Unverified Commit 8af84912 authored by fzyzcjy's avatar fzyzcjy Committed by GitHub
Browse files

Support casting bf16 NextN moe to fp8 (#11613)

parent 505329ca
...@@ -25,13 +25,18 @@ from sglang.srt.eplb.expert_distribution import get_global_expert_distribution_r ...@@ -25,13 +25,18 @@ from sglang.srt.eplb.expert_distribution import get_global_expert_distribution_r
from sglang.srt.layers.dp_attention import is_dp_attention_enabled from sglang.srt.layers.dp_attention import is_dp_attention_enabled
from sglang.srt.layers.layernorm import RMSNorm from sglang.srt.layers.layernorm import RMSNorm
from sglang.srt.layers.logits_processor import LogitsProcessor from sglang.srt.layers.logits_processor import LogitsProcessor
from sglang.srt.layers.quantization import Fp8Config
from sglang.srt.layers.quantization.base_config import QuantizationConfig from sglang.srt.layers.quantization.base_config import QuantizationConfig
from sglang.srt.layers.vocab_parallel_embedding import ( from sglang.srt.layers.vocab_parallel_embedding import (
ParallelLMHead, ParallelLMHead,
VocabParallelEmbedding, VocabParallelEmbedding,
) )
from sglang.srt.model_executor.forward_batch_info import ForwardBatch from sglang.srt.model_executor.forward_batch_info import ForwardBatch
from sglang.srt.models.deepseek_v2 import DeepseekV2DecoderLayer, DeepseekV3ForCausalLM from sglang.srt.models.deepseek_v2 import (
DeepseekV2DecoderLayer,
DeepseekV3ForCausalLM,
enable_nextn_moe_bf16_cast_to_fp8,
)
from sglang.srt.server_args import get_global_server_args from sglang.srt.server_args import get_global_server_args
from sglang.srt.utils import BumpAllocator, add_prefix, is_cuda from sglang.srt.utils import BumpAllocator, add_prefix, is_cuda
...@@ -49,6 +54,16 @@ class DeepseekModelNextN(nn.Module): ...@@ -49,6 +54,16 @@ class DeepseekModelNextN(nn.Module):
prefix: str = "", prefix: str = "",
) -> None: ) -> None:
super().__init__() super().__init__()
if enable_nextn_moe_bf16_cast_to_fp8(quant_config):
# refer to real DeepSeek V3 quant config
moe_quant_config = Fp8Config(
is_checkpoint_fp8_serialized=True,
weight_block_size=[128, 128],
)
else:
moe_quant_config = None
if quant_config is not None and quant_config.get_name() == "modelopt_fp4": if quant_config is not None and quant_config.get_name() == "modelopt_fp4":
logger.warning( logger.warning(
"Overriding DeepseekV3ForCausalLMNextN quant config for modelopt_fp4 Deepseek model." "Overriding DeepseekV3ForCausalLMNextN quant config for modelopt_fp4 Deepseek model."
...@@ -74,6 +89,7 @@ class DeepseekModelNextN(nn.Module): ...@@ -74,6 +89,7 @@ class DeepseekModelNextN(nn.Module):
config, config,
0, 0,
quant_config=quant_config, quant_config=quant_config,
moe_quant_config=moe_quant_config,
is_nextn=True, is_nextn=True,
prefix=add_prefix("decoder", prefix), prefix=add_prefix("decoder", prefix),
alt_stream=self.alt_stream, alt_stream=self.alt_stream,
......
...@@ -26,6 +26,7 @@ from typing import Any, Dict, Iterable, Optional, Tuple, Union ...@@ -26,6 +26,7 @@ from typing import Any, Dict, Iterable, Optional, Tuple, Union
import torch import torch
import torch.nn.functional as F import torch.nn.functional as F
from torch import nn from torch import nn
from tqdm import tqdm, trange
from transformers import PretrainedConfig from transformers import PretrainedConfig
from sglang.srt import single_batch_overlap from sglang.srt import single_batch_overlap
...@@ -82,7 +83,7 @@ from sglang.srt.layers.moe import ( ...@@ -82,7 +83,7 @@ from sglang.srt.layers.moe import (
from sglang.srt.layers.moe.ep_moe.layer import DeepEPMoE, get_moe_impl_class from sglang.srt.layers.moe.ep_moe.layer import DeepEPMoE, get_moe_impl_class
from sglang.srt.layers.moe.fused_moe_triton.layer import FusedMoE from sglang.srt.layers.moe.fused_moe_triton.layer import FusedMoE
from sglang.srt.layers.moe.topk import TopK, TopKOutputFormat from sglang.srt.layers.moe.topk import TopK, TopKOutputFormat
from sglang.srt.layers.quantization import deep_gemm_wrapper from sglang.srt.layers.quantization import Fp8Config, deep_gemm_wrapper
from sglang.srt.layers.quantization.base_config import QuantizationConfig from sglang.srt.layers.quantization.base_config import QuantizationConfig
from sglang.srt.layers.quantization.fp8_kernel import ( from sglang.srt.layers.quantization.fp8_kernel import (
is_fp8_fnuz, is_fp8_fnuz,
...@@ -196,6 +197,15 @@ _is_cublas_ge_129 = is_nvidia_cublas_cu12_version_ge_12_9() ...@@ -196,6 +197,15 @@ _is_cublas_ge_129 = is_nvidia_cublas_cu12_version_ge_12_9()
logger = logging.getLogger(__name__) logger = logging.getLogger(__name__)
def enable_nextn_moe_bf16_cast_to_fp8(quant_config):
return (
quant_config is not None
and quant_config.get_name() == "modelopt_fp4"
and get_moe_a2a_backend().is_deepep()
)
FORWARD_ABSORB_CORE_ATTENTION_BACKENDS = [ FORWARD_ABSORB_CORE_ATTENTION_BACKENDS = [
"fa3", "fa3",
"nsa", "nsa",
...@@ -526,6 +536,7 @@ class DeepseekV2MoE(nn.Module): ...@@ -526,6 +536,7 @@ class DeepseekV2MoE(nn.Module):
self.config = config self.config = config
self.layer_id = layer_id self.layer_id = layer_id
self.alt_stream = alt_stream self.alt_stream = alt_stream
self.is_nextn = is_nextn
if self.tp_size > config.n_routed_experts: if self.tp_size > config.n_routed_experts:
raise ValueError( raise ValueError(
...@@ -2381,6 +2392,7 @@ class DeepseekV2DecoderLayer(nn.Module): ...@@ -2381,6 +2392,7 @@ class DeepseekV2DecoderLayer(nn.Module):
config: PretrainedConfig, config: PretrainedConfig,
layer_id: int, layer_id: int,
quant_config: Optional[QuantizationConfig] = None, quant_config: Optional[QuantizationConfig] = None,
moe_quant_config: Optional[QuantizationConfig] = None,
is_nextn: bool = False, is_nextn: bool = False,
prefix: str = "", prefix: str = "",
alt_stream: Optional[torch.cuda.Stream] = None, alt_stream: Optional[torch.cuda.Stream] = None,
...@@ -2430,7 +2442,7 @@ class DeepseekV2DecoderLayer(nn.Module): ...@@ -2430,7 +2442,7 @@ class DeepseekV2DecoderLayer(nn.Module):
if self.is_layer_sparse: if self.is_layer_sparse:
self.mlp = DeepseekV2MoE( self.mlp = DeepseekV2MoE(
config=config, config=config,
quant_config=quant_config, quant_config=moe_quant_config or quant_config,
prefix=add_prefix("mlp", prefix), prefix=add_prefix("mlp", prefix),
layer_id=self.layer_id, layer_id=self.layer_id,
alt_stream=alt_stream, alt_stream=alt_stream,
...@@ -3109,6 +3121,9 @@ class DeepseekV2ForCausalLM(nn.Module): ...@@ -3109,6 +3121,9 @@ class DeepseekV2ForCausalLM(nn.Module):
): ):
self._weight_requant_ue8m0(is_nextn) self._weight_requant_ue8m0(is_nextn)
if is_nextn and enable_nextn_moe_bf16_cast_to_fp8(self.quant_config):
self._transform_scale_nextn_moe_ue8m0()
def _weight_requant_ue8m0(self, is_nextn=False): def _weight_requant_ue8m0(self, is_nextn=False):
weight_block_size = self.quant_config.weight_block_size weight_block_size = self.quant_config.weight_block_size
...@@ -3174,6 +3189,28 @@ class DeepseekV2ForCausalLM(nn.Module): ...@@ -3174,6 +3189,28 @@ class DeepseekV2ForCausalLM(nn.Module):
module.weight, module.weight_scale_inv, weight_block_size module.weight, module.weight_scale_inv, weight_block_size
) )
# TODO avoid code dup (currently combine from weight_requant_ue8m0 and transform_scale_ue8m0)
def _transform_scale_nextn_moe_ue8m0(self):
layer = self.model.decoder
shared_experts = getattr(layer.mlp, "shared_experts", None)
if shared_experts is not None:
for module in [
shared_experts.gate_up_proj,
shared_experts.down_proj,
]:
transform_scale_ue8m0_inplace(
module.weight_scale_inv, mn=module.weight.shape[-2]
)
experts = layer.mlp.experts
if isinstance(experts, DeepEPMoE):
for w in [
experts.w13_weight_fp8,
experts.w2_weight_fp8,
]:
transform_scale_ue8m0_inplace(w[1], mn=w[0].shape[-2])
def load_weights(self, weights: Iterable[Tuple[str, torch.Tensor]], is_nextn=False): def load_weights(self, weights: Iterable[Tuple[str, torch.Tensor]], is_nextn=False):
if is_nextn: if is_nextn:
...@@ -3189,6 +3226,11 @@ class DeepseekV2ForCausalLM(nn.Module): ...@@ -3189,6 +3226,11 @@ class DeepseekV2ForCausalLM(nn.Module):
else: else:
raise ValueError("num_nextn_predict_layers is not in the config") raise ValueError("num_nextn_predict_layers is not in the config")
if is_nextn and enable_nextn_moe_bf16_cast_to_fp8(self.quant_config):
weights = self._quant_nextn_moe_to_fp8_ue8m0(
weights, nextn_layer_id=nextn_layer_id
)
stacked_params_mapping = [ stacked_params_mapping = [
# (param_name, shard_name, shard_id) # (param_name, shard_name, shard_id)
("gate_up_proj", "gate_proj", 0), ("gate_up_proj", "gate_proj", 0),
...@@ -3418,6 +3460,38 @@ class DeepseekV2ForCausalLM(nn.Module): ...@@ -3418,6 +3460,38 @@ class DeepseekV2ForCausalLM(nn.Module):
self.post_load_weights(is_nextn=is_nextn, weight_names=weight_names) self.post_load_weights(is_nextn=is_nextn, weight_names=weight_names)
# TODO avoid code dup
def _quant_nextn_moe_to_fp8_ue8m0(self, weights, nextn_layer_id: int):
weights_dict = dict(weights)
# temporarily only support DeepSeek V3/R1
weight_block_size = [128, 128]
for layer_id in [nextn_layer_id]:
for expert_sub_name in [
"shared_experts",
*[
f"experts.{expert_id}"
for expert_id in range(self.config.n_routed_experts)
],
]:
for stem in [
"gate_proj",
"up_proj",
"down_proj",
]:
partial_name = (
f"model.layers.{layer_id}.mlp.{expert_sub_name}.{stem}"
)
original_weight = weights_dict[f"{partial_name}.weight"]
out_w, out_s = quant_weight_ue8m0(
original_weight, weight_block_size=weight_block_size
)
weights_dict[f"{partial_name}.weight"] = out_w
weights_dict[f"{partial_name}.weight_scale_inv"] = out_s
return list(weights_dict.items())
def get_embed_and_head(self): def get_embed_and_head(self):
return self.model.embed_tokens.weight, self.lm_head.weight return self.model.embed_tokens.weight, self.lm_head.weight
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment