ecf2fa482e
### What this PR does / why we need it?
The initialization method of expert_map used by the eplb module is
different from that used by the fused_moe module. This PR deletes the
expert_map initialization method used by the eplb module to make the
initialization methods consistent.
#### before bugfix
self._expert_map=tensor([64, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
-1, -1, -1, -1, -1, -1, 1, -1, -1, -1, -1, -1, -1, -1, -1, -1, 0, 1, 2,
3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22,
23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40,
41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58,
59, 60, 61,62, 63], device='npu:1', dtype=torch.int32)
self.shared_dict["expert_maps"][0]=tensor([-1, -1, -1, -1, -1, -1, -1,
-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1,
-1, -1, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17,
18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35,
36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53,
54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64]], dtype=torch.int32)
### How was this patch tested?
#### qwen3-235B-w8a8 aime
| dataset | version | metric | mode | vllm-api-general-chat |
|----- | ----- | ----- | ----- | -----|
| aime2024 | 604a78 | accuracy | gen | 86.67 |
- vLLM version: v0.13.0
- vLLM main:
2f4e6548ef
Signed-off-by: shenchuxiaofugui <1311027364@qq.com>
41 lines
1.4 KiB
Python
41 lines
1.4 KiB
Python
#
|
|
# Copyright (c) 2025 Huawei Technologies Co., Ltd. All Rights Reserved.
|
|
#
|
|
# 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.
|
|
# This file is a part of the vllm-ascend project.
|
|
#
|
|
# Todo: Once https://github.com/vllm-project/vllm/issues/22246 is merged in vllm. Remove this adaptor.
|
|
from abc import abstractmethod
|
|
from typing import Any
|
|
|
|
|
|
class EplbAdaptor():
|
|
|
|
def __init__(self, **args):
|
|
pass
|
|
|
|
@abstractmethod
|
|
def get_rank_expert_workload(self):
|
|
raise NotImplementedError
|
|
|
|
@abstractmethod
|
|
def do_update_expert_map(self, layer_id: Any,
|
|
updated_expert_map: Any) -> Any:
|
|
raise NotImplementedError
|
|
|
|
@abstractmethod
|
|
def do_update_expert_weight(self, layer_id: Any,
|
|
local_expert_to_replace: Any,
|
|
buffer_tensor_id: Any) -> Any:
|
|
raise NotImplementedError
|