[Model Refactoring] Move DeepSeek V4 layers to models/deepseek_v4/ [2/N] (#43039)
Signed-off-by: Woosuk Kwon <woosuk@inferact.ai>
This commit is contained in:
+1
-2
@@ -153,9 +153,8 @@ mkdocs.yaml @hmellor
|
||||
/vllm/model_executor/models/deepseek_mtp.py @luccafong
|
||||
|
||||
# DeepseekV4-specific files
|
||||
/vllm/models/deepseek_v4 @zyongye
|
||||
/vllm/v1/attention/ops/deepseek_v4_ops @zyongye
|
||||
/vllm/model_executor/layers/deepseek_compressor.py @zyongye
|
||||
/vllm/model_executor/layers/deepseek_v4_attention.py @zyongye
|
||||
/vllm/model_executor/layers/sparse_attn_indexer.py @zyongye
|
||||
|
||||
# Mistral-specific files
|
||||
|
||||
+1
-1
@@ -46,7 +46,6 @@ from vllm.forward_context import ForwardContext, get_forward_context
|
||||
from vllm.logger import init_logger
|
||||
from vllm.model_executor.custom_op import PluggableLayer
|
||||
from vllm.model_executor.layers.attention_layer_base import AttentionLayerBase
|
||||
from vllm.model_executor.layers.deepseek_compressor import DeepseekCompressor
|
||||
from vllm.model_executor.layers.layernorm import RMSNorm
|
||||
from vllm.model_executor.layers.quantization import QuantizationConfig
|
||||
from vllm.model_executor.layers.quantization.input_quant_fp8 import (
|
||||
@@ -55,6 +54,7 @@ from vllm.model_executor.layers.quantization.input_quant_fp8 import (
|
||||
from vllm.model_executor.layers.quantization.utils.quant_utils import (
|
||||
GroupShape,
|
||||
)
|
||||
from vllm.models.deepseek_v4.compressor import DeepseekCompressor
|
||||
from vllm.platforms import current_platform
|
||||
from vllm.utils.multi_stream_utils import (
|
||||
execute_in_parallel,
|
||||
@@ -18,11 +18,6 @@ from vllm.distributed import (
|
||||
)
|
||||
from vllm.forward_context import get_forward_context
|
||||
from vllm.model_executor.layers.activation import SiluAndMul, SiluAndMulWithClamp
|
||||
from vllm.model_executor.layers.deepseek_v4_attention import (
|
||||
DeepseekV4Indexer,
|
||||
DeepseekV4MLAModules,
|
||||
DeepseekV4MultiHeadLatentAttentionWrapper,
|
||||
)
|
||||
from vllm.model_executor.layers.fused_moe import FusedMoE
|
||||
from vllm.model_executor.layers.fused_moe.router.fused_topk_bias_router import (
|
||||
fused_topk_bias,
|
||||
@@ -61,6 +56,11 @@ from vllm.model_executor.models.utils import (
|
||||
maybe_prefix,
|
||||
)
|
||||
from vllm.model_executor.utils import set_weight_attrs
|
||||
from vllm.models.deepseek_v4.attention import (
|
||||
DeepseekV4Indexer,
|
||||
DeepseekV4MLAModules,
|
||||
DeepseekV4MultiHeadLatentAttentionWrapper,
|
||||
)
|
||||
from vllm.platforms import current_platform
|
||||
from vllm.sequence import IntermediateTensors
|
||||
from vllm.triton_utils import tl, triton
|
||||
|
||||
@@ -31,9 +31,7 @@ from vllm.v1.attention.ops.rocm_aiter_mla_sparse import (
|
||||
from vllm.v1.worker.workspace import current_workspace_manager
|
||||
|
||||
if TYPE_CHECKING:
|
||||
from vllm.model_executor.layers.deepseek_v4_attention import (
|
||||
DeepseekV4MLAAttention,
|
||||
)
|
||||
from vllm.models.deepseek_v4.attention import DeepseekV4MLAAttention
|
||||
|
||||
|
||||
def _build_indptr_from_lengths(lengths: torch.Tensor) -> torch.Tensor:
|
||||
|
||||
Reference in New Issue
Block a user