# Copyright 2025 The EasyDeL Author @erfanzar (Erfan Zare Chavoshi).
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# https://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import typing as tp
from eformer.common_types import (
EMPTY,
MODE_TRAIN,
TP,
ColumnWise,
DynamicShardingAxes,
Replicated,
RowWise,
)
from easydel.infra.base_module import EasyDeLBaseConfig
from easydel.infra.etils import EasyDeLGradientCheckPointers
from easydel.infra.factory import register_config
from easydel.infra.utils import AttnMaskDetail, AttnMaskType
from easydel.layers.moe.utils import get_moe_partition_spec
[docs]class ExpertTensorParallel(DynamicShardingAxes):
"""Expert Tensor Parallelism (EPxTP) sharding axes."""
axes: tp.ClassVar = [TP, EMPTY, EMPTY]
mode: tp.ClassVar = MODE_TRAIN
[docs]@register_config("mixtral")
class MixtralConfig(EasyDeLBaseConfig):
"""
Configuration objects inherit from [`EasyDeLBaseConfig`] and can be used to control the model outputs. Read
the documentation from [`EasyDeLBaseConfig`] for more information.
Args:
vocab_size (`int`, *optional*, defaults to 32000):
Vocabulary size of the Mixtral model. Defines the number of different tokens that can be represented by the
`inputs_ids` passed to the forward method.
hidden_size (`int`, *optional*, defaults to 4096):
Dimensionality of the encoder layers and the pooler layer.
intermediate_size (`int`, *optional*, defaults to 14336):
Dimensionality of the "intermediate" (i.e., feed-forward) layer in the Transformer encoder.
num_hidden_layers (`int`, *optional*, defaults to 32):
Number of hidden layers in the Transformer encoder.
num_attention_heads (`int`, *optional*, defaults to 32):
Number of attention heads for each attention layer in the Transformer encoder.
num_key_value_heads (`int`, *optional*, defaults to 8):
Number of key and value heads for each attention layer in the Transformer encoder.
hidden_act (`str` or `function`, *optional*, defaults to `"silu"`):
The non-linear activation function (function or string) to use in the encoder and pooler. If string,
`"gelu"`, `"relu"`, `"swish"` and `"gelu_new"` are supported.
max_position_embeddings (`int`, *optional*, defaults to 4096 * 32):
The maximum sequence length that this model might ever be used with. Typically set this to something large
just in case (e.g., 2048 or 4096).
initializer_range (`float`, *optional*, defaults to 0.02):
The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
rms_norm_eps (`float`, *optional*, defaults to 1e-5):
The epsilon used by the rms normalization layers.
use_cache (`bool`, *optional*, defaults to `True`):
Whether or not the model should return the last key/values attentions (not used by all models). Only
relevant if `config.is_decoder=True`.
pad_token_id (`int`, *optional*):
The index of the padding token in the vocabulary.
bos_token_id (`int`, *optional*, defaults to 1):
The id of the *beginning-of-sequence* token.
eos_token_id (`int`, *optional*, defaults to 2):
The id of the *end-of-sequence* token.
tie_word_embeddings (`bool`, *optional*, defaults to `False`):
Whether to tie the weights of the input embeddings and the output embeddings.
rope_theta (`float`, *optional*, defaults to 1e6):
The theta value to use for rotary position embeddings.
sliding_window (`int`, *optional*, defaults to 4096):
The sliding window size.
attention_dropout (`float`, *optional*, defaults to 0.0):
The dropout ratio for the attention probabilities.
num_experts_per_tok (`int`, *optional*, defaults to 2):
The number of experts per token.
num_local_experts (`int`, *optional*, defaults to 8):
The number of local experts.
output_router_logits (`bool`, *optional*, defaults to `False`):
Whether to output router logits.
router_aux_loss_coef (`float`, *optional*, defaults to 0.001):
The router auxiliary loss coefficient.
gradient_checkpointing (`str`, *optional*, defaults to `"nothing_saveable"`):
The gradient checkpointing configuration.
use_scan_mlp (`bool`, *optional*, defaults to `False`):
Whether to use the scan implementation for the MLP.
scan_mlp_chunk_size (`int`, *optional*, defaults to 1024):
The chunk size to use when scanning the MLP.
number_rep_kv (`int`, *optional*, defaults to 1):
Number of repetitions for the key and value vectors.
bits (`int`, *optional*):
The number of bits to quantize the model to.
rope_scaling (`tp.Dict[str, tp.Union[str, float]]`, *optional*):
The configuration for rope scaling.
attention_bias (`bool`, *optional*, defaults to `False`):
Whether to use bias in the attention layer.
initialization_of_moe (`bool`, *optional*, defaults to `False`):
Whether to initialize the MoE layers.
router_jitter_noise (`float`, *optional*, defaults to 0.0):
The jitter noise for the router.
"""
model_type: str = "mixtral"
def __init__(
self,
vocab_size=32000,
hidden_size=4096,
intermediate_size=14336,
num_hidden_layers=32,
num_attention_heads=32,
num_key_value_heads=8,
hidden_act="silu",
max_position_embeddings=4096 * 32,
initializer_range=0.02,
rms_norm_eps=1e-5,
use_cache=True,
pad_token_id=None,
bos_token_id=1,
eos_token_id=2,
tie_word_embeddings=False,
rope_theta=1e6,
sliding_window=4096,
attention_dropout=0.0,
num_experts_per_tok=2,
num_local_experts=8,
output_router_logits=False,
router_aux_loss_coef=0.001,
gradient_checkpointing: EasyDeLGradientCheckPointers = EasyDeLGradientCheckPointers.NONE,
use_scan_mlp: bool = False,
scan_mlp_chunk_size: int = 1024,
number_rep_kv: int = 1,
bits: int | None = None,
rope_scaling: dict[str, str | float] | None = None,
attention_bias: bool = False,
initialization_of_moe: bool = False,
router_jitter_noise=0.0,
head_dim: int | None = None,
layer_types: list[str] | None = None,
**kwargs,
):
"""Initializes a MixtralConfig object.
Args:
vocab_size (int, optional): Vocabulary size. Defaults to 32000.
hidden_size (int, optional): Hidden size. Defaults to 4096.
intermediate_size (int, optional): Intermediate size of the feed-forward network. Defaults to 14336.
num_hidden_layers (int, optional): Number of hidden layers. Defaults to 32.
num_attention_heads (int, optional): Number of attention heads. Defaults to 32.
num_key_value_heads (int, optional): Number of key/value heads (for GQA). Defaults to 8.
hidden_act (str, optional): Activation function. Defaults to "silu".
max_position_embeddings (int, optional): Maximum sequence length. Defaults to 4096 * 32.
initializer_range (float, optional): Initializer range. Defaults to 0.02.
rms_norm_eps (float, optional): Epsilon for RMS normalization. Defaults to 1e-5.
use_cache (bool, optional): Whether to use KV cache. Defaults to True.
pad_token_id (int, optional): Padding token ID. Defaults to None.
bos_token_id (int, optional): Beginning-of-sequence token ID. Defaults to 1.
eos_token_id (int, optional): End-of-sequence token ID. Defaults to 2.
tie_word_embeddings (bool, optional): Whether to tie input/output embeddings. Defaults to False.
rope_theta (float, optional): Base value for RoPE. Defaults to 1e6.
sliding_window (int, optional): Sliding window size for attention. Defaults to 4096.
attention_dropout (float, optional): Dropout probability for attention. Defaults to 0.0.
num_experts_per_tok (int, optional): Number of experts to route per token. Defaults to 2.
num_local_experts (int, optional): Total number of local experts. Defaults to 8.
output_router_logits (bool, optional): Whether to output router logits. Defaults to False.
router_aux_loss_coef (float, optional): Coefficient for router auxiliary loss. Defaults to 0.001.
gradient_checkpointing (EasyDeLGradientCheckPointers, optional): Gradient checkpointing strategy.
Defaults to EasyDeLGradientCheckPointers.NONE.
use_scan_mlp (bool, optional): Whether to use scan for MLP layers. Defaults to False.
scan_mlp_chunk_size (int, optional): Chunk size for scan MLP. Defaults to 1024.
number_rep_kv (int, optional): Number of repetitions for key/value heads. Defaults to 1.
bits (tp.Optional[int], optional): Quantization bits. Defaults to None.
rope_scaling (tp.Dict[str, tp.Union[str, float]], optional): RoPE scaling configuration. Defaults to None.
attention_bias (bool, optional): Whether to use bias in attention layers. Defaults to False.
initialization_of_moe (bool, optional): Whether MoE layers are being initialized. Defaults to False.
router_jitter_noise (float, optional): Jitter noise for router gates. Defaults to 0.0.
**kwargs: Additional keyword arguments.
"""
self.vocab_size = vocab_size
self.max_position_embeddings = max_position_embeddings
self.hidden_size = hidden_size
self.intermediate_size = intermediate_size
self.num_hidden_layers = num_hidden_layers
self.num_attention_heads = num_attention_heads
self.sliding_window = sliding_window
self.bits = bits
self.attention_dropout = attention_dropout
self.num_local_experts = num_local_experts
self.num_experts_per_tok = num_experts_per_tok
self.output_router_logits = output_router_logits
self.router_aux_loss_coef = router_aux_loss_coef
self.attention_bias = attention_bias
# for backward compatibility
self.rope_scaling = rope_scaling
if num_key_value_heads is None:
num_key_value_heads = num_attention_heads
self.num_key_value_heads = num_key_value_heads
self.hidden_act = hidden_act
self.initialization_of_moe = initialization_of_moe
self.initializer_range = initializer_range
self.rms_norm_eps = rms_norm_eps
self.use_cache = use_cache
self.rope_theta = rope_theta
self.number_rep_kv = number_rep_kv
self.gradient_checkpointing = gradient_checkpointing
self.use_scan_mlp = use_scan_mlp
self.scan_mlp_chunk_size = scan_mlp_chunk_size
self.router_jitter_noise = router_jitter_noise
self.layer_types = layer_types
self.head_dim = head_dim or hidden_size // num_attention_heads
if self.layer_types is None:
self.layer_types = [
"sliding_attention" if self.sliding_window is not None else "full_attention"
for i in range(self.num_hidden_layers)
]
super().__init__(
pad_token_id=pad_token_id,
bos_token_id=bos_token_id,
eos_token_id=eos_token_id,
tie_word_embeddings=tie_word_embeddings,
use_scan_mlp=use_scan_mlp,
scan_mlp_chunk_size=scan_mlp_chunk_size,
bits=bits,
**kwargs,
)
[docs] def get_partition_rules(self, *args, **kwargs):
"""
Get the partition rules for the model.
Returns:
`tp.Tuple[tp.Tuple[str, PartitionSpec]]`: The partition rules.
"""
pmag = self.partition_manager
return (
(r"embed_tokens/embedding", pmag.resolve(ColumnWise)),
(r"self_attn/(q_proj|k_proj|v_proj)/kernel", pmag.resolve(ColumnWise)),
(r"self_attn/o_proj/kernel", pmag.resolve(RowWise)),
(r"self_attn/.*proj/bias", pmag.resolve(Replicated)),
(
r"block_sparse_moe/gate/kernel",
pmag.resolve(Replicated if self.use_expert_tensor_mode else ColumnWise),
),
(r"block_sparse_moe/gate/bias", pmag.resolve(Replicated)),
(
r"block_sparse_moe/experts/(w1|w3)/kernel",
get_moe_partition_spec(
partition_manager=self.partition_manager,
direction="column",
tensors_are_expert=self.use_expert_tensor_mode,
is_bias=False,
fsdp_is_ep_bound=self.fsdp_is_ep_bound,
sp_is_ep_bound=self.sp_is_ep_bound,
module_view=True,
),
),
(
r"block_sparse_moe/experts/w2/kernel",
get_moe_partition_spec(
partition_manager=self.partition_manager,
direction="row",
tensors_are_expert=self.use_expert_tensor_mode,
is_bias=False,
fsdp_is_ep_bound=self.fsdp_is_ep_bound,
sp_is_ep_bound=self.sp_is_ep_bound,
module_view=True,
),
),
(r"block_sparse_moe/experts/.*bias", pmag.resolve(Replicated)),
(r".*/(input_layernorm|post_attention_layernorm|norm)/kernel", pmag.resolve(Replicated)),
(r"lm_head/kernel", pmag.resolve(ColumnWise)),
(r"score/kernel", pmag.resolve(RowWise)),
(r".*bias", pmag.resolve(Replicated)),
(r".*", pmag.resolve(Replicated)),
)
@property
def granted_freq_max_position_embedding(self) -> int:
"""Returns the maximum position embedding size specifically for frequency-based position embeddings.
If `freq_max_position_embeddings` is set, it returns that value. Otherwise, it falls back to
`max_position_embeddings`.
Returns:
int: The granted maximum position embedding size for frequency encoding.
"""
return getattr(self, "freq_max_position_embeddings", self.max_position_embeddings)
@property
def granted_mask_max_position_embedding(self) -> int:
"""Returns the maximum position embedding size specifically for mask-based position embeddings.
If `mask_max_position_embeddings` is set, it returns that value. Otherwise, it falls back to
`max_position_embeddings`.
Returns:
int: The granted maximum position embedding size for mask encoding.
"""
return getattr(self, "mask_max_position_embeddings", self.max_position_embeddings)
[docs] def get_mask_details(self) -> dict[int, AttnMaskDetail]:
"""Retrieve attention mask details for each layer in the model.
This method generates a dictionary mapping layer indices to their corresponding attention mask details.
If a sliding window is defined, each layer is assigned a sliding window attention mask with the specified size.
Returns:
dict[int, AttnMaskDetail]: A dictionary where keys are layer indices (int) and values are AttnMaskDetail
objects specifying the attention mask type and size for each layer.
Notes:
- If `self.sliding_window` is None, an empty dictionary is returned.
- The method iterates over `self.num_hidden_layers` to assign mask details for each layer.
- The attention mask type is set to `AttnMaskType.SLIDING` when a sliding window is defined.
"""
mapping = {}
if self.layer_types is not None:
for layer_idx in range(self.num_hidden_layers):
mapping[layer_idx] = AttnMaskDetail(
mask_type=AttnMaskType.from_hf(self.layer_types[layer_idx]),
size=self.sliding_window,
)
return mapping