EngineX/xc-llm-ascend
3
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

[MM][Model] Remove Qwen2-VL modeling files (#4534)

Browse Source 
### What this PR does / why we need it?

Following https://github.com/vllm-project/vllm-ascend/pull/4349, remove
Qwen2-VL modeling files.


- vLLM version: v0.11.2
- vLLM main: https://github.com/vllm-project/vllm/commit/v0.11.2

---------

Signed-off-by: shen-shanshan <467638484@qq.com>
This commit is contained in:
Shanshan Shen
2025-11-29 18:07:01 +08:00
committed by GitHub
parent 6664a4e5ce
commit 2a19215e5f
4 changed files with 212 additions and 590 deletions
Download Patch File Download Diff File Expand all files Collapse all files

200
tests/ut/models/test_qwen2_vl.py
View File

@@ -1,200 +0,0 @@
import pytest
import torch
from pytest_mock import MockerFixture
from vllm.model_executor.layers.activation import QuickGELU
from tests.ut.base import PytestBase
from vllm_ascend.models.qwen2_vl import (AscendQwen2VisionAttention,
AscendQwen2VisionBlock)
class TestAscendQwen2VisionAttention(PytestBase):
def init_attention(
self,
mocker,
embed_dim=1000,
num_heads=10,
projection_size=100,
quant_config=None,
prefix="",
):
mocker_attn = mocker.patch(
"vllm_ascend.models.qwen2_vl.Qwen2VisionAttention.__init__")
attention = AscendQwen2VisionAttention(
embed_dim=embed_dim,
num_heads=num_heads,
projection_size=projection_size,
quant_config=quant_config,
prefix=prefix,
)
args, kwargs = mocker_attn.call_args
assert args == (embed_dim, num_heads, projection_size, None, "")
assert not kwargs
attention.num_attention_heads_per_partition = num_heads
return attention
def test_attn_init_should_normal(self, mocker: MockerFixture):
embed_dim = 1000
num_heads = 10
projection_size = 100
quant_config = None
prefix = ""
vit = self.init_attention(
embed_dim=embed_dim,
num_heads=num_heads,
projection_size=projection_size,
quant_config=quant_config,
prefix=prefix,
mocker=mocker,
)
assert vit.hidden_size_per_attention_head == 10
def test_attn_init_should_raise_error(self, mocker: MockerFixture):
embed_dim = 1000
num_heads = 7
projection_size = 100
quant_config = None
prefix = ""
with pytest.raises(AssertionError):
# projection_size should divided by num heads
self.init_attention(
mocker=mocker,
embed_dim=embed_dim,
num_heads=num_heads,
projection_size=projection_size,
quant_config=quant_config,
prefix=prefix,
)
def test_attn_forward(self, mocker: MockerFixture):
attention = self.init_attention(mocker=mocker)
mocker.patch("torch.nn.Module.__setattr__")
mocker.patch("torch.nn.Module.__getattr__")
mocker.patch("torch.nn.Module.__delattr__")
x = torch.rand((100, 3, 10 * 3 * 128)) # s,b, head*3*head_dim
cu_seqlens = torch.tensor([10, 50, 100])
cos = torch.rand((1, 100, 1, 128))
sin = torch.rand((1, 100, 1, 128))
qkv = lambda x: (x, 0) # noqa
split_qkv = lambda x: [ #noqa
torch.rand((100, 3, 10, 128)) for i in range(3)
] # noqa
npu_rotary_mul = lambda q, cos, sin: q # noqa
_npu_flash_attention_unpad = lambda **kwargs: kwargs["out"] # noqa
proj = lambda x: (x, 0) # noqa
mocker_qkv = mocker.patch.object(attention, "qkv", side_effect=qkv)
mocker_split_qkv = mocker.patch.object(
attention,
"split_qkv",
side_effect=split_qkv,
)
mocker_npu_rotary_mul = mocker.patch("torch_npu.npu_rotary_mul",
side_effect=npu_rotary_mul)
mocker_npu_flash_attention_unpad = mocker.patch(
"torch_npu._npu_flash_attention_unpad",
side_effect=_npu_flash_attention_unpad,
)
mocker_proj = mocker.patch.object(attention, "proj", side_effect=proj)
attention.__dict__["qkv"] = mocker_qkv
attention.__dict__["split_qkv"] = mocker_split_qkv
attention.__dict__["npu_rotary_mul"] = mocker_npu_rotary_mul
attention.__dict__["_npu_flash_attention_unpad"] = (
mocker_npu_flash_attention_unpad)
attention.__dict__["proj"] = mocker_proj
output = attention.forward(
x=x,
cu_seqlens=cu_seqlens,
cos=cos,
sin=sin,
)
qkv_args, qkv_kwargs = mocker_qkv.call_args
assert qkv_args == (x, )
assert not qkv_kwargs
split_qkv_args, split_qkv_kwargs = mocker_split_qkv.call_args
assert split_qkv_args == (x, )
assert not split_qkv_kwargs
npu_rotary_mul_args, npu_rotary_mul_kwargs = mocker_npu_rotary_mul.call_args
assert npu_rotary_mul_args[1:] == (cos, sin)
assert npu_rotary_mul_args[0].shape == torch.Size([3, 100, 10, 128])
assert not npu_rotary_mul_kwargs
assert output.shape == torch.Size([100, 3, 1280])
class TestAscendQwen2VisionBlock(PytestBase):
def init_vision_block(
self,
mocker,
dim=100,
num_heads=10,
mlp_ratio=0.5,
):
mocker_vit = mocker.patch(
"vllm.model_executor.models.qwen2_vl.Qwen2VisionBlock.__init__",
return_value=None,
)
mocker_attn = mocker.patch(
"vllm_ascend.models.qwen2_vl.AscendQwen2VisionAttention.__init__",
return_value=None,
)
mocker.patch("torch.nn.Module.__setattr__")
mocker.patch("torch.nn.Module.__getattr__")
mocker.patch("torch.nn.Module.__delattr__")
vision_block = AscendQwen2VisionBlock(
dim=dim,
num_heads=num_heads,
mlp_ratio=mlp_ratio,
)
args, kwargs = mocker_vit.call_args
assert args == (dim, num_heads, mlp_ratio, QuickGELU, None, None, "")
assert not kwargs
args1, kwargs1 = mocker_attn.call_args
assert not args1
assert kwargs1 == {
"embed_dim": dim,
"num_heads": num_heads,
"projection_size": dim,
"quant_config": None,
"prefix": ".attn",
}
return vision_block
def test_init_vision_block_should_normal(
self,
mocker: MockerFixture,
):
vision_block = self.init_vision_block(mocker)
assert isinstance(vision_block, AscendQwen2VisionBlock)
def test_vision_block_forward(self, mocker: MockerFixture):
x = torch.randint(1, 100, (100, 3, 1280)) # s,b,d
cu_seqlens = torch.tensor([10, 50, 100])
cos = torch.rand((1, 100, 1, 128))
sin = torch.rand((1, 100, 1, 128))
vision_block = self.init_vision_block(mocker)
mocker_attn = mocker.patch.object(vision_block, "attn", return_value=x)
mocker_mlp = mocker.patch.object(vision_block, "mlp", return_value=x)
vision_block.__dict__["attn"] = mocker_attn
vision_block.__dict__["mlp"] = mocker_mlp
output = vision_block.forward(x.clone(), cu_seqlens, cos, sin)
_, attn_kwargs = mocker_attn.call_args
assert attn_kwargs == {
"cu_seqlens": cu_seqlens,
"cos": cos,
"sin": sin,
}
assert torch.all(x * 3 == output)

4
vllm_ascend/models/__init__.py
View File

@@ -2,10 +2,6 @@ from vllm import ModelRegistry
def register_model():
ModelRegistry.register_model(
"Qwen2VLForConditionalGeneration",
"vllm_ascend.models.qwen2_vl:AscendQwen2VLForConditionalGeneration")
ModelRegistry.register_model(
"Qwen3VLMoeForConditionalGeneration",
"vllm_ascend.models.qwen3_vl:AscendQwen3VLMoeForConditionalGeneration")

373
vllm_ascend/models/qwen2_vl.py
View File

@@ -1,373 +0,0 @@
#
# Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
# Copyright 2023 The vLLM team.
#
#
# 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
#
# http://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.
# Adapted from vllm/model_executor/models/qwen2_vl.py
# This file is a part of the vllm-ascend project.
from collections.abc import Iterable
from functools import partial
from typing import Callable, Optional, Set, Tuple, Type
import torch
import torch.nn as nn
import torch_npu
from einops import rearrange
from transformers.models.qwen2_vl.configuration_qwen2_vl import \
Qwen2VLVisionConfig
from vllm.config import VllmConfig
from vllm.distributed import utils as dist_utils
from vllm.model_executor.layers.activation import QuickGELU
from vllm.model_executor.layers.quantization import QuantizationConfig
from vllm.model_executor.model_loader.weight_utils import default_weight_loader
from vllm.model_executor.models.qwen2_vl import (
Qwen2VisionAttention, Qwen2VisionBlock, Qwen2VisionPatchEmbed,
Qwen2VisionTransformer, Qwen2VLDummyInputsBuilder,
Qwen2VLForConditionalGeneration, Qwen2VLMultiModalProcessor,
Qwen2VLProcessingInfo)
from vllm.model_executor.models.utils import maybe_prefix
from vllm.model_executor.models.vision import conv3d_to_linear_weight
from vllm.multimodal import MULTIMODAL_REGISTRY
from vllm_ascend.utils import ACL_FORMAT_FRACTAL_ND, is_enable_nz
MIN_PAD_SIZE = 64 # min_size to pad weight
MAX_PAD_SIZE = 128 # max_size to pad weight
class AscendQwen2VisionAttention(Qwen2VisionAttention):
def __init__(
self,
embed_dim: int,
num_heads: int,
projection_size: int,
quant_config: Optional[QuantizationConfig] = None,
prefix: str = "",
) -> None:
super().__init__(
embed_dim,
num_heads,
projection_size,
quant_config,
prefix,
)
self.cu_seqlens = None
self.hidden_size_per_attention_head = dist_utils.divide(
projection_size, num_heads)
self.origin_hidden_size_per_attention_head = self.hidden_size_per_attention_head
if self.hidden_size_per_attention_head > MIN_PAD_SIZE and self.hidden_size_per_attention_head < MAX_PAD_SIZE:
self.hidden_size_per_attention_head = MAX_PAD_SIZE
def forward(
self,
x: torch.Tensor,
cu_seqlens: torch.Tensor,
cos: torch.Tensor,
sin: torch.Tensor,
) -> torch.Tensor:
self.cu_seqlens = cu_seqlens
# [s, b, c] --> [s, b, 3 * head * head_dim]
x, _ = self.qkv(x)
# [s, b, 3 * head * head_dim] -> 3 * [s, b, head, head_dim]
q, k, v = self.split_qkv(x)
batch_size = q.shape[1]
q, k, v = [
rearrange(x, "s b ... -> b s ...").contiguous() for x in (q, k, v)
]
q = torch_npu.npu_rotary_mul(q, cos, sin)
k = torch_npu.npu_rotary_mul(k, cos, sin)
q, k, v = [
rearrange(x, "b s h d -> (b s) h d").contiguous()
for x in (q, k, v)
]
context_layer = torch.empty_like(q)
# operator requires pta version >= 2.5.1
torch_npu._npu_flash_attention_unpad(
query=q,
key=k,
value=v,
seq_len=self.cu_seqlens,
scale_value=self.origin_hidden_size_per_attention_head**-0.5,
num_heads=self.num_attention_heads_per_partition,
num_kv_heads=self.num_attention_heads_per_partition,
out=context_layer)
context_layer = rearrange(context_layer,
"(b s) h d -> s b (h d)",
b=batch_size).contiguous()
output, _ = self.proj(context_layer)
return output
class AscendQwen2VisionBlock(Qwen2VisionBlock):
def __init__(
self,
dim: int,
num_heads: int,
mlp_ratio: float,
act_layer: Type[nn.Module] = QuickGELU,
norm_layer: Optional[Callable[[int], nn.Module]] = None,
quant_config: Optional[QuantizationConfig] = None,
prefix: str = "",
) -> None:
super().__init__(dim, num_heads, mlp_ratio, act_layer, norm_layer,
quant_config, prefix)
self.attn = AscendQwen2VisionAttention(embed_dim=dim,
num_heads=num_heads,
projection_size=dim,
quant_config=quant_config,
prefix=f"{prefix}.attn")
def forward(
self,
x: torch.Tensor,
cu_seqlens: torch.Tensor,
cos: torch.Tensor,
sin: torch.Tensor,
) -> torch.Tensor:
x = x + self.attn(
self.norm1(x),
cu_seqlens=cu_seqlens,
cos=cos,
sin=sin,
)
x = x + self.mlp(self.norm2(x))
return x
class AscendQwen2VisionPatchEmbed(Qwen2VisionPatchEmbed):
def forward(self, x: torch.Tensor) -> torch.Tensor:
x = x.matmul(
self.proj.weight.data.view(self.embed_dim, -1).transpose(0, 1))
return x
class AscendQwen2VisionTransformer(Qwen2VisionTransformer):
def __init__(
self,
vision_config: Qwen2VLVisionConfig,
norm_eps: float = 1e-6,
quant_config: Optional[QuantizationConfig] = None,
prefix: str = "",
interleaved=False,
) -> None:
super().__init__(vision_config, norm_eps, quant_config, prefix)
self.interleaved = interleaved
self.enable_pad = False
self.depth = vision_config.depth
self.hidden_size = vision_config.embed_dim
self.num_heads = vision_config.num_heads
self.patch_embed = AscendQwen2VisionPatchEmbed(
patch_size=vision_config.patch_size,
temporal_patch_size=vision_config.temporal_patch_size,
in_channels=vision_config.in_channels,
embed_dim=vision_config.embed_dim,
)
self.blocks = nn.ModuleList([
AscendQwen2VisionBlock(dim=self.embed_dim,
num_heads=self.num_heads,
mlp_ratio=vision_config.mlp_ratio,
norm_layer=partial(nn.LayerNorm,
eps=norm_eps),
quant_config=quant_config,
prefix=f"{prefix}.blocks.{layer_idx}")
for layer_idx in range(vision_config.depth)
])
self.hidden_size_per_attention_head = dist_utils.divide(
self.hidden_size, self.num_heads)
if self.hidden_size_per_attention_head > MIN_PAD_SIZE and self.hidden_size_per_attention_head < MAX_PAD_SIZE:
self.enable_pad = True
self.origin_hidden_size_per_attention_head = self.hidden_size_per_attention_head
self.half_origin_hidden_size_per_attention_head = self.hidden_size_per_attention_head // 2
self.half_pad_hidden_size_per_attention_head = (
MAX_PAD_SIZE - self.hidden_size_per_attention_head) // 2
self.hidden_size_per_attention_head = MAX_PAD_SIZE
def cal_cos_sin(self, rotary_pos_emb):
cos = rotary_pos_emb.cos() # [seqlen, rotary_dim / 2]
sin = rotary_pos_emb.sin()
if self.enable_pad:
cos = torch.nn.functional.pad(
cos, (0, self.half_pad_hidden_size_per_attention_head))
sin = torch.nn.functional.pad(
sin, (0, self.half_pad_hidden_size_per_attention_head))
if not self.interleaved:
cos_new = torch.cat((cos, cos), dim=-1)
sin_new = torch.cat((sin, sin), dim=-1)
else:
cos_new = rearrange(torch.stack((cos, cos), dim=-1),
"... d two -> ...(d two)",
two=2)
sin_new = rearrange(torch.stack((sin, sin), dim=-1),
"... d two -> ...(d two)",
two=2)
cos_new = cos_new.reshape(1, -1, 1,
self.hidden_size_per_attention_head)
sin_new = sin_new.reshape(1, -1, 1,
self.hidden_size_per_attention_head)
return cos_new, sin_new
def pad_qkv_bias(self, bias):
first_half = bias.reshape(
-1, 3, self.origin_hidden_size_per_attention_head
)[:, :, :self.half_origin_hidden_size_per_attention_head]
second_half = bias.reshape(
-1, 3, self.origin_hidden_size_per_attention_head
)[:, :, self.half_origin_hidden_size_per_attention_head:]
first_half_padded = torch.nn.functional.pad(
first_half, (0, self.half_pad_hidden_size_per_attention_head))
second_half_padded = torch.nn.functional.pad(
second_half, (0, self.half_pad_hidden_size_per_attention_head))
bias_padded = torch.cat([first_half_padded, second_half_padded], dim=2)
bias_final = bias_padded.reshape(-1)
return bias_final
def pad_qkv_weight(self, data):
qkv_weight_first_half = data.reshape(
-1, 3, self.origin_hidden_size_per_attention_head, self.hidden_size
)[:, :, :self.half_origin_hidden_size_per_attention_head, :]
qkv_weight_second_half = data.reshape(
-1, 3, self.origin_hidden_size_per_attention_head, self.hidden_size
)[:, :, self.half_origin_hidden_size_per_attention_head:, :]
qkv_weight_first_half_padded = torch.nn.functional.pad(
qkv_weight_first_half,
(0, 0, 0, self.half_pad_hidden_size_per_attention_head))
qkv_weight_second_half_padded = torch.nn.functional.pad(
qkv_weight_second_half,
(0, 0, 0, self.half_pad_hidden_size_per_attention_head))
qkv_weight_padded = torch.cat(
[qkv_weight_first_half_padded, qkv_weight_second_half_padded],
dim=2)
qkv_weight_final = qkv_weight_padded.reshape(-1, self.hidden_size)
if is_enable_nz():
qkv_weight_final_copy = torch.empty_like(qkv_weight_final).copy_(
qkv_weight_final)
qkv_weight_final_copy = torch_npu.npu_format_cast(
qkv_weight_final_copy, ACL_FORMAT_FRACTAL_ND)
return qkv_weight_final_copy
return qkv_weight_final
def pad_proj_weight(self, data):
out_weight = torch.nn.functional.pad(
data.reshape(self.hidden_size, -1,
self.half_origin_hidden_size_per_attention_head),
(0, self.half_pad_hidden_size_per_attention_head, 0, 0)).reshape(
self.hidden_size, -1)
if is_enable_nz():
out_weight_copy = torch.empty_like(out_weight).copy_(out_weight)
out_weight_copy = torch_npu.npu_format_cast(
out_weight_copy, ACL_FORMAT_FRACTAL_ND)
return out_weight_copy
return out_weight
def load_weights(self, weights: Iterable[Tuple[str,
torch.Tensor]]) -> Set[str]:
stacked_params_mapping = [
# (param_name, shard_name, shard_id)
("qkv_proj", "q_proj", "q"),
("qkv_proj", "k_proj", "k"),
("qkv_proj", "v_proj", "v"),
]
params_dict = dict(self.named_parameters(remove_duplicate=False))
loaded_params: Set[str] = set()
for name, loaded_weight in weights:
if name.endswith("patch_embed.proj.weight"):
loaded_weight = conv3d_to_linear_weight(loaded_weight)
for (param_name, weight_name, shard_id) in stacked_params_mapping:
if weight_name not in name:
continue
name = name.replace(weight_name, param_name)
param = params_dict[name]
weight_loader = param.weight_loader
weight_loader(param, loaded_weight, shard_id)
break
else:
param = params_dict[name]
weight_loader = getattr(param, "weight_loader",
default_weight_loader)
weight_loader(param, loaded_weight)
if ("attn.proj.weight" in name) and self.enable_pad:
param.data = self.pad_proj_weight(param.data)
if ("attn.qkv.weight" in name) and self.enable_pad:
param.data = self.pad_qkv_weight(param.data)
if ("attn.qkv.bias" in name) and self.enable_pad:
param.data = self.pad_qkv_bias(param.data)
loaded_params.add(name)
return loaded_params
def forward(
self,
x: torch.Tensor,
grid_thw: torch.Tensor,
) -> torch.Tensor:
grid_thw = torch.tensor(grid_thw, dtype=torch.int32)
# compute cu_seqlens and avoid cumsum to fit operator unpadFA
cu_seqlens = torch.repeat_interleave(grid_thw[:, 1] * grid_thw[:, 2],
grid_thw[:,
0]).cpu().to(torch.int32)
# patchify
x = x.to(device=self.device, dtype=self.dtype)
x = self.patch_embed(x)
# compute position embedding
rotary_pos_emb = self.rot_pos_emb(grid_thw)
cos, sin = self.cal_cos_sin(rotary_pos_emb)
x = x.unsqueeze(1)
for blk in self.blocks:
x = blk(x, cu_seqlens=cu_seqlens, cos=cos, sin=sin)
# adapter
x = self.merger(x)
return x
@MULTIMODAL_REGISTRY.register_processor(Qwen2VLMultiModalProcessor,
info=Qwen2VLProcessingInfo,
dummy_inputs=Qwen2VLDummyInputsBuilder)
class AscendQwen2VLForConditionalGeneration(Qwen2VLForConditionalGeneration):
def __init__(self, *, vllm_config: VllmConfig, prefix: str = ""):
super().__init__(vllm_config=vllm_config, prefix=prefix)
self.visual = AscendQwen2VisionTransformer(
self.config.vision_config,
norm_eps=getattr(self.config, "rms_norm_eps", 1e-6),
quant_config=vllm_config.quant_config,
prefix=maybe_prefix(prefix, "visual"),
)

225
vllm_ascend/patch/worker/patch_qwen2_5_vl.py
View File

@@ -24,17 +24,27 @@ import torch.nn.functional as F
import torch_npu
from transformers.models.qwen2_5_vl.configuration_qwen2_5_vl import \
Qwen2_5_VLVisionConfig
from transformers.models.qwen2_vl.configuration_qwen2_vl import \
Qwen2VLVisionConfig
from vllm.attention.backends.registry import AttentionBackendEnum
from vllm.attention.layer import maybe_get_vit_flash_attn_backend
from vllm.attention.layer import (check_upstream_fa_availability,
maybe_get_vit_flash_attn_backend)
from vllm.model_executor.layers.activation import get_act_and_mul_fn
from vllm.model_executor.layers.layernorm import RMSNorm
from vllm.model_executor.layers.quantization import QuantizationConfig
from vllm.model_executor.layers.rotary_embedding import get_rope
from vllm.model_executor.layers.rotary_embedding.common import (
apply_rotary_emb_torch, dispatch_rotary_emb_function)
from vllm.model_executor.models.qwen2_5_vl import (
Qwen2_5_VisionAttention, Qwen2_5_VisionBlock, Qwen2_5_VisionPatchEmbed,
Qwen2_5_VisionPatchMerger, Qwen2_5_VisionTransformer,
Qwen2_5_VLForConditionalGeneration, Qwen2_5_VLImageInputs,
Qwen2_5_VLVideoInputs)
from vllm.model_executor.models.qwen2_vl import (Qwen2VisionAttention,
Qwen2VisionBlock,
Qwen2VisionPatchEmbed,
Qwen2VisionPatchMerger,
Qwen2VisionTransformer)
from vllm.model_executor.models.utils import cast_overflow_tensors
from vllm.model_executor.models.vision import (
get_vit_attn_backend, run_dp_sharded_mrope_vision_model)
@@ -74,18 +84,8 @@ class AscendQwen2_5_VisionAttention(nn.Module):
# Convert cumulative tensor to intervals and move it to cpu.
cu_seqlens = torch.diff(cu_seqlens).to("cpu")
cos = rotary_pos_emb_cos
sin = rotary_pos_emb_sin
cos = einops.rearrange(
torch.stack((cos, cos), dim=-1),
"... d two -> ...(d two)",
two=2,
)
sin = einops.rearrange(
torch.stack((sin, sin), dim=-1),
"... d two -> ...(d two)",
two=2,
)
cos = torch.cat((rotary_pos_emb_cos, rotary_pos_emb_cos), dim=-1)
sin = torch.cat((rotary_pos_emb_sin, rotary_pos_emb_sin), dim=-1)
cos = cos.reshape(1, -1, 1, self.hidden_size_per_attention_head)
sin = sin.reshape(1, -1, 1, self.hidden_size_per_attention_head)
q = torch_npu.npu_rotary_mul(q, cos, sin)
@@ -132,6 +132,191 @@ class AscendQwen2_5_VisionAttention(nn.Module):
return output
class AscendQwen2VisionBlock(nn.Module):
def forward(
self,
x: torch.Tensor,
cu_seqlens: torch.Tensor,
rotary_pos_emb_cos: torch.Tensor,
rotary_pos_emb_sin: torch.Tensor,
max_seqlen: int | None = None, # Only used for Flash Attention
seqlens: list[int] | None = None, # Only used for xFormers
) -> torch.Tensor:
x = x + self.attn(
self.norm1(x),
cu_seqlens=cu_seqlens,
rotary_pos_emb_cos=rotary_pos_emb_cos,
rotary_pos_emb_sin=rotary_pos_emb_sin,
max_seqlen=max_seqlen,
seqlens=seqlens,
)
x = x + self.mlp(self.norm2(x))
return x
class AscendQwen2VisionTransformer(nn.Module):
def __init__(
self,
vision_config: Qwen2VLVisionConfig,
norm_eps: float = 1e-6,
quant_config: QuantizationConfig | None = None,
prefix: str = "",
use_data_parallel: bool = False,
attn_backend_override: AttentionBackendEnum | None = None,
) -> None:
nn.Module.__init__(self)
patch_size = vision_config.patch_size
temporal_patch_size = vision_config.temporal_patch_size
spatial_merge_size = vision_config.spatial_merge_size
in_channels = vision_config.in_channels
hidden_size = vision_config.hidden_size
embed_dim = vision_config.embed_dim
depth = vision_config.depth
num_heads = vision_config.num_heads
mlp_ratio = vision_config.mlp_ratio
self.use_data_parallel = use_data_parallel
self.out_hidden_size = vision_config.hidden_size
self.spatial_merge_size = spatial_merge_size
self.num_heads = num_heads
self.embed_dim = embed_dim
self.patch_embed = Qwen2VisionPatchEmbed(
patch_size=patch_size,
temporal_patch_size=temporal_patch_size,
in_channels=in_channels,
embed_dim=embed_dim,
)
norm_layer = partial(nn.LayerNorm, eps=norm_eps)
head_dim = embed_dim // num_heads
self.rotary_pos_emb = get_rope(
head_size=head_dim,
rotary_dim=head_dim // 2,
max_position=8192,
base=10000.0,
is_neox_style=True,
)
self.blocks = nn.ModuleList([
Qwen2VisionBlock(
dim=embed_dim,
num_heads=num_heads,
mlp_ratio=mlp_ratio,
norm_layer=norm_layer,
quant_config=quant_config,
prefix=f"{prefix}.blocks.{layer_idx}",
use_data_parallel=use_data_parallel,
attn_backend_override=attn_backend_override,
) for layer_idx in range(depth)
])
self.merger = Qwen2VisionPatchMerger(
d_model=hidden_size,
context_dim=embed_dim,
norm_layer=norm_layer,
quant_config=quant_config,
prefix=f"{prefix}.merger",
use_data_parallel=use_data_parallel,
)
self.attn_backend = get_vit_attn_backend(
head_size=head_dim,
dtype=torch.get_default_dtype(),
attn_backend_override=attn_backend_override,
)
if (self.attn_backend != AttentionBackendEnum.FLASH_ATTN
and check_upstream_fa_availability(torch.get_default_dtype())):
self.attn_backend = AttentionBackendEnum.FLASH_ATTN
def rot_pos_emb(
self,
grid_thw: list[list[int]]) -> tuple[torch.Tensor, torch.Tensor]:
pos_ids = []
max_grid_size = 0
for t, h, w in grid_thw:
hpos_ids = torch.arange(h).unsqueeze(1).expand(-1, w)
wpos_ids = torch.arange(w).unsqueeze(0).expand(h, -1)
hpos_ids = (hpos_ids.reshape(
h // self.spatial_merge_size,
self.spatial_merge_size,
w // self.spatial_merge_size,
self.spatial_merge_size,
).permute(0, 2, 1, 3).flatten())
wpos_ids = (wpos_ids.reshape(
h // self.spatial_merge_size,
self.spatial_merge_size,
w // self.spatial_merge_size,
self.spatial_merge_size,
).permute(0, 2, 1, 3).flatten())
pos_ids.append(
torch.stack([hpos_ids, wpos_ids], dim=-1).repeat(t, 1))
max_grid_size = max(max_grid_size, h, w)
pos_ids = torch.cat(pos_ids, dim=0)
# Use pre-computed cos_sin_cache from RotaryEmbedding
cos, sin = self.rotary_pos_emb.get_cos_sin(max_grid_size)
# (num_tokens, rotary_dim // 2)
cos_h = cos[pos_ids[:, 0]] # type: ignore
cos_w = cos[pos_ids[:, 1]] # type: ignore
sin_h = sin[pos_ids[:, 0]] # type: ignore
sin_w = sin[pos_ids[:, 1]] # type: ignore
cos_combined = torch.cat([cos_h, cos_w], dim=-1)
sin_combined = torch.cat([sin_h, sin_w], dim=-1)
return cos_combined, sin_combined
def forward(
self,
x: torch.Tensor,
grid_thw: torch.Tensor | list[list[int]],
) -> torch.Tensor:
# patchify
x = x.to(device=self.device, dtype=self.dtype)
x = self.patch_embed(x)
if isinstance(grid_thw, list):
grid_thw_list = grid_thw
grid_thw = torch.tensor(grid_thw, dtype=torch.int32)
else:
grid_thw_list = grid_thw.tolist()
# compute position embedding
rotary_pos_emb_cos, rotary_pos_emb_sin = self.rot_pos_emb(
grid_thw_list)
# compute cu_seqlens
cu_seqlens = torch.repeat_interleave(grid_thw[:, 1] * grid_thw[:, 2],
grid_thw[:, 0]).cumsum(
dim=0, dtype=torch.int32)
cu_seqlens = torch.cat([cu_seqlens.new_zeros(1), cu_seqlens])
cu_seqlens = cu_seqlens.to(self.device, non_blocking=True)
# transformers
x = x.unsqueeze(1)
# pre-compute seqlens for attn mask to reduce cuMemcpy operations
max_seqlen, seqlens = self.compute_attn_mask_seqlen(cu_seqlens)
for blk in self.blocks:
x = blk(
x,
cu_seqlens=cu_seqlens,
rotary_pos_emb_cos=rotary_pos_emb_cos,
rotary_pos_emb_sin=rotary_pos_emb_sin,
max_seqlen=max_seqlen,
seqlens=seqlens,
)
# adapter
x = self.merger(x)
return x
class AscendQwen2_5_VisionBlock(nn.Module):
def forward(
@@ -486,7 +671,16 @@ class AscendQwen2_5_VLForConditionalGeneration(nn.Module):
return video_embeds.split(sizes)
def _apply_rotary_pos_emb_vision(t: torch.Tensor, cos: torch.Tensor,
sin: torch.Tensor) -> torch.Tensor:
rotary_emb_function = dispatch_rotary_emb_function(
default=partial(apply_rotary_emb_torch, is_neox_style=True))
output = rotary_emb_function(t, cos, sin).type_as(t)
return output
# NOTE: This will be removed after MMEncoderAttention has been extract as a CustomOp in vllm.
Qwen2VisionAttention.forward = AscendQwen2_5_VisionAttention.forward
Qwen2_5_VisionAttention.forward = AscendQwen2_5_VisionAttention.forward
# NOTE: These will be removed after https://github.com/vllm-project/vllm/pull/29388 is merged.
@@ -494,8 +688,13 @@ Qwen2_5_VLForConditionalGeneration._process_image_input = AscendQwen2_5_VLForCon
Qwen2_5_VLForConditionalGeneration._process_video_input = AscendQwen2_5_VLForConditionalGeneration._process_video_input
# NOTE: These will be removed after vllm-ascend is aligned with vllm latest main.
Qwen2VisionBlock.forward = AscendQwen2VisionBlock.forward
Qwen2VisionTransformer.__init__ = AscendQwen2VisionTransformer.__init__
Qwen2VisionTransformer.rot_pos_emb = AscendQwen2VisionTransformer.rot_pos_emb
Qwen2VisionTransformer.forward = AscendQwen2VisionTransformer.forward
Qwen2_5_VisionBlock.forward = AscendQwen2_5_VisionBlock.forward
Qwen2_5_VisionTransformer.__init__ = AscendQwen2_5_VisionTransformer.__init__
Qwen2_5_VisionTransformer.rotary_pos_emb_thw = AscendQwen2_5_VisionTransformer.rotary_pos_emb_thw
Qwen2_5_VisionTransformer.get_rope_by_thw = AscendQwen2_5_VisionTransformer.get_rope_by_thw
Qwen2_5_VisionTransformer.forward = AscendQwen2_5_VisionTransformer.forward
apply_rotary_pos_emb_vision = _apply_rotary_pos_emb_vision