| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
fix: isolate dist train parallel state getters Co-authored-by: Liz<lizhi166@huawei.com> # message auto-generated for no-merge-commit merge: !3941 merge fix/dist-train-state-getter-ut into master fix: isolate dist train parallel state getters Created-by: Liz_ Commit-by: Liz Merged-by: ascend-robot Description: ## What this PR does / why we need it? 见关联issue 修复： 把测试从“绕过装饰器”改回“走公开入口” - 原来用 __wrapped__()，现在直接调用 dps.get_tensor_model_parallel_group()、dps.get_data_parallel_group()、dps.get_expert_model_parallel_group()、dps.get_expert_model_parallel_world_size()。 - 这样保留了 @subwrold_decorator 的原始测试意图。 补了装饰器所需的最小上下文 - 不再只手工改全局变量，而是构造一个 DetachedSubWorld - 然后把 _CUR_SUB_WORLD 指到这个 subworld - 让装饰器先恢复全局状态，再执行 getter ## Does this PR introduce any user-facing change? Please describe whether the PR will result in any user-facing usage changes. If there is related documentation, please specify its path. 无 ## How was this patch tested? 流水线通过 See merge request: Ascend/MindSpeed!3941 | 1 个月前 | |
fix: add open component Copyright Co-authored-by: GuoHaifeng1999<guohaifeng12@huawei.com> # message auto-generated for no-merge-commit merge: !3611 merge master into master fix: add open component Copyright Created-by: GuoHaifeng1999 Commit-by: GuoHaifeng1999 Merged-by: ascend-robot Description: ## What this PR does / why we need it? add open component Copyright ## Does this PR introduce any user-facing change? nothing ## How was this patch tested? don't needs See merge request: Ascend/MindSpeed!3611 | 3 个月前 | |
fix: restore non-load-balanced KVAllGather THD path Co-authored-by: Liz_<lizhi166@huawei.com> # message auto-generated for no-merge-commit merge: !4043 merge revert-balanced-tnd-master into master fix: restore non-load-balanced KVAllGather THD path Created-by: Liz_ Commit-by: Liz_ Merged-by: ascend-robot Description: fix:回退采用非负载均衡KVAllgather TND实现 ## What this PR does / why we need it? 回退采用非负载均衡KVAllgather TND实现 ## Does this PR introduce any user-facing change? NAN ## How was this patch tested? 自验证通过 See merge request: Ascend/MindSpeed!4043 | 9 天前 | |
feat(triton): use expert-tile sort chunk map Co-authored-by: guofanfeng<guofanfeng1@huawei.com> # message auto-generated for no-merge-commit merge: !3956 merge feat/triton-sort-chunks-expert-tile into master feat(triton): use expert-tile sort chunk map Created-by: guofanfeng23 Commit-by: guofanfeng Merged-by: ascend-robot Description: --- title: "feat(triton): use expert-tile sort chunk map" labels: ["feat"] assignees: 'guofanfeng23' --- ## What this PR does / why we need it? 修改_make_chunk_sort_map_kernel 实现方式，采用expert-tile调度方式，减少ub占用，消除ub overflow风险。同时新算法有性能优化 ## Does this PR introduce any user-facing change? No ## How was this patch tested? A2上通过精度测试和性能测试。可以见issue中说明 https://gitcode.com/Ascend/MindSpeed/issues/323 [#323](https://gitcode.com/Ascend/MindSpeed/issues/323) See merge request: Ascend/MindSpeed!3956 | 28 天前 | |
Distributed test framework | 2 年前 | |
fix【问题】use_custom_fsdp特性和dualpipev不兼容 Co-authored-by: lizhi<lizhi166@huawei.com> # message auto-generated for no-merge-commit merge: !3080 merge bug-master into master fix【问题】use_custom_fsdp特性和dualpipev不兼容 Created-by: liz_hw Commit-by: lizhi Merged-by: ascend-robot Description: fix【问题】use_custom_fsdp特性和dualpipev不兼容 See merge request: Ascend/MindSpeed!3080 | 9 个月前 | |
feature(ut): uses third-party verification for mc2 UT. Co-authored-by: Muu<koimuu@163.com> # message auto-generated for no-merge-commit merge: !3247 merge feature_mc2 into master feature(ut): uses third-party verification for mc2 UT. Created-by: Muuyo Commit-by: Muu Merged-by: ascend-robot Description: 采用三方验证 输入值域 [-5,5] 验证文档: https://wiki.huawei.com/domains/76578/wiki/233229/WIKI2026013110011445 See merge request: Ascend/MindSpeed!3247 | 7 个月前 |