【Hackathon 10th Spring No.47】【RFC】新增 MiniMax-M1 模型设计文档#1252
Open
cloudforge1 wants to merge 2 commits intoPaddlePaddle:masterfrom
Open
【Hackathon 10th Spring No.47】【RFC】新增 MiniMax-M1 模型设计文档#1252cloudforge1 wants to merge 2 commits intoPaddlePaddle:masterfrom
cloudforge1 wants to merge 2 commits intoPaddlePaddle:masterfrom
Conversation
Collaborator
Open
10 tasks
5 tasks
Collaborator
如果觉得原来的RFC没问题，可以直接提代码PR。如果觉得RFC还需要修改，在原来的基础上修改即可，无需提一个新的文件。 |
cloudforge1
force-pushed
the
task/047-rfc-minimax-m1
branch
from
792fb53 to
7abad1f
Compare
Contributor
Author
|
了解，将调整为更新原 RFC 文件，不新增文件。代码 PR 同步推进中：PaddlePaddle/FastDeploy#6994 |
Contributor
Author
|
已更新，改为直接修改原 RFC 文件。请查看。 |
cloudforge1
force-pushed
the
task/047-rfc-minimax-m1
branch
from
7abad1f to
940b245
Compare
cloudforge1
force-pushed
the
task/047-rfc-minimax-m1
branch
from
940b245 to
8210271
Compare
Contributor
Author
|
已按实际实现全面更新 RFC（同一文件名，190 行更新）。主要变更：Triton 算子替代 CUDA C++ 内核、移除 MambaBackend 抽象、补充量化支持与权重加载双路径。原始作者已在元数据中标注。 |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
概述
本 RFC 设计 FastDeploy 新增 MiniMax-M1 模型的推理支持方案。
核心挑战
MiniMax-M1 采用 80 层混合架构（标准 GQA + 线性注意力/Mamba），需要：
MambaBackend(AttentionBackend)支持线性注意力attn_type_list选择 GQA 或 Mamba 后端主要工作量分布
复用与创新
相关链接
vllm/model_executor/layers/lightning_attn.py