Files
sglang/python/sglang/srt/layers/quantization/gptq.py

1098 lines
38 KiB
Python

from __future__ import annotations
import logging
from dataclasses import dataclass
from fractions import Fraction
from typing import TYPE_CHECKING, Any, Callable, Dict, List, Optional, Union
import torch
from sglang.srt.layers.parameter import (
BasevLLMParameter,
ChannelQuantScaleParameter,
GroupQuantScaleParameter,
PackedColumnParameter,
PackedvLLMParameter,
RowvLLMParameter,
permute_param_layout_,
)
from sglang.srt.layers.quantization.base_config import (
FusedMoEMethodBase,
LinearMethodBase,
QuantizationConfig,
QuantizeMethodBase,
)
from sglang.srt.layers.quantization.marlin_utils import (
apply_gptq_marlin_linear,
check_marlin_supported,
check_marlin_supports_shape,
marlin_is_k_full,
marlin_make_empty_g_idx,
marlin_make_workspace,
marlin_moe_permute_scales,
marlin_permute_scales,
marlin_repeat_scales_on_all_ranks,
marlin_sort_g_idx,
marlin_zero_points,
verify_marlin_supported,
)
from sglang.srt.layers.quantization.utils import (
get_linear_quant_method,
get_scalar_types,
replace_parameter,
unpack_cols,
)
if TYPE_CHECKING:
from sglang.srt.layers.moe.moe_runner import MoeRunnerConfig
from sglang.srt.layers.moe.token_dispatcher import (
StandardDispatchOutput,
CombineInput,
)
from sglang.srt.utils import is_cuda
_is_cuda = is_cuda()
if _is_cuda:
from sgl_kernel import fused_marlin_moe, gptq_gemm, gptq_marlin_repack, gptq_shuffle
logger = logging.getLogger(__name__)
ScalarType, scalar_types = get_scalar_types()
def check_marlin_format(hf_quant_cfg: Dict[str, Any]) -> bool:
# compat: gptqmodel and autogptq (eol) main use checkpoint_format: str
# compat: autogptq <=0.7.1 is_marlin_format: bool
return hf_quant_cfg.get("checkpoint_format") == "marlin" or hf_quant_cfg.get(
"is_marlin_format", False
)
def gptq_marlin_moe_repack(
b_q_weight: torch.Tensor,
perm: torch.Tensor,
size_k: int,
size_n: int,
num_bits: int,
) -> torch.Tensor:
num_experts = b_q_weight.shape[0]
assert size_k % 16 == 0
output = torch.empty(
(num_experts, size_k // 16, size_n * (num_bits // 2)),
device=b_q_weight.device,
dtype=b_q_weight.dtype,
)
for e in range(num_experts):
output[e] = gptq_marlin_repack(b_q_weight[e], perm[e], size_k, size_n, num_bits)
return output
@dataclass
class MarlinLinearLayerConfig:
full_weight_shape: tuple[int, int] # [in, out]
partition_weight_shape: tuple[int, int]
weight_type: ScalarType
act_type: torch.dtype
group_size: int
zero_points: bool
has_g_idx: bool
class GPTQConfig(QuantizationConfig):
"""Config class for GPTQ.
Reference: https://arxiv.org/abs/2210.17323
"""
def __init__(
self,
weight_bits: int,
group_size: int,
desc_act: bool,
lm_head_quantized: bool,
dynamic: Dict[str, Dict[str, Union[int, bool]]],
) -> None:
# GPTQModel use `dynamic` config property to allow per module
# quantization config so each module can be individually optimized.
# Format is Dict[str, Dict] where key is a regex string that can
# perform both positive ("+:" prefixed) or negative ("-:" prefixed)
# matching of a module.
# Default to positive match, override base quant config mode, if no
# prefix is used. Value is in dict format of field key and override
# value.
# Negative matching will skip quantization init for this module
# entirely:
# non-quantized inference. More details and quantization examples can be
# found at: https://github.com/ModelCloud/GPTQModel
# Example:
# # last 1/2 of the layers 10-21 has 8bit vs 4bit for 0-9
# # last 1/4 of the layers 16-21 has 8bit and group_size 64
# dynamic = {
# #`.*\.` matches the layers_node prefix
# # positive match layer 10-15
# r"+:.*\.(?:1[0-5])\..*": {"bits": 8,},
# # positive match layer 16-21
# r"+:.*\.(?:1[6-9]|20|21)\..*": {"bits": 8, "group_size": 64,},
# r"-:.*\.moe\..*": {}, # negative match (skip) all `moe` layers
# }
super().__init__()
self.dynamic = dynamic
self.weight_bits = weight_bits
self.group_size = group_size
self.desc_act = desc_act
self.lm_head_quantized = lm_head_quantized
self.pack_factor = Fraction(32, self.weight_bits)
if self.weight_bits not in [2, 3, 4, 8]:
raise ValueError(
"Currently, only 2/3/4/8-bit weight quantization is "
f"supported for GPTQ, but got {self.weight_bits} bits."
)
def __repr__(self) -> str:
return (
f"GPTQConfig(weight_bits={self.weight_bits}, "
f"group_size={self.group_size}, "
f"desc_act={self.desc_act}),"
f"lm_head_quantized={self.lm_head_quantized}), "
f"dynamic={self.dynamic}"
)
def get_scaled_act_names(self) -> List[str]:
"""Returns the activation function names that should be post-scaled.
For now, this is only used by AWQ.
"""
raise NotImplementedError
@classmethod
def get_name(cls) -> str:
return "gptq"
@classmethod
def get_supported_act_dtypes(cls) -> List[torch.dtype]:
return [torch.half]
@classmethod
# Need to figure it out
def get_min_capability(cls) -> int:
return 60
@classmethod
def get_config_filenames(cls) -> List[str]:
return ["quantize_config.json"]
@classmethod
def from_config(cls, config: Dict[str, Any]) -> GPTQConfig:
dynamic = cls.get_from_keys_or(config, ["dynamic"], default={})
dynamic = {} if dynamic is None else dynamic
weight_bits = cls.get_from_keys(config, ["bits"])
group_size = cls.get_from_keys(config, ["group_size"])
desc_act = cls.get_from_keys(config, ["desc_act"])
lm_head_quantized = cls.get_from_keys_or(config, ["lm_head"], default=False)
return cls(weight_bits, group_size, desc_act, lm_head_quantized, dynamic)
def get_quant_method(
self, layer: torch.nn.Module, prefix: str
) -> Optional[LinearMethodBase]:
# Delay the import to avoid circular dependency
from sglang.srt.layers.linear import LinearBase
from sglang.srt.layers.moe.fused_moe_triton import FusedMoE
if isinstance(layer, FusedMoE):
raise TypeError("GPTQ Method does not support MoE, please use gptq_marlin")
else:
return get_linear_quant_method(
self, layer, prefix=prefix, linear_method_cls=GPTQLinearMethod
)
class GPTQMarlinConfig(QuantizationConfig):
"""Config class for GPTQ Marlin"""
# (num_bits, is_sym) -> quant_type
TYPE_MAP = {
(4, True): scalar_types.uint4b8,
(8, True): scalar_types.uint8b128,
}
def __init__(
self,
weight_bits: int,
group_size: int,
desc_act: bool,
is_sym: bool,
lm_head_quantized: bool,
dynamic: Dict[str, Dict[str, Union[int, bool]]],
full_config: Dict[str, Any],
) -> None:
super().__init__()
if desc_act and group_size == -1:
# In this case, act_order == True is the same as act_order == False
# (since we have only one group per output channel)
desc_act = False
# GPTQModel use `dynamic` config property to allow per module
# quantization config so each module can be individually optimized.
# Format is Dict[str, Dict] where key is a regex string that can
# perform both positive ("+:" prefixed) or negative ("-:" prefixed)
# matching of a module.
# Default to positive match, override base quant config mode, if no
# prefix is used. Value is in dict format of field key and override
# value.
# Negative matching will skip quantization init for this module
# entirely:
# non-quantized inference. More details and quantization examples can be
# found at: https://github.com/ModelCloud/GPTQModel
# Example:
# # last 1/2 of the layers 10-21 has 8bit vs 4bit for 0-9
# # last 1/4 of the layers 16-21 has 8bit and group_size 64
# dynamic = {
# #`.*\.` matches the layers_node prefix
# # positive match layer 10-15
# r"+:.*\.(?:1[0-5])\..*": {"bits": 8,},
# # positive match layer 16-21
# r"+:.*\.(?:1[6-9]|20|21)\..*": {"bits": 8, "group_size": 64,},
# r"-:.*\.moe\..*": {}, # negative match (skip) all `moe` layers
# }
self.dynamic = dynamic
self.weight_bits = weight_bits
self.is_sym = is_sym
self.pack_factor = 32 // weight_bits # packed into int32
self.group_size = group_size
self.desc_act = desc_act
self.lm_head_quantized = lm_head_quantized
self.full_config = full_config
if (weight_bits, is_sym) not in self.TYPE_MAP:
raise ValueError(
"Unsupported quantization config: " f"bits={weight_bits}, sym={is_sym}"
)
# (num_bits, is_sym) -> quant_type
self.quant_type = self.TYPE_MAP[(weight_bits, is_sym)]
def __repr__(self) -> str:
return (
f"GPTQMarlinConfig(quant_type={self.quant_type}, "
f"group_size={self.group_size}, "
f"desc_act={self.desc_act}, "
f"lm_head_quantized={self.lm_head_quantized}), "
f"dynamic={self.dynamic}"
)
def get_scaled_act_names(self) -> List[str]:
"""Returns the activation function names that should be post-scaled.
For now, this is only used by AWQ.
"""
raise NotImplementedError
@classmethod
def get_name(cls) -> str:
return "gptq_marlin"
@classmethod
def get_supported_act_dtypes(cls) -> List[torch.dtype]:
return [torch.half, torch.bfloat16]
@classmethod
def get_min_capability(cls) -> int:
return 80
@classmethod
def get_config_filenames(cls) -> List[str]:
return ["quantize_config.json"]
@classmethod
def from_config(cls, config: Dict[str, Any]) -> GPTQMarlinConfig:
dynamic = cls.get_from_keys_or(config, ["dynamic"], default={})
dynamic = {} if dynamic is None else dynamic
weight_bits = cls.get_from_keys(config, ["bits"])
group_size = cls.get_from_keys(config, ["group_size"])
desc_act = cls.get_from_keys(config, ["desc_act"])
is_sym = cls.get_from_keys(config, ["sym"])
lm_head_quantized = cls.get_from_keys_or(config, ["lm_head"], default=False)
return cls(
weight_bits,
group_size,
desc_act,
is_sym,
lm_head_quantized,
dynamic,
config,
)
@classmethod
def override_quantization_method(cls, hf_quant_cfg, user_quant) -> Optional[str]:
is_marlin_format = check_marlin_format(hf_quant_cfg)
can_convert = cls.is_gptq_marlin_compatible(hf_quant_cfg)
is_valid_user_quant = (
user_quant is None or user_quant == "marlin" or user_quant == "gptq_marlin"
)
if not is_marlin_format and can_convert and is_valid_user_quant:
msg = (
"The model is convertible to {} during runtime."
" Using {} kernel.".format(cls.get_name(), cls.get_name())
)
logger.info(msg)
return cls.get_name()
if not is_marlin_format and can_convert and user_quant == "gptq":
logger.info(
"Detected that the model can run with gptq_marlin"
", however you specified quantization=gptq explicitly,"
" so forcing gptq. Use quantization=gptq_marlin for"
" faster inference"
)
return None
def get_quant_method(
self, layer: torch.nn.Module, prefix: str
) -> Optional[QuantizeMethodBase]:
# Delay the import to avoid circular dependency
from sglang.srt.layers.moe.fused_moe_triton import FusedMoE
if isinstance(layer, FusedMoE):
return GPTQMarlinMoEMethod(self)
return get_linear_quant_method(self, layer, prefix, GPTQMarlinLinearMethod)
@classmethod
def is_gptq_marlin_compatible(cls, quant_config: Dict[str, Any]):
quant_method = quant_config.get("quant_method", "").lower()
num_bits = quant_config.get("bits")
group_size = quant_config.get("group_size")
sym = quant_config.get("sym")
desc_act = quant_config.get("desc_act")
if not _is_cuda:
return False
if quant_method != "gptq":
return False
# Marlin conversion is only valid if required properties are found
if num_bits is None or group_size is None or sym is None or desc_act is None:
return False
if (num_bits, sym) not in cls.TYPE_MAP:
return False
return check_marlin_supported(
quant_type=cls.TYPE_MAP[(num_bits, sym)], group_size=group_size
)
class GPTQLinearMethod(LinearMethodBase):
"""Linear method for GPTQ.
Args:
quant_config: The GPTQ quantization config.
"""
def __init__(self, quant_config: GPTQConfig):
self.quant_config = quant_config
def create_weights(
self,
layer: torch.nn.Module,
input_size_per_partition: int,
output_partition_sizes: list[int],
input_size: int,
output_size: int,
params_dtype: torch.dtype,
**extra_weight_attrs,
):
del output_size # Unused.
weight_loader = extra_weight_attrs.get("weight_loader")
if input_size_per_partition % self.quant_config.group_size != 0:
raise ValueError(
"The input size is not aligned with the quantized "
"weight shape. This can be caused by too large "
"tensor parallel size."
)
output_size_per_partition = sum(output_partition_sizes)
if output_size_per_partition % self.quant_config.pack_factor.numerator != 0:
raise ValueError(
"The output size is not aligned with the quantized "
"weight shape. This can be caused by too large "
"tensor parallel size."
)
if self.quant_config.group_size != -1:
group_size = self.quant_config.group_size
else:
group_size = input_size
self.use_shuffle = True
scale_and_zero_size = input_size // group_size
scale_and_zero_input_dim = None
if (
input_size != input_size_per_partition
and self.quant_config.group_size != -1
):
if self.quant_config.desc_act:
self.use_shuffle = False
else:
# we need to partition qzeros and scales for exllama kernel
scale_and_zero_size = input_size_per_partition // group_size
scale_and_zero_input_dim = 0
qweight = PackedvLLMParameter(
data=torch.empty(
input_size_per_partition // self.quant_config.pack_factor,
output_size_per_partition,
dtype=torch.int32,
),
input_dim=0,
output_dim=1,
packed_dim=0,
packed_factor=self.quant_config.pack_factor,
weight_loader=weight_loader,
)
g_idx = RowvLLMParameter(
data=torch.tensor(
[
i // self.quant_config.group_size
for i in range(input_size_per_partition)
],
dtype=torch.int32,
),
input_dim=0,
weight_loader=weight_loader,
)
qzeros_args = {
"data": torch.empty(
scale_and_zero_size,
output_size_per_partition // self.quant_config.pack_factor,
dtype=torch.int32,
),
"weight_loader": weight_loader,
}
weight_scale_args = {
"data": torch.empty(
scale_and_zero_size,
output_size_per_partition,
dtype=params_dtype,
),
"weight_loader": weight_loader,
}
if scale_and_zero_input_dim is None:
scales = ChannelQuantScaleParameter(output_dim=1, **weight_scale_args)
qzeros = PackedColumnParameter(
output_dim=1,
packed_dim=1,
packed_factor=self.quant_config.pack_factor,
**qzeros_args,
)
else:
scales = GroupQuantScaleParameter(
output_dim=1, input_dim=0, **weight_scale_args
)
qzeros = PackedvLLMParameter(
input_dim=0,
output_dim=1,
packed_dim=1,
packed_factor=self.quant_config.pack_factor,
**qzeros_args,
)
layer.register_parameter("qweight", qweight)
layer.register_parameter("g_idx", g_idx)
layer.register_parameter("qzeros", qzeros)
layer.register_parameter("scales", scales)
def process_weights_after_loading(self, layer: torch.nn.Module) -> None:
# for torch.compile
layer.qzeros = torch.nn.Parameter(layer.qzeros.data, requires_grad=False)
layer.qweight = torch.nn.Parameter(layer.qweight.data, requires_grad=False)
layer.g_idx = torch.nn.Parameter(layer.g_idx.data, requires_grad=False)
layer.scales = torch.nn.Parameter(layer.scales.data, requires_grad=False)
# exllama needs to shuffle the weight after the weight is loaded
# here we do the shuffle on first forward pass
if self.use_shuffle:
if self.quant_config.desc_act:
layer.g_idx.data = torch.argsort(layer.g_idx).to(torch.int)
else:
layer.g_idx.data = torch.empty(
(0,), dtype=torch.int, device=layer.g_idx.device
)
gptq_shuffle(layer.qweight, layer.g_idx, self.quant_config.weight_bits)
def apply(
self,
layer: torch.nn.Module,
x: torch.Tensor,
bias: Optional[torch.Tensor] = None,
) -> torch.Tensor:
out_shape = x.shape[:-1] + (layer.qweight.shape[-1],)
reshaped_x = x.reshape(-1, x.shape[-1])
output = gptq_gemm(
reshaped_x,
layer.qweight,
layer.qzeros,
layer.scales,
layer.g_idx,
self.use_shuffle,
self.quant_config.weight_bits,
)
if bias is not None:
output.add_(bias)
return output.reshape(out_shape)
class GPTQMarlinLinearMethod(LinearMethodBase):
"""Linear method for GPTQ Marlin.
Args:
quant_config: The GPTQ Marlin quantization config.
"""
_kernel_backends_being_used: set[str] = set()
def __init__(self, quant_config: GPTQMarlinConfig) -> None:
self.quant_config = quant_config
# Verify supported on platform.
verify_marlin_supported(
quant_type=self.quant_config.quant_type,
group_size=self.quant_config.group_size,
)
def create_weights(
self,
layer: torch.nn.Module,
input_size_per_partition: int,
output_partition_sizes: list[int],
input_size: int,
output_size: int,
params_dtype: torch.dtype,
**extra_weight_attrs,
) -> None:
output_size_per_partition = sum(output_partition_sizes)
is_row_parallel = input_size != input_size_per_partition
weight_loader = extra_weight_attrs.get("weight_loader")
self.kernel_config = MarlinLinearLayerConfig(
full_weight_shape=(input_size, output_size),
partition_weight_shape=(
input_size_per_partition,
output_size_per_partition,
),
weight_type=self.quant_config.quant_type,
act_type=params_dtype,
group_size=self.quant_config.group_size,
zero_points=False,
has_g_idx=self.quant_config.desc_act,
)
# Normalize group_size
if self.quant_config.group_size != -1:
group_size = self.quant_config.group_size
else:
group_size = input_size
# Determine sharding
if marlin_repeat_scales_on_all_ranks(
self.quant_config.desc_act, self.quant_config.group_size, is_row_parallel
):
# By setting scale_dim == None, weight_loader will
# repeat the scales on each GPU in TP>1 case.
scales_and_zp_input_dim = None
scales_and_zp_size = input_size // group_size
else:
# By setting scale_dim == 0, weight_loader will
# shard the scales in TP>1 case.
scales_and_zp_input_dim = 0
scales_and_zp_size = input_size_per_partition // group_size
# Quantized weights
qweight = PackedvLLMParameter(
data=torch.empty(
input_size_per_partition // self.quant_config.pack_factor,
output_size_per_partition,
dtype=torch.int32,
),
input_dim=0,
output_dim=1,
packed_dim=0,
packed_factor=self.quant_config.pack_factor,
weight_loader=weight_loader,
)
# Activation order
g_idx = RowvLLMParameter(
data=torch.empty(
input_size_per_partition,
dtype=torch.int32,
),
input_dim=0,
weight_loader=weight_loader,
)
qzeros_args = {
"data": torch.empty(
scales_and_zp_size,
output_size_per_partition // self.quant_config.pack_factor,
dtype=torch.int32,
),
"weight_loader": weight_loader,
}
weight_scale_args = {
"data": torch.empty(
scales_and_zp_size,
output_size_per_partition,
dtype=params_dtype,
),
"weight_loader": weight_loader,
}
if scales_and_zp_input_dim is None:
scales = ChannelQuantScaleParameter(output_dim=1, **weight_scale_args)
qzeros = PackedColumnParameter(
output_dim=1,
packed_dim=1,
packed_factor=self.quant_config.pack_factor,
**qzeros_args,
)
else:
scales = GroupQuantScaleParameter(
output_dim=1, input_dim=0, **weight_scale_args
)
qzeros = PackedvLLMParameter(
input_dim=0,
output_dim=1,
packed_dim=1,
packed_factor=self.quant_config.pack_factor,
**qzeros_args,
)
layer.register_parameter("qweight", qweight)
layer.register_parameter("g_idx", g_idx)
layer.register_parameter("scales", scales)
layer.register_parameter("qzeros", qzeros)
def process_weights_after_loading(self, layer: torch.nn.Module) -> None:
device = getattr(layer, "qweight").device
c = self.kernel_config
check_marlin_supports_shape(
c.partition_weight_shape[1], # out_features
c.partition_weight_shape[0], # in_features
c.full_weight_shape[0], # in_features
c.group_size,
)
row_parallel = c.partition_weight_shape[0] != c.full_weight_shape[0]
self.is_k_full = marlin_is_k_full(c.has_g_idx, row_parallel)
# Allocate marlin workspace.
self.workspace = marlin_make_workspace(device)
# Default names since marlin requires empty parameters for these,
# TODO: remove this requirement from marlin (allow optional tensors)
self.w_q_name = "qweight"
self.w_s_name = "scales"
self.w_zp_name = "qzeros"
self.w_gidx_name = "g_idx"
def _transform_param(
layer: torch.nn.Module, name: Optional[str], fn: Callable
) -> None:
if name is not None and getattr(layer, name, None) is not None:
old_param = getattr(layer, name)
new_param = fn(old_param)
# replace the parameter with torch.nn.Parameter for TorchDynamo
# compatibility
replace_parameter(
layer, name, torch.nn.Parameter(new_param.data, requires_grad=False)
)
def transform_w_q(x):
assert isinstance(x, BasevLLMParameter)
permute_param_layout_(x, input_dim=0, output_dim=1, packed_dim=0)
x.data = gptq_marlin_repack(
x.data.contiguous(),
perm=layer.g_idx_sort_indices,
size_k=c.partition_weight_shape[0],
size_n=c.partition_weight_shape[1],
num_bits=c.weight_type.size_bits,
)
return x
def transform_w_s(x):
assert isinstance(x, BasevLLMParameter)
permute_param_layout_(x, input_dim=0, output_dim=1)
x.data = marlin_permute_scales(
x.data.contiguous(),
size_k=c.partition_weight_shape[0],
size_n=c.partition_weight_shape[1],
group_size=c.group_size,
)
return x
if c.has_g_idx:
g_idx, g_idx_sort_indices = marlin_sort_g_idx(
getattr(layer, self.w_gidx_name)
)
_transform_param(layer, self.w_gidx_name, lambda _: g_idx)
layer.g_idx_sort_indices = g_idx_sort_indices
else:
setattr(layer, self.w_gidx_name, marlin_make_empty_g_idx(device))
layer.g_idx_sort_indices = marlin_make_empty_g_idx(device)
if c.zero_points:
grouped_k = (
c.partition_weight_shape[0] // c.group_size if c.group_size != -1 else 1
)
_transform_param(
layer,
self.w_zp_name,
lambda x: marlin_zero_points(
unpack_cols(
x.t(),
c.weight_type.size_bits,
grouped_k,
c.partition_weight_shape[1],
),
size_k=grouped_k,
size_n=c.partition_weight_shape[1],
num_bits=c.weight_type.size_bits,
),
)
else:
setattr(layer, self.w_zp_name, marlin_make_empty_g_idx(device))
_transform_param(layer, self.w_q_name, transform_w_q)
_transform_param(layer, self.w_s_name, transform_w_s)
def apply(
self,
layer: torch.nn.Module,
x: torch.Tensor,
bias: Optional[torch.Tensor] = None,
) -> torch.Tensor:
c = self.kernel_config
def _get_weight_params(
layer: torch.nn.Module,
) -> tuple[
torch.Tensor, # w_q
torch.Tensor, # w_s
Optional[torch.Tensor], # w_zp,
Optional[torch.Tensor], # w_gidx
]:
return (
getattr(layer, self.w_q_name),
getattr(layer, self.w_s_name),
getattr(layer, self.w_zp_name or "", None),
getattr(layer, self.w_gidx_name or "", None),
)
w_q, w_s, w_zp, w_gidx = _get_weight_params(layer)
# `process_weights_after_loading` will ensure w_zp and w_gidx are not
# None for marlin
return apply_gptq_marlin_linear(
input=x,
weight=w_q,
weight_scale=w_s,
weight_zp=w_zp, # type: ignore
g_idx=w_gidx, # type: ignore
g_idx_sort_indices=layer.g_idx_sort_indices,
workspace=self.workspace,
wtype=c.weight_type,
input_size_per_partition=c.partition_weight_shape[0],
output_size_per_partition=c.partition_weight_shape[1],
is_k_full=self.is_k_full,
bias=bias,
)
class GPTQMarlinMoEMethod(FusedMoEMethodBase):
"""MoE Marlin method with quantization."""
def __init__(self, quant_config: GPTQMarlinConfig) -> None:
self.quant_config = quant_config
def create_weights(
self,
layer: torch.nn.Module,
num_experts: int,
hidden_size: int,
intermediate_size_per_partition: int,
params_dtype: torch.dtype,
**extra_weight_attrs,
):
# Delay the import to avoid circular dependency
from sglang.srt.layers.linear import set_weight_attrs
from sglang.srt.layers.moe.fused_moe_triton import FusedMoeWeightScaleSupported
self.is_k_full = (not self.quant_config.desc_act) or layer.moe_tp_size == 1
if self.quant_config.group_size != -1:
scales_size13 = hidden_size // self.quant_config.group_size
if self.quant_config.desc_act:
w2_scales_size = intermediate_size_per_partition
else:
w2_scales_size = intermediate_size_per_partition * layer.moe_tp_size
scales_size2 = w2_scales_size // self.quant_config.group_size
strategy = FusedMoeWeightScaleSupported.GROUP.value
else:
scales_size13 = 1
scales_size2 = 1
strategy = FusedMoeWeightScaleSupported.CHANNEL.value
extra_weight_attrs.update({"quant_method": strategy, "is_transposed": True})
# Fused gate_up_proj (column parallel)
w13_qweight = torch.nn.Parameter(
torch.empty(
num_experts,
hidden_size // self.quant_config.pack_factor,
2 * intermediate_size_per_partition,
dtype=torch.int32,
),
requires_grad=False,
)
layer.register_parameter("w13_qweight", w13_qweight)
set_weight_attrs(w13_qweight, extra_weight_attrs)
# down_proj (row parallel)
w2_qweight = torch.nn.Parameter(
torch.empty(
num_experts,
intermediate_size_per_partition // self.quant_config.pack_factor,
hidden_size,
dtype=torch.int32,
),
requires_grad=False,
)
layer.register_parameter("w2_qweight", w2_qweight)
set_weight_attrs(w2_qweight, extra_weight_attrs)
# up_proj scales
w13_scales = torch.nn.Parameter(
torch.empty(
num_experts,
scales_size13,
2 * intermediate_size_per_partition,
dtype=torch.half,
),
requires_grad=False,
)
layer.register_parameter("w13_scales", w13_scales)
set_weight_attrs(w13_scales, extra_weight_attrs)
# down_proj scales
w2_scales = torch.nn.Parameter(
torch.empty(num_experts, scales_size2, hidden_size, dtype=torch.half),
requires_grad=False,
)
layer.register_parameter("w2_scales", w2_scales)
set_weight_attrs(w2_scales, extra_weight_attrs)
# dont shard the w2 scales when running act order
set_weight_attrs(w2_scales, {"load_full_w2": self.quant_config.desc_act})
# up_proj scales
w13_qzeros = torch.nn.Parameter(
torch.empty(
num_experts,
scales_size13,
2 * intermediate_size_per_partition // self.quant_config.pack_factor,
dtype=params_dtype,
),
requires_grad=False,
)
layer.register_parameter("w13_qzeros", w13_qzeros)
set_weight_attrs(w13_qzeros, extra_weight_attrs)
# down_proj scales
w2_qzeros = torch.nn.Parameter(
torch.empty(
num_experts,
scales_size2,
hidden_size // self.quant_config.pack_factor,
dtype=params_dtype,
),
requires_grad=False,
)
layer.register_parameter("w2_qzeros", w2_qzeros)
set_weight_attrs(w2_qzeros, extra_weight_attrs)
# dont shard the w2 scales when running act order
set_weight_attrs(w2_qzeros, {"load_full_w2": self.quant_config.desc_act})
w13_g_idx = torch.nn.Parameter(
torch.empty(
num_experts,
hidden_size,
dtype=torch.int32,
),
requires_grad=False,
)
layer.register_parameter("w13_g_idx", w13_g_idx)
set_weight_attrs(w13_g_idx, extra_weight_attrs)
w2_g_idx = torch.nn.Parameter(
torch.empty(
num_experts,
intermediate_size_per_partition,
dtype=torch.int32,
),
requires_grad=False,
)
layer.register_parameter("w2_g_idx", w2_g_idx)
set_weight_attrs(w2_g_idx, extra_weight_attrs)
w13_g_idx_sort_indices = torch.nn.Parameter(
torch.empty(
num_experts,
hidden_size,
dtype=torch.int32,
),
requires_grad=False,
)
layer.register_parameter("w13_g_idx_sort_indices", w13_g_idx_sort_indices)
set_weight_attrs(w13_g_idx_sort_indices, extra_weight_attrs)
w2_g_idx_sort_indices = torch.nn.Parameter(
torch.empty(
num_experts,
intermediate_size_per_partition,
dtype=torch.int32,
),
requires_grad=False,
)
layer.register_parameter("w2_g_idx_sort_indices", w2_g_idx_sort_indices)
set_weight_attrs(w2_g_idx_sort_indices, extra_weight_attrs)
def process_weights_after_loading(self, layer: torch.nn.Module) -> None:
# Process act_order
if self.quant_config.desc_act:
# Get sorting based on g_idx
num_experts = layer.w13_g_idx.shape[0]
w13_g_idx_sort_indices = torch.empty_like(layer.w13_g_idx)
w2_g_idx_sort_indices = torch.empty_like(layer.w2_g_idx)
w13_sorted_g_idx = torch.empty_like(layer.w13_g_idx)
w2_sorted_g_idx = torch.empty_like(layer.w2_g_idx)
for e in range(num_experts):
w13_g_idx_sort_indices[e] = torch.argsort(layer.w13_g_idx[e]).to(
torch.int32
)
w2_g_idx_sort_indices[e] = torch.argsort(layer.w2_g_idx[e]).to(
torch.int32
)
w13_sorted_g_idx[e] = layer.w13_g_idx[e][w13_g_idx_sort_indices[e]]
w2_sorted_g_idx[e] = layer.w2_g_idx[e][w2_g_idx_sort_indices[e]]
replace_parameter(layer, "w13_g_idx", w13_sorted_g_idx)
replace_parameter(layer, "w2_g_idx", w2_sorted_g_idx)
replace_parameter(layer, "w13_g_idx_sort_indices", w13_g_idx_sort_indices)
replace_parameter(layer, "w2_g_idx_sort_indices", w2_g_idx_sort_indices)
else:
# Reset g_idx related tensors
num_experts = layer.w13_g_idx.shape[0]
device = layer.w13_g_idx.device
layer.w13_g_idx = torch.nn.Parameter(
torch.empty((num_experts, 0), dtype=torch.int32, device=device),
requires_grad=False,
)
layer.w2_g_idx = torch.nn.Parameter(
torch.empty((num_experts, 0), dtype=torch.int32, device=device),
requires_grad=False,
)
layer.w13_g_idx_sort_indices = torch.nn.Parameter(
torch.empty((num_experts, 0), dtype=torch.int32, device=device),
requires_grad=False,
)
layer.w2_g_idx_sort_indices = torch.nn.Parameter(
torch.empty((num_experts, 0), dtype=torch.int32, device=device),
requires_grad=False,
)
# Repack weights
marlin_w13_qweight = gptq_marlin_moe_repack(
layer.w13_qweight,
layer.w13_g_idx_sort_indices,
layer.w13_qweight.shape[1] * self.quant_config.pack_factor,
layer.w13_qweight.shape[2],
self.quant_config.weight_bits,
)
replace_parameter(layer, "w13_qweight", marlin_w13_qweight)
marlin_w2_qweight = gptq_marlin_moe_repack(
layer.w2_qweight,
layer.w2_g_idx_sort_indices,
layer.w2_qweight.shape[1] * self.quant_config.pack_factor,
layer.w2_qweight.shape[2],
self.quant_config.weight_bits,
)
replace_parameter(layer, "w2_qweight", marlin_w2_qweight)
# Repack scales
marlin_w13_scales = marlin_moe_permute_scales(
s=layer.w13_scales,
size_k=layer.intermediate_size_per_partition,
size_n=layer.w13_scales.shape[2],
group_size=self.quant_config.group_size,
)
replace_parameter(layer, "w13_scales", marlin_w13_scales)
marlin_w2_scales = marlin_moe_permute_scales(
s=layer.w2_scales,
size_k=layer.w2_scales.shape[1]
* (
self.quant_config.group_size
if self.quant_config.group_size != -1
else self.quant_config.pack_factor
),
size_n=layer.w2_scales.shape[2],
group_size=self.quant_config.group_size,
)
replace_parameter(layer, "w2_scales", marlin_w2_scales)
def create_moe_runner(
self, layer: torch.nn.Module, moe_runner_config: MoeRunnerConfig
):
self.moe_runner_config = moe_runner_config
def apply(
self,
layer: torch.nn.Module,
dispatch_output: StandardDispatchOutput,
) -> CombineInput:
from sglang.srt.layers.moe.token_dispatcher import StandardCombineInput
x = dispatch_output.hidden_states
topk_output = dispatch_output.topk_output
# Delay the import to avoid circular dependency
assert (
self.moe_runner_config.activation == "silu"
), "Only SiLU activation is supported."
# The input must currently be float16
orig_dtype = x.dtype
x = x.half()
topk_weights, topk_ids, router_logits = topk_output
output = fused_marlin_moe(
x,
layer.w13_qweight,
layer.w2_qweight,
layer.w13_scales,
layer.w2_scales,
router_logits,
topk_weights,
topk_ids,
g_idx1=layer.w13_g_idx,
g_idx2=layer.w2_g_idx,
sort_indices1=layer.w13_g_idx_sort_indices,
sort_indices2=layer.w2_g_idx_sort_indices,
num_bits=self.quant_config.weight_bits,
is_k_full=self.is_k_full,
).to(orig_dtype)
return StandardCombineInput(hidden_states=output)