = "whole_layer"
+    """
+    Checkpoint every transformer layer.
+    """
+
+    one_in_two = "one_in_two"
+    """
+    Checkpoint one in two transformer layers.
+    """
+
+    one_in_three = "one_in_three"
+    """
+    Checkpoint one in three transformer layers.
+    """
+
+    one_in_four = "one_in_four"
+    """
+    Checkpoint one in four transformer layers.
+    """
+    
+    two_in_three = "two_in_three"
+    """
+    Checkpoint two out of every three transformer layers.
+    """
+
+    three_in_four = "three_in_four"
+    """
+    Checkpoint three out of four of every transformer layers.
+    """
+
+    four_in_five = "four_in_five"
+    """
+    Checkpoint four out of five of every transformer layers.
+    """
+
+    nine_in_ten = "nine_in_ten"
+    """
+    Checkpoint nine out of ten of every transformer layers.
+    """
+
+    fine_grained = "fine_grained"
+    """
+    Focus checkpointing on where it is cheap to recompute and saves most memory.
+    """
+
+
+class LLaDAConfig(PretrainedConfig):
+    model_type = "llada"
+    keys_to_ignore_at_inference = ["past_key_values"]  # TODO: confirm
+
+    def __init__(self, use_cache: bool = False, **kwargs):
+        model_config = ModelConfig()
+        all_kwargs = model_config.__dict__
+        all_kwargs.update(kwargs)
+        all_kwargs.update({"use_cache": use_cache})
+        all_kwargs.update(
+            {
+                "architectures": all_kwargs.get("architectures", ["LLaDAModelLM"])
+            }
+        )
+        super().__init__(**all_kwargs)
+
+    @property
+    def num_attention_heads(self):
+        return self.n_heads
+
+    @property
+    def num_hidden_layers(self):
+        return self.n_layers
+
+    @property
+    def hidden_size(self):
+        return self.d_model
+
+
+# Register the config class so that it is available for transformer pipelines, auto-loading etc.
+AutoConfig.register("llada", LLaDAConfig)

generation_config.json

@@ -0,0 +1,6 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 126080,
+  "eos_token_id": 126081,
+  "transformers_version": "4.52.4"
+}

model-00001-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:02cfb6407a937165b6cdbd7d07bca4a2f19120cbff1d9f13692db83556cf332c
+size 4995581672

model-00002-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f7b5bffa735d6199981358c6fb3158d89cffb50480f2854cb5b00bb73fc92d0a
+size 4999819552

model-00003-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8af05b84d02d6a4fcf0a521b94f12c8abf2fc03b89f77dee9d8057060b0aab8a
+size 4999802728

model-00004-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91475a1e5d06f7e264c1943bdd91a8435067a9df122f25b2859b4c4024c43e41
+size 1035993232

model.safetensors.index.json

@@ -0,0 +1,298 @@
+{
+  "metadata": {
+    "total_size": 16031162368
+  },
+  "weight_map": {
+    "model.transformer.blocks.0.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.0.attn_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.0.ff_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.0.ff_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.0.ff_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.0.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.0.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.0.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.0.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.1.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.1.attn_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.1.ff_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.1.ff_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.1.ff_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.1.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.1.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.1.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.1.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.10.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.10.attn_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.10.ff_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.10.ff_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.10.ff_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.10.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.10.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.10.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.10.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.11.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.11.attn_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.11.ff_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.11.ff_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.11.ff_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.11.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.11.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.11.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.11.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.12.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.12.attn_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.12.ff_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.12.ff_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.12.ff_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.12.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.12.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.12.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.12.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.13.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.13.attn_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.13.ff_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.13.ff_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.13.ff_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.13.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.13.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.13.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.13.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.14.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.14.attn_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.14.ff_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.14.ff_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.14.ff_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.14.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.14.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.14.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.14.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.15.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.15.attn_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.15.ff_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.15.ff_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.15.ff_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.15.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.15.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.15.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.15.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.16.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.16.attn_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.16.ff_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.16.ff_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.16.ff_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.16.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.16.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.16.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.16.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.17.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.17.attn_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.17.ff_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.17.ff_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.17.ff_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.17.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.17.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.17.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.17.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.18.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.18.attn_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.18.ff_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.18.ff_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.18.ff_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.18.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.18.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.18.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.18.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.19.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.19.attn_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.19.ff_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.19.ff_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.19.ff_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.19.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.19.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.19.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.19.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.2.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.2.attn_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.2.ff_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.2.ff_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.2.ff_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.2.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.2.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.2.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.2.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.20.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.20.attn_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.20.ff_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.20.ff_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.20.ff_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.20.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.20.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.20.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.20.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.21.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.21.attn_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.21.ff_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.21.ff_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.21.ff_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.21.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.21.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.21.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.21.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.22.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.22.attn_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.22.ff_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.22.ff_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.22.ff_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.22.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.22.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.22.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.22.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.23.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.23.attn_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.23.ff_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.23.ff_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.23.ff_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.23.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.23.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.23.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.23.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.24.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.24.attn_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.24.ff_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.24.ff_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.24.ff_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.24.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.24.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.24.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.24.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.25.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.25.attn_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.25.ff_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.25.ff_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.25.ff_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.25.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.25.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.25.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.25.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.26.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.26.attn_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.26.ff_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.26.ff_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.26.ff_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.26.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.26.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.26.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.26.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.27.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.27.attn_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.27.ff_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.27.ff_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.27.ff_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.27.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.27.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.27.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.27.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.28.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.28.attn_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.28.ff_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.28.ff_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.28.ff_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.28.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.28.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.28.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.28.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.29.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.29.attn_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.29.ff_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.29.ff_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.29.ff_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.29.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.29.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.29.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.29.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.3.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.3.attn_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.3.ff_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.3.ff_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.3.ff_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.3.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.3.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.3.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.3.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.30.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.30.attn_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.30.ff_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.30.ff_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.30.ff_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.30.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.30.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.30.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.30.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.31.attn_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.31.attn_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.31.ff_norm.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.31.ff_out.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.31.ff_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.31.k_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.31.q_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.31.up_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.31.v_proj.weight": "model-00003-of-00004.safetensors",
+    "model.transformer.blocks.4.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.4.attn_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.4.ff_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.4.ff_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.4.ff_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.4.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.4.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.4.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.4.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.5.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.5.attn_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.5.ff_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.5.ff_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.5.ff_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.5.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.5.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.5.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.5.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.6.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.6.attn_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.6.ff_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.6.ff_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.6.ff_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.6.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.6.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.6.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.6.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.7.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.7.attn_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.7.ff_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.7.ff_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.7.ff_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.7.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.7.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.7.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.7.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.8.attn_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.8.attn_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.8.ff_norm.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.8.ff_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.8.ff_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.8.k_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.8.q_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.8.up_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.8.v_proj.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.9.attn_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.9.attn_out.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.blocks.9.ff_norm.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.9.ff_out.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.9.ff_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.9.k_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.9.q_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.9.up_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.blocks.9.v_proj.weight": "model-00002-of-00004.safetensors",
+    "model.transformer.ff_out.weight": "model-00004-of-00004.safetensors",
+    "model.transformer.ln_f.weight": "model-00001-of-00004.safetensors",
+    "model.transformer.wte.weight": "model-00001-of-00004.safetensors"
+  }
+}

modeling_llada.py

@@ -0,0 +1,1506 @@
+from __future__ import annotations
+
+import logging
+import math
+import sys
+from abc import abstractmethod
+from collections import defaultdict
+from functools import partial
+from typing import (
+    Callable,
+    Dict,
+    Iterable,
+    List,
+    NamedTuple,
+    Optional,
+    Sequence,
+    Set,
+    Tuple,
+    cast,
+)
+from dataclasses import fields
+from typing import List, Optional, Tuple, Union
+
+import torch
+import torch.backends.cuda
+import torch.nn as nn
+import torch.nn.functional as F
+from torch import einsum
+from transformers import PreTrainedModel
+from transformers.modeling_outputs import CausalLMOutputWithPast
+from transformers.models.auto import AutoModel
+from transformers.cache_utils import Cache
+
+from .configuration_llada import (
+    LLaDAConfig,
+    StrEnum,
+    InitFnType,
+    ActivationType,
+    BlockType,
+    LayerNormType,
+    ModelConfig,
+    ActivationCheckpointingStrategy,
+)
+
+if sys.version_info.minor > 8:
+    from collections.abc import MutableMapping
+elif sys.version_info.minor == 8:
+    from typing import MutableMapping
+else:
+    raise SystemExit("This script supports Python 3.8 or higher")
+
+__all__ = [
+    "LayerNormBase",
+    "LayerNorm",
+    "RMSLayerNorm",
+    "GemmaRMSLayerNorm",
+    "RotaryEmbedding",
+    "Activation",
+    "GELU",
+    "ReLU",
+    "SwiGLU",
+    "LLaDABlock",
+    "LLaDASequentialBlock",
+    "LLaDAModel",
+    "LLaDAOutput",
+    "LLaDAGenerateOutput",
+]
+
+
+log = logging.getLogger(__name__)
+
+
+class ModuleType(StrEnum):
+    in_module = "in"
+    out_module = "out"
+    emb = "emb"
+    final_out = "final_out"
+
+
+def init_weights(
+    config: ModelConfig,
+    module: Union[nn.Linear, nn.Embedding],
+    d: Optional[int] = None,
+    layer_id: Optional[int] = None,
+    std_factor: float = 1.0,
+    type_of_module: Optional[ModuleType] = None,
+) -> None:
+    """
+    Initialize weights of a linear or embedding module.
+
+    :param config: The model config.
+    :param module: The linear or embedding submodule to initialize.
+    :param d: The effective input dimensionality of the weights. This could be smaller than the actual dimensions
+        for fused layers.
+    :param layer_id: When set, the standard deviation for the "mitchell" method will be adjusted by
+        ``1 / sqrt(2 * (layer_id + 1))``.
+    """
+    d = d if d is not None else config.d_model
+    if config.init_fn == InitFnType.normal:
+        std = config.init_std * std_factor
+        if config.init_cutoff_factor is not None:
+            cutoff_value = config.init_cutoff_factor * std
+            nn.init.trunc_normal_(module.weight, mean=0.0, std=std, a=-cutoff_value, b=cutoff_value)
+        else:
+            nn.init.normal_(module.weight, mean=0.0, std=std)
+    elif config.init_fn == InitFnType.mitchell:
+        std = std_factor / math.sqrt(d)
+        if layer_id is not None:
+            std = std / math.sqrt(2 * (layer_id + 1))
+        nn.init.trunc_normal_(module.weight, mean=0.0, std=std, a=-3 * std, b=3 * std)
+    elif config.init_fn == InitFnType.kaiming_normal:
+        nn.init.kaiming_normal_(module.weight, nonlinearity="relu")
+    elif config.init_fn == InitFnType.fan_in:
+        std = std_factor / math.sqrt(d)
+        nn.init.normal_(module.weight, mean=0.0, std=std)
+    elif config.init_fn == InitFnType.full_megatron:
+        if type_of_module is None:
+            raise RuntimeError(f"When using the {InitFnType.full_megatron} init, every module must have a type.")
+
+        cutoff_factor = config.init_cutoff_factor
+        if cutoff_factor is None:
+            cutoff_factor = 3
+
+        if type_of_module == ModuleType.in_module:
+            # for att_proj (same as QKV), ff_proj
+            std = config.init_std
+        elif type_of_module == ModuleType.out_module:
+            # for attn_out, ff_out
+            std = config.init_std / math.sqrt(2.0 * config.n_layers)
+        elif type_of_module == ModuleType.emb:
+            # positional embeddings (wpe)
+            # token embeddings (wte)
+            std = config.init_std
+        elif type_of_module == ModuleType.final_out:
+            # final output (ff_out)
+            std = config.d_model**-0.5
+        else:
+            raise RuntimeError(f"Unknown module type '{type_of_module}'")
+        nn.init.trunc_normal_(
+            module.weight,
+            mean=0.0,
+            std=std,
+            a=-cutoff_factor * std,
+            b=cutoff_factor * std,
+        )
+    else:
+        raise NotImplementedError(config.init_fn)
+
+    if isinstance(module, nn.Linear):
+        if module.bias is not None:
+            nn.init.zeros_(module.bias)
+
+        if config.init_fn == InitFnType.normal and getattr(module, "_is_residual", False):
+            with torch.no_grad():
+                module.weight.div_(math.sqrt(2 * config.n_layers))
+
+
+def ensure_finite_(x: torch.Tensor, check_neg_inf: bool = True, check_pos_inf: bool = False):
+    """
+    Modify ``x`` in place to replace ``float("-inf")`` with the minimum value of the dtype when ``check_neg_inf``
+    is ``True`` and to replace ``float("inf")`` with the maximum value of the dtype when ``check_pos_inf`` is ``True``.
+    """
+    if check_neg_inf:
+        x.masked_fill_(x == float("-inf"), torch.finfo(x.dtype).min)
+    if check_pos_inf:
+        x.masked_fill_(x == float("inf"), torch.finfo(x.dtype).max)
+
+
+def activation_checkpoint_function(cfg: ModelConfig):
+    preserve_rng_state = (
+        (cfg.attention_dropout == 0.0) and (cfg.embedding_dropout == 0.0) and (cfg.residual_dropout == 0.0)
+    )
+    from torch.utils.checkpoint import checkpoint
+
+    return partial(
+        checkpoint,
+        preserve_rng_state=preserve_rng_state,
+        use_reentrant=False,
+    )
+
+
+class BufferCache(dict, MutableMapping[str, torch.Tensor]):
+    """
+    Cache for attention biases and other things that would normally be stored as buffers.
+    We avoid using buffers because we've run into various issues doing so with FSDP.
+    In general it appears the way FSDP handles buffers is not well-defined.
+    It doesn't shard them but apparently it does synchronize them across processes, which we want to avoid
+    since (A) it isn't necessary, and (B) we sometimes have `-inf` in these biases which might get turned into
+    NaNs when they're synchronized due to casting or some other issue.
+    """
+
+
+def _non_meta_init_device(config: ModelConfig) -> torch.device:
+    if config.init_device is not None and config.init_device != "meta":
+        return torch.device(config.init_device)
+    else:
+        return torch.device("cuda" if torch.cuda.is_available() else "cpu")
+
+
+class Dropout(nn.Dropout):
+    def forward(self, input: torch.Tensor) -> torch.Tensor:
+        if self.p == 0.0:
+            return input
+        else:
+            return F.dropout(input, self.p, self.training, self.inplace)
+
+
+class LayerNormBase(nn.Module):
+    def __init__(
+        self,
+        config: ModelConfig,
+        *,
+        size: Optional[int] = None,
+        elementwise_affine: Optional[bool] = True,
+        eps: float = 1e-05,
+    ):
+        super().__init__()
+        self.config = config
+        self.eps = eps
+        self.normalized_shape = (size or config.d_model,)
+        if elementwise_affine or (elementwise_affine is None and self.config.layer_norm_with_affine):
+            self.weight = nn.Parameter(torch.ones(self.normalized_shape, device=config.init_device))
+            use_bias = self.config.bias_for_layer_norm
+            if use_bias is None:
+                use_bias = self.config.include_bias
+            if use_bias:
+                self.bias = nn.Parameter(torch.zeros(self.normalized_shape, device=config.init_device))
+            else:
+                self.register_parameter("bias", None)
+        else:
+            self.register_parameter("bias", None)
+            self.register_parameter("weight", None)
+
+    @abstractmethod
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        raise NotImplementedError
+
+    @classmethod
+    def build(cls, config: ModelConfig, size: Optional[int] = None, **kwargs) -> LayerNormBase:
+        if config.layer_norm_type == LayerNormType.default:
+            return LayerNorm(config, size=size, low_precision=False, **kwargs)
+        elif config.layer_norm_type == LayerNormType.low_precision:
+            return LayerNorm(config, size=size, low_precision=True, **kwargs)
+        elif config.layer_norm_type == LayerNormType.rms:
+            return RMSLayerNorm(config, size=size, **kwargs)
+        elif config.layer_norm_type == LayerNormType.gemma_rms:
+            return GemmaRMSLayerNorm(config, size=size, **kwargs)
+        else:
+            raise NotImplementedError(f"Unknown LayerNorm type: '{config.layer_norm_type}'")
+
+    def _cast_if_autocast_enabled(self, tensor: torch.Tensor, dtype: Optional[torch.dtype] = None) -> torch.Tensor:
+        # NOTE: `is_autocast_enabled()` only checks for CUDA autocast, so we use the separate function
+        # `is_autocast_cpu_enabled()` for CPU autocast.
+        # See https://github.com/pytorch/pytorch/issues/110966.
+        if tensor.device.type == "cuda" and torch.is_autocast_enabled():
+            return tensor.to(dtype=dtype if dtype is not None else torch.get_autocast_gpu_dtype())
+        elif tensor.device.type == "cpu" and torch.is_autocast_cpu_enabled():
+            return tensor.to(dtype=dtype if dtype is not None else torch.get_autocast_cpu_dtype())
+        else:
+            return tensor
+
+    def reset_parameters(self):
+        if self.weight is not None:
+            torch.nn.init.ones_(self.weight)  # type: ignore
+        if self.bias is not None:
+            torch.nn.init.zeros_(self.bias)  # type: ignore
+
+
+class LayerNorm(LayerNormBase):
+    """
+    The default :class:`LayerNorm` implementation which can optionally run in low precision.
+    """
+
+    def __init__(
+        self,
+        config: ModelConfig,
+        size: Optional[int] = None,
+        low_precision: bool = False,
+        elementwise_affine: Optional[bool] = None,
+        eps: float = 1e-05,
+    ):
+        super().__init__(config, size=size, elementwise_affine=elementwise_affine, eps=eps)
+        self.low_precision = low_precision
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        if self.low_precision:
+            module_device = x.device
+            downcast_x = self._cast_if_autocast_enabled(x)
+            downcast_weight = (
+                self._cast_if_autocast_enabled(self.weight) if self.weight is not None else self.weight
+            )
+            downcast_bias = self._cast_if_autocast_enabled(self.bias) if self.bias is not None else self.bias
+            with torch.autocast(enabled=False, device_type=module_device.type):
+                return F.layer_norm(
+                    downcast_x, self.normalized_shape, weight=downcast_weight, bias=downcast_bias, eps=self.eps
+                )
+        else:
+            return F.layer_norm(x, self.normalized_shape, weight=self.weight, bias=self.bias, eps=self.eps)
+
+
+class RMSLayerNorm(LayerNormBase):
+    """
+    RMS layer norm, a simplified :class:`LayerNorm` implementation
+    """
+
+    def __init__(
+        self,
+        config: ModelConfig,
+        size: Optional[int] = None,
+        elementwise_affine: Optional[bool] = None,
+        eps: float = 1e-5,
+    ):
+        super().__init__(config, size=size, elementwise_affine=elementwise_affine, eps=config.rms_norm_eps)
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        with torch.autocast(enabled=False, device_type=x.device.type):
+            og_dtype = x.dtype
+            x = x.to(torch.float32)
+            variance = x.pow(2).mean(-1, keepdim=True)
+            x = x * torch.rsqrt(variance + self.eps)
+            x = x.to(og_dtype)
+
+        if self.weight is not None:
+            if self.bias is not None:
+                return self.weight * x + self.bias
+            else:
+                return self.weight * x
+        else:
+            return x
+
+
+class GemmaRMSLayerNorm(LayerNormBase):
+    """
+    Gemma RMS layer norm, a simplified :class:`LayerNorm` implementation
+    """
+
+    def __init__(
+        self,
+        config: ModelConfig,
+        size: Optional[int] = None,
+        elementwise_affine: Optional[bool] = None,
+        eps: float = 1e-5,
+    ):
+        super().__init__(config, size=size, elementwise_affine=elementwise_affine, eps=config.rms_norm_eps)
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        with torch.autocast(enabled=False, device_type=x.device.type):
+            og_dtype = x.dtype
+            x = x.to(torch.float32)
+            variance = x.pow(2).mean(-1, keepdim=True)
+            x = x * torch.rsqrt(variance + self.eps)
+            x = x.to(og_dtype)
+
+        if self.weight is not None:
+            if self.bias is not None:
+                return x * (1 + self.weight) + self.bias
+            else:
+                return x * (1 + self.weight)
+        else:
+            return x
+
+
+class RotaryEmbedding(nn.Module):
+    """
+    [Rotary positional embeddings (RoPE)](https://arxiv.org/abs/2104.09864).
+    """
+
+    def __init__(self, config: ModelConfig, cache: BufferCache):
+        super().__init__()
+        self.config = config
+        self.__cache = cache
+        # Warm up cache.
+        self.rope_theta = config.rope_theta
+        self.get_rotary_embedding(config.max_sequence_length, _non_meta_init_device(config))
+
+    def get_rotary_embedding(self, seq_len: int, device: torch.device) -> Tuple[torch.Tensor, torch.Tensor]:
+        if (
+            (pos_sin := self.__cache.get("rope_pos_sin")) is not None
+            and (pos_cos := self.__cache.get("rope_pos_cos")) is not None
+            and pos_sin.shape[-2] >= seq_len
+            and pos_cos.shape[-2] >= seq_len
+        ):
+            if pos_sin.device != device:
+                pos_sin = pos_sin.to(device)
+                self.__cache["rope_pos_sin"] = pos_sin
+            if pos_cos.device != device:
+                pos_cos = pos_cos.to(device)
+                self.__cache["rope_pos_cos"] = pos_cos
+            return pos_sin[:, :, :seq_len, :], pos_cos[:, :, :seq_len, :]
+
+        with torch.autocast(device.type, enabled=False):
+            dim = self.config.d_model // self.config.n_heads
+            inv_freq = 1.0 / (self.rope_theta ** (torch.arange(0, dim, 2, device=device, dtype=torch.float) / dim))
+            seq = torch.arange(seq_len, device=device, dtype=torch.float)
+            freqs = einsum("i , j -> i j", seq, inv_freq)
+            positions = torch.cat((freqs, freqs), dim=-1)
+            pos_sin, pos_cos = positions.sin()[None, None, :, :], positions.cos()[None, None, :, :]
+        self.__cache["rope_pos_sin"] = pos_sin
+        self.__cache["rope_pos_cos"] = pos_cos
+        return pos_sin, pos_cos
+
+    def rotate_half(self, x: torch.Tensor) -> torch.Tensor:
+        B, nh, T, hs = x.size()
+        x = x.view(B, nh, T, 2, hs // 2)
+        x1, x2 = x.unbind(dim=-2)
+        return torch.cat((-x2, x1), dim=-1)
+
+    def apply_rotary_pos_emb(self, pos_sin: torch.Tensor, pos_cos: torch.Tensor, t: torch.Tensor) -> torch.Tensor:
+        return ((t * pos_cos) + (self.rotate_half(t) * pos_sin)).to(t.dtype)
+
+    def forward(self, q: torch.Tensor, k: torch.Tensor) -> Tuple[torch.Tensor, torch.Tensor]:
+        if self.config.rope_full_precision:
+            q_, k_ = q.float(), k.float()
+        else:
+            q_, k_ = q, k
+
+        with torch.autocast(q.device.type, enabled=False):
+            query_len, key_len = q_.shape[-2], k_.shape[-2]  # could be different if layer_past not None
+            pos_sin, pos_cos = self.get_rotary_embedding(key_len, q_.device)
+            pos_sin = pos_sin.type_as(q_)
+            pos_cos = pos_cos.type_as(q_)
+            q_ = self.apply_rotary_pos_emb(
+                pos_sin[:, :, key_len - query_len : key_len, :],
+                pos_cos[:, :, key_len - query_len : key_len, :],
+                q_,
+            )
+            k_ = self.apply_rotary_pos_emb(pos_sin, pos_cos, k_)
+        return q_.type_as(q), k_.type_as(k)
+
+
+class Activation(nn.Module):
+    def __init__(self, config: ModelConfig):
+        super().__init__()
+        self.config = config
+
+    @abstractmethod
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        raise NotImplementedError
+
+    @property
+    @abstractmethod
+    def output_multiplier(self) -> float:
+        raise NotImplementedError
+
+    @classmethod
+    def build(cls, config: ModelConfig) -> Activation:
+        if config.activation_type == ActivationType.gelu:
+            return cast(Activation, GELU(approximate="none"))
+        elif config.activation_type == ActivationType.relu:
+            return cast(Activation, ReLU(inplace=False))
+        elif config.activation_type == ActivationType.silu:
+            return cast(Activation, SiLU(inplace=False))
+        elif config.activation_type == ActivationType.swiglu:
+            return SwiGLU(config)
+        else:
+            raise NotImplementedError(f"Unknown activation: '{config.activation_type}'")
+
+
+class GELU(nn.GELU):
+    @property
+    def output_multiplier(self) -> float:
+        return 1.0
+
+
+class ReLU(nn.ReLU):
+    @property
+    def output_multiplier(self) -> float:
+        return 1.0
+
+class SiLU(nn.SiLU):
+    @property
+    def output_multiplier(self) -> float:
+        return 1.0
+
+class SwiGLU(Activation):
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        x, gate = x.chunk(2, dim=-1)
+        return F.silu(gate) * x
+
+    @property
+    def output_multiplier(self) -> float:
+        return 0.5
+
+
+def causal_attention_bias(seq_len: int, device: torch.device) -> torch.FloatTensor:
+    att_bias = torch.triu(
+        torch.ones(seq_len, seq_len, device=device, dtype=torch.float),
+        diagonal=1,
+    )
+    att_bias.masked_fill_(att_bias == 1, torch.finfo(att_bias.dtype).min)
+    return att_bias.view(1, 1, seq_len, seq_len)  # type: ignore
+
+
+def get_causal_attention_bias(cache: BufferCache, seq_len: int, device: torch.device) -> torch.Tensor:
+    if (causal_bias := cache.get("causal_attention_bias")) is not None and causal_bias.shape[-1] >= seq_len:
+        if causal_bias.device != device:
+            causal_bias = causal_bias.to(device)
+            cache["causal_attention_bias"] = causal_bias
+        return causal_bias
+    with torch.autocast(device.type, enabled=False):
+        causal_bias = causal_attention_bias(seq_len, device)
+    cache["causal_attention_bias"] = causal_bias
+    return causal_bias
+
+
+def alibi_attention_bias(seq_len: int, config: ModelConfig, device: torch.device) -> torch.FloatTensor:
+    alibi_bias = torch.arange(1 - seq_len, 1, dtype=torch.float, device=device).view(1, 1, 1, seq_len)
+
+    # shape: (1, 1, seq_len, seq_len)
+    alibi_bias = alibi_bias - torch.arange(1 - seq_len, 1, dtype=torch.float, device=device).view(1, 1, seq_len, 1)
+    alibi_bias.abs_().mul_(-1)
+
+    # shape: (n_heads,)
+    m = torch.arange(1, config.n_heads + 1, dtype=torch.float, device=device)
+    m.mul_(config.alibi_bias_max / config.n_heads)
+
+    # shape: (1, n_heads, seq_len, seq_len)
+    return alibi_bias * (1.0 / (2 ** m.view(1, config.n_heads, 1, 1)))  # type: ignore
+
+
+class LLaDABlock(nn.Module):
+    """
+    A base class for transformer block implementations.
+    """
+
+    def __init__(self, layer_id: int, config: ModelConfig, cache: BufferCache):
+        super().__init__()
+        self.layer_id = layer_id
+        self.config = config
+        self.hidden_size = (
+            config.mlp_hidden_size if config.mlp_hidden_size is not None else config.mlp_ratio * config.d_model
+        )
+        self.__cache = cache
+        assert config.d_model % config.n_heads == 0
+
+        self._activation_checkpoint_fn = None
+
+        # Dropout.
+        self.dropout = Dropout(config.residual_dropout)
+
+        # Layer norms.
+        self.k_norm: Optional[LayerNormBase] = None
+        self.q_norm: Optional[LayerNormBase] = None
+        if config.attention_layer_norm:
+            self.k_norm = LayerNormBase.build(
+                config,
+                size=(config.d_model // config.n_heads) * config.effective_n_kv_heads,
+                elementwise_affine=config.attention_layer_norm_with_affine,
+            )
+            self.q_norm = LayerNormBase.build(config, elementwise_affine=config.attention_layer_norm_with_affine)
+
+        # Activation function.
+        self.act = Activation.build(config)
+        assert (self.act.output_multiplier * self.hidden_size) % 1 == 0
+
+        # Attention output projection.
+        self.attn_out = nn.Linear(
+            config.d_model, config.d_model, bias=config.include_bias, device=config.init_device
+        )
+
+        # Feed-forward output projection.
+        self.ff_out = nn.Linear(
+            int(self.act.output_multiplier * self.hidden_size),
+            config.d_model,
+            bias=config.include_bias,
+            device=config.init_device,
+        )
+        self.ff_out._is_residual = True  # type: ignore
+
+        # Rotary embeddings.
+        if self.config.rope:
+            self.rotary_emb = RotaryEmbedding(config, self.__cache)
+
+        self.flash_attn_func = None
+        if config.flash_attention:
+            try:
+                from flash_attn import flash_attn_func  # type: ignore
+
+                self.flash_attn_func = flash_attn_func
+            except ModuleNotFoundError:
+                pass
+
+    def reset_parameters(self):
+        if self.k_norm is not None:
+            self.k_norm.reset_parameters()
+        if self.q_norm is not None:
+            self.q_norm.reset_parameters()
+        init_weights(
+            self.config,
+            self.attn_out,
+            d=self.config.d_model,
+            layer_id=self.layer_id,
+            type_of_module=ModuleType.out_module,
+        )
+        init_weights(
+            self.config,
+            self.ff_out,
+            d=self.ff_out.in_features,
+            layer_id=self.layer_id,
+            type_of_module=ModuleType.out_module,
+        )
+
+    def set_activation_checkpointing(self, strategy: Optional[ActivationCheckpointingStrategy]):
+        if strategy == ActivationCheckpointingStrategy.fine_grained:
+            self._activation_checkpoint_fn = activation_checkpoint_function(self.config)
+        else:
+            self._activation_checkpoint_fn = None
+
+    @classmethod
+    def _cast_attn_bias(cls, bias: torch.Tensor, input_dtype: torch.dtype) -> torch.Tensor:
+        target_dtype = input_dtype
+        # NOTE: `is_autocast_enabled()` only checks for CUDA autocast, so we use the separate function
+        # `is_autocast_cpu_enabled()` for CPU autocast.
+        # See https://github.com/pytorch/pytorch/issues/110966.
+        if bias.device.type == "cuda" and torch.is_autocast_enabled():
+            target_dtype = torch.get_autocast_gpu_dtype()
+        elif bias.device.type == "cpu" and torch.is_autocast_cpu_enabled():
+            target_dtype = torch.get_autocast_cpu_dtype()
+        if bias.dtype != target_dtype:
+            bias = bias.to(target_dtype)
+            ensure_finite_(bias, check_neg_inf=True, check_pos_inf=False)
+        return bias
+
+    def _scaled_dot_product_attention(
+        self,
+        q: torch.Tensor,
+        k: torch.Tensor,
+        v: torch.Tensor,
+        attn_mask: Optional[torch.Tensor] = None,
+        dropout_p: float = 0.0,
+        is_causal: bool = False,
+    ) -> torch.Tensor:
+        """
+        Computes scaled dot product attention on query, key and value tensors, using an optional
+        attention mask if passed, and applying dropout if a probability greater than 0.0 is specified.
+        """
+        if self.flash_attn_func is not None and attn_mask is None:
+            r = self.flash_attn_func(
+                q.transpose(1, 2), k.transpose(1, 2), v.transpose(1, 2), dropout_p=dropout_p, causal=False
+            )
+            return r.transpose(1, 2)
+        else:
+            # torch's sdpa doesn't support GQA, so we're doing this
+            assert k.size(1) == v.size(1)
+            num_kv_heads = k.size(1)
+            num_q_heads = q.size(1)
+            if num_q_heads != num_kv_heads:
+                assert num_q_heads % num_kv_heads == 0
+                k = k.repeat_interleave(num_q_heads // num_kv_heads, dim=1, output_size=num_q_heads)
+                v = v.repeat_interleave(num_q_heads // num_kv_heads, dim=1, output_size=num_q_heads)
+
+            # Modify: MDM set causal to False.
+            return F.scaled_dot_product_attention(
+                q,
+                k,
+                v,
+                attn_mask=attn_mask,
+                dropout_p=dropout_p,
+                is_causal=False,
+            )
+
+    def attention(
+        self,
+        q: torch.Tensor,
+        k: torch.Tensor,
+        v: torch.Tensor,
+        attention_bias: Optional[torch.Tensor] = None,
+        layer_past: Optional[Tuple[torch.Tensor, torch.Tensor]] = None,
+        use_cache: bool = False,
+    ) -> Tuple[torch.Tensor, Optional[Tuple[torch.Tensor, torch.Tensor]]]:
+        B, T, C = q.size()  # batch size, sequence length, d_model
+        dtype = k.dtype
+
+        # Optionally apply layer norm to keys and queries.
+        if self.q_norm is not None and self.k_norm is not None:
+            q = self.q_norm(q).to(dtype=dtype)
+            k = self.k_norm(k).to(dtype=dtype)
+
+        # Move head forward to be next to the batch dim.
+        # shape: (B, nh, T, hs)
+        q = q.view(B, T, self.config.n_heads, C // self.config.n_heads).transpose(1, 2)
+        # shape: (B, n_kv_h, T, hs)
+        k = k.view(B, T, self.config.effective_n_kv_heads, C // self.config.n_heads).transpose(1, 2)
+        # shape: (B, n_kv_h, T, hs)
+        v = v.view(B, T, self.config.effective_n_kv_heads, C // self.config.n_heads).transpose(1, 2)
+
+        if layer_past is not None:
+            past_key, past_value = layer_past
+            k = torch.cat((past_key, k), dim=-2)
+            v = torch.cat((past_value, v), dim=-2)
+
+        present = (k, v) if use_cache else None
+        query_len, key_len = q.shape[-2], k.shape[-2]  # could be different if layer_past not None
+
+        if self.config.rope:
+            # Apply rotary embeddings.
+            q, k = self.rotary_emb(q, k)
+
+        if attention_bias is not None:
+            # Resize and cast attention bias.
+            # The current dtype of the attention bias might not match the dtype that the SDP attn function will
+            # run in if AMP is enabled, and this can be a problem if some tokens are masked out due to padding
+            # as down-casting the attention bias to the autocast precision will result in -infs, which will
+            # cause the SDP attn function to produce NaNs.
+            attention_bias = self._cast_attn_bias(
+                attention_bias[:, :, key_len - query_len : key_len, :key_len], dtype
+            )
+
+        # Get the attention scores.
+        # shape: (B, nh, T, hs)
+        att = self._scaled_dot_product_attention(
+            q,
+            k,
+            v,
+            attn_mask=attention_bias,
+            dropout_p=0.0 if not self.training else self.config.attention_dropout,
+            is_causal=False,
+        )
+
+        # Re-assemble all head outputs side-by-side.
+        att = att.transpose(1, 2).contiguous().view(B, T, C)
+
+        # Apply output projection.
+        return self.attn_out(att), present
+
+    @abstractmethod
+    def forward(
+        self,
+        x: torch.Tensor,
+        attention_bias: Optional[torch.FloatTensor] = None,
+        layer_past: Optional[Tuple[torch.Tensor, torch.Tensor]] = None,
+        use_cache: bool = False,
+    ) -> Tuple[torch.Tensor, Optional[Tuple[torch.Tensor, torch.Tensor]]]:
+        raise NotImplementedError
+
+    @classmethod
+    def build(cls, layer_id: int, config: ModelConfig, cache: BufferCache) -> LLaDABlock:
+        if config.block_type == BlockType.sequential:
+            return LLaDASequentialBlock(layer_id, config, cache)
+        elif config.block_type == BlockType.llama:
+            return LLaDALlamaBlock(layer_id, config, cache)
+        else:
+            raise NotImplementedError(f"Unknown block type: '{config.block_type}'")
+
+
+class LLaDASequentialBlock(LLaDABlock):
+    """
+    This is a typical transformer block where the output is computed as ``MLP(LN(x + Attention(LN(x))))``
+    (plus another skip connection).
+    """
+
+    def __init__(self, layer_id: int, config: ModelConfig, cache: BufferCache):
+        super().__init__(layer_id, config, cache)
+        # Layer norms.
+        self.attn_norm = LayerNorm.build(config)
+        self.ff_norm = LayerNorm.build(config)
+        # Attention input projection. Projects x -> (q, k, v)
+        head_dim = config.d_model // config.n_heads
+        self.fused_dims = (
+            config.d_model,
+            config.effective_n_kv_heads * head_dim,
+            config.effective_n_kv_heads * head_dim,
+        )
+        self.att_proj = nn.Linear(
+            config.d_model, sum(self.fused_dims), bias=config.include_bias | config.include_qkv_bias, device=config.init_device
+        )
+        # Feed-forward input projection.
+        self.ff_proj = nn.Linear(
+            config.d_model, self.hidden_size, bias=config.include_bias, device=config.init_device
+        )
+
+    def reset_parameters(self):
+        super().reset_parameters()
+        self.attn_norm.reset_parameters()
+        self.ff_norm.reset_parameters()
+        # NOTE: the standard deviation for these weights does not depend on the layer.
+        init_weights(
+            self.config, self.att_proj, d=self.config.d_model, layer_id=None, type_of_module=ModuleType.in_module
+        )
+        init_weights(
+            self.config, self.ff_proj, d=self.config.d_model, layer_id=None, type_of_module=ModuleType.in_module
+        )
+
+    def forward(
+        self,
+        x: torch.Tensor,
+        attention_bias: Optional[torch.Tensor] = None,
+        layer_past: Optional[Tuple[torch.Tensor, torch.Tensor]] = None,
+        use_cache: bool = False,
+    ) -> Tuple[torch.Tensor, Optional[Tuple[torch.Tensor, torch.Tensor]]]:
+        # Get query, key, value projections.
+        # shape:
+        #  - for regular attn q, k, v: (batch_size, seq_len, d_model)
+        #  - for multi-query attn q: (batch_size, seq_len, d_model)
+        #                      k, v: (batch_size, seq_len, d_model // n_heads)
+        #  - for group query attn q: (batch_size, seq_len, d_model)
+        #                      k, v: (batch_size, seq_len, d_model // n_kv_heads)
+        if self._activation_checkpoint_fn is not None:
+            q, k, v = self.att_proj(self._activation_checkpoint_fn(self.attn_norm, x)).split(
+                self.fused_dims, dim=-1
+            )
+        else:
+            q, k, v = self.att_proj(self.attn_norm(x)).split(self.fused_dims, dim=-1)
+
+        # Get attention scores.
+        if self._activation_checkpoint_fn is not None:
+            att, cache = self._activation_checkpoint_fn(  # type: ignore
+                self.attention, q, k, v, attention_bias, layer_past=layer_past, use_cache=use_cache
+            )
+        else:
+            att, cache = self.attention(q, k, v, attention_bias, layer_past=layer_past, use_cache=use_cache)
+
+        # Add attention scores.
+        # shape: (B, T, C)
+        x = x + self.dropout(att)
+
+        # Add feed-forward projection.
+        # shape: (batch_size, seq_len, d_model)
+        og_x = x
+        if self._activation_checkpoint_fn is not None:
+            x = self._activation_checkpoint_fn(self.ff_norm, x)  # type: ignore
+        else:
+            x = self.ff_norm(x)
+        x = self.ff_proj(x)
+        if self._activation_checkpoint_fn is not None:
+            x = self._activation_checkpoint_fn(self.act, x)  # type: ignore
+        else:
+            x = self.act(x)
+        x = self.ff_out(x)
+        x = self.dropout(x)
+        x = og_x + x
+
+        return x, cache
+
+
+class LLaDALlamaBlock(LLaDABlock):
+    """
+    This is a transformer block where the output is computed as ``MLP(LN(x + Attention(LN(x))))``
+    (plus another skip connection). This block is similar to `LLaDASequentialBlock`
+    but some operations have slightly different implementations to imitate the
+    behavior of Llama.
+    """
+
+    def __init__(self, layer_id: int, config: ModelConfig, cache: BufferCache):
+        super().__init__(layer_id, config, cache)
+        # Layer norms.
+        self.attn_norm = LayerNorm.build(config)
+        self.ff_norm = LayerNorm.build(config)
+        self.__cache = cache
+
+        # Attention input projection. Projects x -> (q, k, v)
+        head_dim = config.d_model // config.n_heads
+        q_proj_out_dim = config.d_model
+        k_proj_out_dim = config.effective_n_kv_heads * head_dim
+        v_proj_out_dim = config.effective_n_kv_heads * head_dim
+        self.q_proj = nn.Linear(
+            config.d_model, q_proj_out_dim, bias=config.include_bias | config.include_qkv_bias, device=config.init_device
+        )
+        self.k_proj = nn.Linear(
+            config.d_model, k_proj_out_dim, bias=config.include_bias | config.include_qkv_bias, device=config.init_device
+        )
+        self.v_proj = nn.Linear(
+            config.d_model, v_proj_out_dim, bias=config.include_bias | config.include_qkv_bias, device=config.init_device
+        )
+
+        # Feed-forward input projection.
+        self.ff_proj = nn.Linear(
+            config.d_model, self.hidden_size, bias=config.include_bias, device=config.init_device
+        )
+        # new add
+        self.up_proj = nn.Linear(
+            config.d_model, self.hidden_size, bias=config.include_bias, device=config.init_device
+        )
+
+    def reset_parameters(self):
+        super().reset_parameters()
+        self.attn_norm.reset_parameters()
+        self.ff_norm.reset_parameters()
+        # NOTE: the standard deviation for these weights does not depend on the layer.
+        init_weights(self.config, self.q_proj, d=self.config.d_model, layer_id=None)
+        init_weights(self.config, self.k_proj, d=self.config.d_model, layer_id=None)
+        init_weights(self.config, self.v_proj, d=self.config.d_model, layer_id=None)
+        init_weights(self.config, self.ff_proj, d=self.config.d_model, layer_id=None)
+        init_weights(self.config, self.up_proj, d=self.config.d_model, layer_id=None)  # new add
+
+    def forward(
+        self,
+        x: torch.Tensor,
+        attention_bias: Optional[torch.Tensor] = None,
+        layer_past: Optional[Tuple[torch.Tensor, torch.Tensor]] = None,
+        use_cache: bool = False,
+    ) -> Tuple[torch.Tensor, Optional[Tuple[torch.Tensor, torch.Tensor]]]:
+        # Get query, key, value projections.
+        # shape:
+        #  - for regular attn q, k, v: (batch_size, seq_len, d_model)
+        #  - for multi-query attn q: (batch_size, seq_len, d_model)
+        #                      k, v: (batch_size, seq_len, d_model // n_heads)
+        #  - for group query attn q: (batch_size, seq_len, d_model)
+        #                      k, v: (batch_size, seq_len, d_model // n_kv_heads)
+        x_normed = self.attn_norm(x)
+        q = self.q_proj(x_normed)
+        k = self.k_proj(x_normed)
+        v = self.v_proj(x_normed)
+
+        # Get attention scores.
+        if self._activation_checkpoint_fn is not None:
+            att, cache = self._activation_checkpoint_fn(  # type: ignore
+                self.attention, q, k, v, attention_bias, layer_past=layer_past, use_cache=use_cache
+            )
+        else:
+            att, cache = self.attention(q, k, v, attention_bias, layer_past=layer_past, use_cache=use_cache)
+
+        # Add attention scores.
+        # shape: (B, T, C)
+        x = x + self.dropout(att)
+
+        # Add feed-forward projection.
+        # shape: (batch_size, seq_len, d_model)
+        og_x = x
+        if self._activation_checkpoint_fn is not None:
+            x = self._activation_checkpoint_fn(self.ff_norm, x)  # type: ignore
+        else:
+            x = self.ff_norm(x)
+        x, x_up = self.ff_proj(x), self.up_proj(x) # new add
+        if self._activation_checkpoint_fn is not None:
+            x = self._activation_checkpoint_fn(self.act, x)  # type: ignore
+        else:
+            x = self.act(x)
+        x = x * x_up # new add
+        x = self.ff_out(x)
+        x = self.dropout(x)
+        x = og_x + x
+
+        return x, cache
+
+
+class LLaDAOutput(NamedTuple):
+    logits: torch.FloatTensor
+    """
+    A tensor of shape `(batch_size, seq_len, vocab_size)` representing the log probabilities
+    for the next token *before* normalization via (log) softmax.
+    """
+
+    attn_key_values: Optional[List[Tuple[torch.Tensor, torch.Tensor]]]
+    """
+    Attention keys and values from each block.
+    """
+
+    hidden_states: Optional[Tuple[torch.Tensor]]
+    """
+    Hidden states from each block.
+    """
+
+
+class LLaDAGenerateOutput(NamedTuple):
+    token_ids: torch.LongTensor
+    """
+    The generated token IDs, a tensor of shape `(batch_size, beam_size, max_steps)`.
+    These do *not* include the original input IDs.
+    """
+
+    scores: torch.FloatTensor
+    """
+    The scores of the generated sequences, a tensor of shape `(batch_size, beam_size)`.
+    """
+
+
+class LLaDABlockGroup(nn.ModuleList):
+    def __init__(self, config: ModelConfig, layer_offset: int, modules: Optional[Iterable[nn.Module]] = None):
+        super().__init__(modules)
+        self.config = config
+        self.layer_offset = layer_offset
+        self.activation_checkpointing_strategy: Optional[ActivationCheckpointingStrategy] = None
+        self._activation_checkpoint_fn = activation_checkpoint_function(self.config)
+
+    def forward(
+        self,
+        x: torch.Tensor,
+        attention_bias: Optional[torch.FloatTensor] = None,
+        layers_past: Optional[List[Tuple[torch.Tensor, torch.Tensor]]] = None,
+        use_cache: bool = False,
+    ) -> Tuple[torch.Tensor, Optional[List[Tuple[torch.Tensor, torch.Tensor]]]]:
+        attn_key_values: Optional[List[Tuple[torch.Tensor, torch.Tensor]]] = [] if use_cache else None
+        for block_idx, block in enumerate(self):
+            layer_past = None if layers_past is None else layers_past[block_idx]
+            block_idx += self.layer_offset
+            if (
+                (self.activation_checkpointing_strategy == ActivationCheckpointingStrategy.whole_layer)
+                or (
+                    self.activation_checkpointing_strategy == ActivationCheckpointingStrategy.one_in_two
+                    and block_idx % 2 == 0
+                )
+                or (
+                    self.activation_checkpointing_strategy == ActivationCheckpointingStrategy.one_in_three
+                    and block_idx % 3 == 0
+                )
+                or (
+                    self.activation_checkpointing_strategy == ActivationCheckpointingStrategy.one_in_four
+                    and block_idx % 4 == 0
+                )
+            ):
+                # shape: (batch_size, seq_len, d_model)
+                x, cache = self._activation_checkpoint_fn(  # type: ignore
+                    block, x, attention_bias=attention_bias, layer_past=layer_past, use_cache=use_cache
+                )
+            else:
+                # shape: (batch_size, seq_len, d_model)
+                x, cache = block(x, attention_bias=attention_bias, layer_past=layer_past, use_cache=use_cache)
+            if attn_key_values is not None:
+                assert cache is not None
+                attn_key_values.append(cache)
+        return x, attn_key_values
+
+    def reset_parameters(self):
+        for block in self:
+            block.reset_parameters()
+
+    def set_activation_checkpointing(self, strategy: Optional[ActivationCheckpointingStrategy]):
+        self.activation_checkpointing_strategy = strategy
+        for block in self:
+            block.set_activation_checkpointing(strategy)
+
+
+class LLaDAModel(nn.Module):
+    def __init__(self, config: ModelConfig, init_params: bool = True):
+        super().__init__()
+        self.config = config
+        self.__cache = BufferCache()
+
+        # Validate config.
+        if self.config.alibi and self.config.flash_attention:
+            raise Exception("ALiBi is currently not supported with FlashAttention")
+
+        if self.config.alibi and self.config.rope:
+            raise Exception("ALiBi and RoPE are mutually exclusive")
+
+        if self.config.embedding_size is not None and self.config.embedding_size != self.config.vocab_size:
+            if self.config.embedding_size < self.config.vocab_size:
+                raise Exception("embedding size should be at least as big as vocab size")
+            elif self.config.embedding_size % 128 != 0:
+                import warnings
+
+                warnings.warn(
+                    "Embedding size is not a multiple of 128! This could hurt throughput performance.", UserWarning
+                )
+
+        self.activation_checkpointing_strategy: Optional[ActivationCheckpointingStrategy] = None
+        self._activation_checkpoint_fn: Callable = activation_checkpoint_function(self.config)
+
+        if not (
+            0 < self.config.block_group_size <= self.config.n_layers
+            and self.config.n_layers % self.config.block_group_size == 0
+        ):
+            raise Exception("n layers must be divisible by block group size")
+
+        torch.backends.cuda.enable_flash_sdp(True)
+        torch.backends.cuda.enable_mem_efficient_sdp(False)  # this is super slow so make sure torch won't use it
+
+        self.transformer = nn.ModuleDict(
+            dict(
+                wte=nn.Embedding(
+                    config.embedding_size or config.vocab_size, config.d_model, device=config.init_device
+                ),
+                emb_drop=Dropout(config.embedding_dropout),
+                ln_f=LayerNorm.build(config),
+            )
+        )
+
+        blocks = [LLaDABlock.build(i, config, self.__cache) for i in range(config.n_layers)]
+        if self.config.block_group_size > 1:
+            block_groups = [
+                LLaDABlockGroup(config, i, blocks[i : i + config.block_group_size])
+                for i in range(0, config.n_layers, config.block_group_size)
+            ]
+            self.transformer.update({"block_groups": nn.ModuleList(block_groups)})
+        else:
+            self.transformer.update({"blocks": nn.ModuleList(blocks)})
+
+        if not (self.config.alibi or self.config.rope):
+            self.transformer.update(
+                {"wpe": nn.Embedding(config.max_sequence_length, config.d_model, device=config.init_device)}
+            )
+        if not config.weight_tying:
+            self.transformer.update(
+                {
+                    "ff_out": nn.Linear(
+                        config.d_model,
+                        config.embedding_size or config.vocab_size,
+                        bias=config.include_bias,
+                        device=config.init_device,
+                    )
+                }
+            )
+        # When `init_device="meta"` FSDP will call `reset_parameters()` to initialize weights.
+        if init_params and self.config.init_device != "meta":
+            self.reset_parameters()
+        self.__num_fwd_flops: Optional[int] = None
+
+        # Warm up cache.
+        if self.config.alibi:
+            get_causal_attention_bias(self.__cache, config.max_sequence_length, _non_meta_init_device(config))
+            self.get_alibi_attention_bias(config.max_sequence_length, _non_meta_init_device(config))
+
+    def set_activation_checkpointing(self, strategy: Optional[ActivationCheckpointingStrategy]):
+        self.activation_checkpointing_strategy = strategy
+        if self.config.block_group_size != 1:
+            for block_group in self.transformer.block_groups:
+                block_group.set_activation_checkpointing(strategy)
+        else:
+            for block in self.transformer.blocks:
+                block.set_activation_checkpointing(strategy)
+
+    @property
+    def device(self) -> torch.device:
+        device: torch.device = self.transformer.wte.weight.device  # type: ignore
+        if device.type == "meta":
+            return _non_meta_init_device(self.config)
+        else:
+            return device
+
+    def reset_parameters(self):
+        log.info("Initializing model parameters...")
+        # Top-level embeddings / linear layers.
+        init_weights(
+            self.config,
+            self.transformer.wte,  # type: ignore
+            std_factor=(0.5 * math.sqrt(self.config.d_model)) if self.config.scale_logits else 1.0,
+            type_of_module=ModuleType.emb,
+        )
+        if hasattr(self.transformer, "wpe"):
+            init_weights(self.config, self.transformer.wpe, type_of_module=ModuleType.emb)  # type: ignore
+
+        # Top-level layer norm.
+        self.transformer.ln_f.reset_parameters()  # type: ignore
+
+        # Output weights.
+        if hasattr(self.transformer, "ff_out"):
+            init_weights(self.config, self.transformer.ff_out, type_of_module=ModuleType.final_out)  # type: ignore
+
+        # Let the blocks handle themselves.
+        if self.config.block_group_size == 1:
+            for block in self.transformer.blocks:
+                block.reset_parameters()
+        else:
+            for block_group in self.transformer.block_groups:
+                block_group.reset_parameters()
+
+    def get_alibi_attention_bias(self, seq_len: int, device: torch.device) -> torch.Tensor:
+        if (alibi_bias := self.__cache.get("alibi_attention_bias")) is not None and alibi_bias.shape[
+            -1
+        ] >= seq_len:
+            if alibi_bias.device != device:
+                alibi_bias = alibi_bias.to(device)
+                self.__cache["alibi_attention_bias"] = alibi_bias
+            return alibi_bias
+        with torch.autocast(device.type, enabled=False):
+            alibi_bias = alibi_attention_bias(seq_len, self.config, device)
+        self.__cache["alibi_attention_bias"] = alibi_bias
+        return alibi_bias
+    
+    def get_bidirectional_attention_bias(self, seq_len: int, device: torch.device) -> torch.Tensor:
+        if (bidirectional_bias := self.__cache.get("bidirectional_attention_bias")) is not None and bidirectional_bias.shape[
+            -1
+        ] >= seq_len:
+            if bidirectional_bias.device != device:
+                bidirectional_bias = bidirectional_bias.to(device)
+                self.__cache["bidirectional_attention_bias"] = bidirectional_bias
+            return bidirectional_bias
+        with torch.autocast(device.type, enabled=False):
+            bidirectional_bias = torch.zeros((1, 1, seq_len, seq_len), device=device, dtype=torch.float)
+        self.__cache["bidirectional_attention_bias"] = bidirectional_bias
+        return bidirectional_bias
+    
+    def forward(
+        self,
+        input_ids: torch.LongTensor,
+        input_embeddings: Optional[torch.FloatTensor] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        attention_bias: Optional[torch.Tensor] = None,
+        past_key_values: Optional[Sequence[Tuple[torch.Tensor, torch.Tensor]]] = None,
+        use_cache: bool = False,
+        last_logits_only: bool = False,
+        output_hidden_states: Optional[bool] = None,
+    ) -> LLaDAOutput:
+        """
+        :param input_ids: A tensor of shape `(batch_size, seq_len)`.
+        :param input_embeddings: A tensor of shape `(batch_size, seq_len, d_model)` with input
+            embeddings. When provided, it is treated as the output of the input embedding layer.
+        :param attention_mask: A tensor of shape `(batch_size, seq_len)` that indicates
+            which input IDs are masked. A `1` value in the mask means that
+            the corresponding input ID should *not* be ignored. A `0` means
+            that the corresponding input ID is masked.
+
+            This has the same meaning as the `attention_mask` in HuggingFace's `transformers`
+            library.
+        :param attention_bias: A tensor of shape `(batch_size, 1, seq_len, seq_len)`,
+            `(1, 1, seq_len, seq_len)`, or `(seq_len, seq_len)`. This is used
+            to introduce causal or other biases.
+
+            If the tensor is a bool or byte tensor, a `True` or `1` at `attention_bias[:, :, i, j]`
+            indicates that the i-th element in the sequence is allowed to attend to the j-th
+            element in the sequence.
+
+            If the tensor is a float tensor, it will just be added to the attention
+            scores before the softmax.
+
+            The default is causal, which corresponds to a lower-diagonal byte matrix of ones.
+        :param past_key_values: Pre-computed keys and values for each attention block.
+            Can be used to speed up sequential decoding. The `input_ids` which have
+            their past given to this model should not be passed as `input_ids` as they have already been computed.
+        :param use_cache: If `True`, return key and value tensors for each block.
+        :param last_logits_only: If `True`, only compute the logits for the last token of each sequence.
+            This can speed up decoding when you only care about the next token.
+        """
+        # Add Basic MDM Model config check
+        assert not self.config.alibi, "Alibi length extrapolation is not supported for MDM."
+        assert self.config.rope, "Rope must be used in Llama-Encoder for MDM."
+        assert (past_key_values is None and not use_cache), "The kvcache is not suppotred for MDM."
+
+        output_hidden_states = output_hidden_states if output_hidden_states is not None else False
+
+        if past_key_values:
+            assert len(past_key_values) == self.config.n_layers
+
+        batch_size, seq_len = input_ids.size() if input_embeddings is None else input_embeddings.size()[:2]
+        if past_key_values is None:
+            past_length = 0
+        else:
+            past_length = past_key_values[0][0].size(-2)
+
+        # Get embeddings of input.
+        # shape: (batch_size, seq_len, d_model)
+        x = self.transformer.wte(input_ids) if input_embeddings is None else input_embeddings  # type: ignore
+
+        if self.config.input_emb_norm:
+            x = x * (self.config.d_model**0.5)
+
+        if not (self.config.alibi or self.config.rope):
+            # Get positional embeddings.
+            # shape: (1, seq_len)
+            pos = torch.arange(past_length, past_length + seq_len, dtype=torch.long, device=x.device).unsqueeze(0)
+            # shape: (1, seq_len, d_model)
+            pos_emb = self.transformer.wpe(pos)  # type: ignore
+            x = pos_emb + x
+
+        # Add input + positional embeddings and apply dropout.
+        # shape: (batch_size, seq_len, d_model)
+        x = self.transformer.emb_drop(x)  # type: ignore
+
+        # Transform the attention mask into what the blocks expect.
+        if attention_mask is not None and 0.0 in attention_mask:
+            # shape: (batch_size, 1, 1, seq_len)
+            attention_mask = attention_mask.to(dtype=torch.float).view(batch_size, -1)[:, None, None, :]
+            attention_mask = (1.0 - attention_mask) * torch.finfo(attention_mask.dtype).min
+        else:
+            attention_mask = None
+
+        # Merge attention mask with attention bias.
+        if (
+            attention_bias is not None
+            or attention_mask is not None
+            or self.config.alibi
+            # NOTE (epwalsh): we need to initialize the attn bias in order for attn to work properly
+            # with key+value cache. Otherwise `F.scaled_dot_product_attention()` doesn't seem to compute
+            # scores correctly.
+            or past_key_values is not None
+        ):
+            if attention_bias is None and self.config.alibi:
+                attention_bias = get_causal_attention_bias(
+                    self.__cache, past_length + seq_len, x.device
+                ) + self.get_alibi_attention_bias(past_length + seq_len, x.device)
+            elif attention_bias is None:
+                attention_bias = self.get_bidirectional_attention_bias(past_length + seq_len, x.device)
+            elif attention_bias.dtype in (torch.int8, torch.bool):
+                attention_bias = attention_bias.to(dtype=torch.float)
+                attention_bias.masked_fill_(attention_bias == 0.0, torch.finfo(attention_bias.dtype).min)
+
+            # Transform to the right shape and data type.
+            mask_len = seq_len
+            if attention_mask is not None:
+                mask_len = attention_mask.shape[-1]
+            elif past_key_values is not None:
+                mask_len = past_key_values[0][0].shape[-2] + seq_len
+            attention_bias = attention_bias[:, :, :mask_len, :mask_len].to(dtype=torch.float)
+
+            # Add in the masking bias.
+            if attention_mask is not None:
+                attention_bias = attention_bias + attention_mask
+                # Might get -infs after adding attention mask, since dtype.min + dtype.min = -inf.
+                # `F.scaled_dot_product_attention()` doesn't handle -inf like you'd expect, instead
+                # it can produce NaNs.
+                ensure_finite_(attention_bias, check_neg_inf=True, check_pos_inf=False)
+
+        attn_key_values: Optional[List[Tuple[torch.Tensor, torch.Tensor]]] = [] if use_cache else None
+
+        # decoder layers
+        all_hidden_states = []
+
+        # Apply blocks one-by-one.
+        if self.config.block_group_size == 1:
+            for block_idx, block in enumerate(self.transformer.blocks):
+                if output_hidden_states:
+                    # add hidden states
+                    all_hidden_states.append(x)
+
+                layer_past = None if past_key_values is None else past_key_values[block_idx]
+                if (
+                    (self.activation_checkpointing_strategy == ActivationCheckpointingStrategy.whole_layer)
+                    or (
+                        self.activation_checkpointing_strategy == ActivationCheckpointingStrategy.one_in_two
+                        and block_idx % 2 == 0
+                    )
+                    or (
+                        self.activation_checkpointing_strategy == ActivationCheckpointingStrategy.one_in_three
+                        and block_idx % 3 == 0
+                    )
+                    or (
+                        self.activation_checkpointing_strategy == ActivationCheckpointingStrategy.one_in_four
+                        and block_idx % 4 == 0
+                    )
+                ):
+                    # shape: (batch_size, seq_len, d_model)
+                    x, cache = self._activation_checkpoint_fn(
+                        block, x, attention_bias=attention_bias, layer_past=layer_past, use_cache=use_cache
+                    )
+                else:
+                    # shape: (batch_size, seq_len, d_model)
+                    x, cache = block(x, attention_bias=attention_bias, layer_past=layer_past, use_cache=use_cache)
+                if attn_key_values is not None:
+                    assert cache is not None
+                    attn_key_values.append(cache)
+        else:
+            for group_idx, block_group in enumerate(self.transformer.block_groups):
+                if output_hidden_states:
+                    # add hidden states
+                    all_hidden_states.append(x)
+
+                layers_past = (
+                    None
+                    if past_key_values is None
+                    else past_key_values[
+                        group_idx * self.config.block_group_size : (group_idx + 1) * self.config.block_group_size
+                    ]
+                )
+                x, cache = block_group(
+                    x, attention_bias=attention_bias, layers_past=layers_past, use_cache=use_cache
+                )
+                if attn_key_values is not None:
+                    assert cache is not None
+                    attn_key_values.extend(cache)
+
+        if last_logits_only:
+            # shape: (batch_size, 1, d_model)
+            x = x[:, -1, :].unsqueeze(1)
+
+        # Apply final layer norm.
+        # shape: (batch_size, seq_len or 1, d_model)
+        x = self.transformer.ln_f(x)  # type: ignore
+        if output_hidden_states:
+            # add final hidden state post-final-layernorm, following HuggingFace's convention
+            all_hidden_states.append(x)
+
+        # Get logits.
+        # shape: (batch_size, seq_len or 1, vocab_size)
+        if self.config.weight_tying:
+            logits = F.linear(x, self.transformer.wte.weight, None)  # type: ignore
+        else:
+            logits = self.transformer.ff_out(x)  # type: ignore
+        if self.config.scale_logits:
+            logits.mul_(1 / math.sqrt(self.config.d_model))
+
+        return LLaDAOutput(logits=logits, attn_key_values=attn_key_values, hidden_states=tuple(all_hidden_states) if output_hidden_states else None)  # type: ignore[arg-type]
+
+
+def create_model_config_from_pretrained_config(config: LLaDAConfig):
+    """
+    Utility function
+    """
+
+    kwargs = {}
+    for field in fields(ModelConfig):
+        kwargs[field.name] = getattr(config, field.name)
+
+    model_config = ModelConfig(**kwargs)
+    return model_config
+
+
+class LLaDAModelLM(PreTrainedModel):
+    """
+    Extremely barebones HF model wrapper.
+    """
+
+    config_class = LLaDAConfig
+    base_model_prefix = "model"
+    _no_split_modules = ["LLaDABlock", "LLaDASequentialBlock", "LLaDALlamaBlock"]
+
+    def __init__(self, config: LLaDAConfig, model: Optional[LLaDAModel] = None, init_params: bool = False):
+        super().__init__(config)
+
+        if not model:
+            model_config = create_model_config_from_pretrained_config(config)
+            # Initialize model (always on CPU to start with so we don't run out of GPU memory).
+            model_config.init_device = "cpu"
+            self.model = LLaDAModel(model_config, init_params=init_params)
+        else:
+            self.model = model
+
+    def forward(
+        self,
+        input_ids: torch.LongTensor = None,
+        inputs_embeds: Optional[torch.FloatTensor] = None,
+        attention_mask: Optional[torch.Tensor] = None,
+        attention_bias: Optional[torch.Tensor] = None,
+        past_key_values: Optional[List[torch.FloatTensor]] = None,
+        labels: Optional[torch.LongTensor] = None,
+        use_cache: Optional[bool] = None,
+        output_attentions: Optional[bool] = None,
+        output_hidden_states: Optional[bool] = None,
+        return_dict: Optional[bool] = None,
+        cache_position: Optional[Cache] = None,  # This is a hack mitigation of an issue in transformers `4.39.x`
+    ) -> Union[Tuple, CausalLMOutputWithPast]:
+        if use_cache is None:
+            use_cache = self.config.use_cache
+
+        if output_attentions:
+            raise ValueError("output_attentions is not yet supported in LLaDA")
+
+        return_dict = return_dict if return_dict is not None else self.config.use_return_dict
+
+        # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn)
+        outputs = self.model.forward(
+            input_ids=input_ids,
+            input_embeddings=inputs_embeds,
+            attention_mask=attention_mask,
+            attention_bias=attention_bias,
+            past_key_values=past_key_values,
+            use_cache=use_cache,
+            output_hidden_states=output_hidden_states,
+        )
+
+        logits = outputs.logits
+        hidden_states = outputs.hidden_states
+
+        loss = None
+        if labels is not None:
+            import warnings
+            warnings.warn("Note that for LLaDA, you cannot calculate the loss here.", UserWarning)
+        if not return_dict:
+            output = (logits,) + outputs[1:]
+            return (loss,) + output if loss is not None else output
+
+        return CausalLMOutputWithPast(
+            logits=logits,
+            past_key_values=outputs.attn_key_values,
+            hidden_states=hidden_states,
+        )
+
+    def can_generate(self) -> bool:
+        return True
+
+    def prepare_inputs_for_generation(
+        self, input_ids: torch.LongTensor, past_key_values: Optional[List[Tuple]] = None, **kwargs
+    ):
+        if past_key_values:
+            # This is because we want the model to only process the last generated token.
+            input_ids = input_ids[:, -1:]
+        model_inputs = {"input_ids": input_ids, "past_key_values": past_key_values}
+
+        model_inputs.update(kwargs)
+        model_inputs["use_cache"] = kwargs.pop("use_cache", self.config.use_cache)
+        return model_inputs
+
+    # TODO: these are required to make the implementation complete.
+    # def resize_position_embeddings(self, new_num_position_embeddings: int):
+    #     pass
+    #
+    # def get_position_embeddings(self) -> Union[nn.Embedding, Tuple[nn.Embedding]]:
+    #     pass
+    #
+    # def _reorder_cache(self, past_key_values, beam_idx):
+    #     pass
+
+    def get_input_embeddings(self) -> torch.nn.Module:
+        return self.model.transformer.wte
+
+    def set_input_embeddings(self, value: torch.nn.Module):
+        self.model.transformer.wte = value
+
+    def get_output_embeddings(self):
+        if self.config.weight_tying:
+            return self.model.transformer.wte
+        else:
+            return self.model.transformer.ff_out
+
+    def set_output_embeddings(self, value: torch.nn.Module):
+        if self.config.weight_tying:
+            self.model.transformer.wte = value
+        else:
+            self.model.transformer.ff_out = value
+
+    def tie_weights(self):
+        if self.config.weight_tying:
+            self.model.transformer.ff_out = self.model.transformer.wte
+
+# Register the model so that it is available for transformer pipelines, auto-loading, etc.
+AutoModel.register(LLaDAConfig, LLaDAModelLM)