cleanup models dependencies 1/n (#2948)

python/sglang/srt/layers/moe/ep_moe/layer.py

@@ -5,7 +5,6 @@
 from torch.nn import Module
 from vllm import _custom_ops as ops
 from vllm.model_executor.custom_op import CustomOp
-from vllm.model_executor.layers.quantization.fp8 import Fp8Config, Fp8MoEMethod
 
 from sglang.srt.distributed import (
     get_tensor_model_parallel_rank,
@@ -25,6 +24,7 @@
     QuantizationConfig,
     QuantizeMethodBase,
 )
+from sglang.srt.layers.quantization.fp8 import Fp8Config, Fp8MoEMethod
 from sglang.srt.utils import is_hip, set_weight_attrs
 
 logger = logging.getLogger(__name__)

python/sglang/srt/lora/lora.py

@@ -19,26 +19,18 @@
 # https://github.com/vllm-project/vllm/blob/4abf6336ec65c270343eb895e7b18786e9274176/vllm/lora/layers.py
 
 
-import json
-import os
 import re
-from typing import Any, Dict, List, Optional, Tuple
 
-import safetensors.torch
 import torch
 from torch import nn
-from vllm.model_executor.layers.vocab_parallel_embedding import (
-    ParallelLMHead,
-    VocabParallelEmbedding,
-)
+from vllm.model_executor.layers.vocab_parallel_embedding import VocabParallelEmbedding
 
 from sglang.srt.layers.linear import (
     ColumnParallelLinear,
     MergedColumnParallelLinear,
     QKVParallelLinear,
     RowParallelLinear,
 )
-from sglang.srt.model_executor.forward_batch_info import ForwardBatch, ForwardMode
 from sglang.srt.model_loader.loader import DefaultModelLoader
 
 

python/sglang/srt/models/baichuan.py

@@ -24,11 +24,6 @@
 import torch
 from torch import nn
 from transformers import PretrainedConfig
-from vllm.model_executor.layers.linear import (
-    MergedColumnParallelLinear,
-    QKVParallelLinear,
-    RowParallelLinear,
-)
 from vllm.model_executor.layers.rotary_embedding import get_rope
 
 from sglang.srt.distributed import (
@@ -37,6 +32,11 @@
 )
 from sglang.srt.layers.activation import SiluAndMul
 from sglang.srt.layers.layernorm import RMSNorm
+from sglang.srt.layers.linear import (
+    MergedColumnParallelLinear,
+    QKVParallelLinear,
+    RowParallelLinear,
+)
 from sglang.srt.layers.logits_processor import LogitsProcessor
 from sglang.srt.layers.quantization.base_config import QuantizationConfig
 from sglang.srt.layers.radix_attention import RadixAttention

python/sglang/srt/models/gpt2.py

@@ -22,10 +22,9 @@
 import torch
 from torch import nn
 from transformers import GPT2Config
-from vllm.model_executor.layers.activation import get_act_fn
-from vllm.model_executor.layers.vocab_parallel_embedding import VocabParallelEmbedding
 
 from sglang.srt.distributed.parallel_state import get_tensor_model_parallel_world_size
+from sglang.srt.layers.activation import get_act_fn
 
 # from sglang.srt.layers.activation import get_act_fn
 from sglang.srt.layers.linear import (

python/sglang/srt/models/minicpm3.py

@@ -19,17 +19,17 @@
 import torch
 from torch import nn
 from transformers import PretrainedConfig
-from vllm.model_executor.layers.linear import (
-    ColumnParallelLinear,
-    MergedColumnParallelLinear,
-    ReplicatedLinear,
-    RowParallelLinear,
-)
 from vllm.model_executor.layers.rotary_embedding import get_rope
 
 from sglang.srt.distributed import get_tensor_model_parallel_world_size
 from sglang.srt.layers.activation import SiluAndMul
 from sglang.srt.layers.layernorm import RMSNorm
+from sglang.srt.layers.linear import (
+    ColumnParallelLinear,
+    MergedColumnParallelLinear,
+    ReplicatedLinear,
+    RowParallelLinear,
+)
 from sglang.srt.layers.logits_processor import LogitsProcessor
 from sglang.srt.layers.quantization.base_config import QuantizationConfig
 from sglang.srt.layers.radix_attention import RadixAttention

python/sglang/srt/models/olmo2.py

@@ -22,7 +22,6 @@
 from torch import nn
 from transformers import PretrainedConfig
 from vllm.model_executor.layers.rotary_embedding import get_rope
-from vllm.model_executor.model_loader.weight_utils import default_weight_loader
 
 from sglang.srt.distributed import (
     get_tensor_model_parallel_rank,
@@ -45,6 +44,7 @@
     VocabParallelEmbedding,
 )
 from sglang.srt.model_executor.forward_batch_info import ForwardBatch
+from sglang.srt.model_loader.weight_utils import default_weight_loader
 from sglang.srt.utils import make_layers
 
 

python/sglang/srt/models/olmoe.py

@@ -23,12 +23,6 @@
 import torch.nn.functional as F
 from torch import nn
 from transformers import PretrainedConfig
-from vllm.model_executor.layers.linear import (
-    MergedColumnParallelLinear,
-    QKVParallelLinear,
-    ReplicatedLinear,
-    RowParallelLinear,
-)
 from vllm.model_executor.layers.rotary_embedding import get_rope
 
 from sglang.srt.distributed import (
@@ -37,6 +31,11 @@
 )
 from sglang.srt.layers.activation import SiluAndMul
 from sglang.srt.layers.layernorm import RMSNorm
+from sglang.srt.layers.linear import (
+    QKVParallelLinear,
+    ReplicatedLinear,
+    RowParallelLinear,
+)
 from sglang.srt.layers.logits_processor import LogitsProcessor, LogitsProcessorOutput
 from sglang.srt.layers.moe.fused_moe_triton import FusedMoE
 from sglang.srt.layers.quantization.base_config import QuantizationConfig

python/sglang/srt/models/qwen2_vl.py

@@ -22,6 +22,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 """Inference-only Qwen2-VL model compatible with HuggingFace weights."""
+import logging
 from functools import lru_cache, partial
 from typing import Iterable, List, Optional, Tuple, Type, TypedDict
 
@@ -30,7 +31,6 @@
 import torch.nn as nn
 import torch.nn.functional as F
 from einops import rearrange, repeat
-from vllm.logger import init_logger
 from vllm.model_executor.layers.activation import QuickGELU
 
 from sglang.srt.configs import Qwen2VLConfig, Qwen2VLVisionConfig
@@ -50,7 +50,7 @@
 from sglang.srt.model_loader.weight_utils import default_weight_loader
 from sglang.srt.models.qwen2 import Qwen2Model
 
-logger = init_logger(__name__)
+logger = logging.getLogger(__name__)
 
 # === Vision Inputs === #
 

python/sglang/srt/models/xverse.py

@@ -21,16 +21,16 @@
 import torch
 from torch import nn
 from transformers import LlamaConfig
-from vllm.model_executor.layers.activation import SiluAndMul
-from vllm.model_executor.layers.layernorm import RMSNorm
-from vllm.model_executor.layers.linear import (
+from vllm.model_executor.layers.rotary_embedding import get_rope
+
+from sglang.srt.distributed import get_tensor_model_parallel_world_size
+from sglang.srt.layers.activation import SiluAndMul
+from sglang.srt.layers.layernorm import RMSNorm
+from sglang.srt.layers.linear import (
     MergedColumnParallelLinear,
     QKVParallelLinear,
     RowParallelLinear,
 )
-from vllm.model_executor.layers.rotary_embedding import get_rope
-
-from sglang.srt.distributed import get_tensor_model_parallel_world_size
 from sglang.srt.layers.logits_processor import LogitsProcessor
 from sglang.srt.layers.quantization.base_config import QuantizationConfig
 from sglang.srt.layers.radix_attention import RadixAttention

python/sglang/srt/models/xverse_moe.py

@@ -18,21 +18,21 @@
 import torch
 from torch import nn
 from transformers import PretrainedConfig
-from vllm.model_executor.layers.activation import SiluAndMul
-from vllm.model_executor.layers.layernorm import RMSNorm
-from vllm.model_executor.layers.linear import (
-    MergedColumnParallelLinear,
-    QKVParallelLinear,
-    ReplicatedLinear,
-    RowParallelLinear,
-)
 from vllm.model_executor.layers.rotary_embedding import get_rope
 
 from sglang.srt.distributed import (
     get_tensor_model_parallel_rank,
     get_tensor_model_parallel_world_size,
     tensor_model_parallel_all_reduce,
 )
+from sglang.srt.layers.activation import SiluAndMul
+from sglang.srt.layers.layernorm import RMSNorm
+from sglang.srt.layers.linear import (
+    MergedColumnParallelLinear,
+    QKVParallelLinear,
+    ReplicatedLinear,
+    RowParallelLinear,
+)
 from sglang.srt.layers.logits_processor import LogitsProcessor
 from sglang.srt.layers.moe.fused_moe_triton import fused_moe
 from sglang.srt.layers.quantization.base_config import QuantizationConfig