main

ascend-robotfix: adapt device_capability for PT212

6e42e83a创建于 17 小时前历史提交

文件	最后提交记录	最后更新时间
features_manager	feat: add fla adaptor and testcase Co-authored-by: GuoHaifeng1999<guohaifeng12@huawei.com> # message auto-generated for no-merge-commit merge: !46 merge main into main feat: add fla adaptor and testcase Created-by: GuoHaifeng1999 Commit-by: GuoHaifeng1999 Merged-by: ascend-robot Description: feat: add fla adaptor feat: add has_nvrx_async_support patch fix: optimization_level parse error fix feat: add transformer_engine_torch patch FLA算子仅提供基于torch原生小算子的实现，用于保障Megatron基础训练流程，NPU大算子实现基于Mindspeed。 导入Mindspeed时，自动覆盖Megatron的GDN类并使用NPU大算子代替MA的小算子 MA不感知Mindspeed，Mindspeed自动覆盖MA的小算子实现 测试验证： 1、测试用例通过，基础功能正确 2、实际训练验证通过，功能正确 See merge request: Ascend/MegatronAdaptor!46	5 天前
patches	fix: adapt device_capability for PT212 Co-authored-by: yulelanmei<huangyijie8@huawei.com> # message auto-generated for no-merge-commit merge: !51 merge main into main fix: adapt device_capability for PT212 Created-by: yulelanmei Commit-by: yulelanmei Merged-by: ascend-robot Description: ## What this PR does / why we need it? Unable to run training with Pytorch2.12 because interface is not compatible. ## Does this PR introduce any user-facing change? N/A ## How was this patch tested? Run any core test case with Pytorch2.12 version. See merge request: Ascend/MegatronAdaptor!51	17 小时前
utils	feat(fp8): add HIF8 format Co-authored-by: Muuyo<koimuu@163.com>	20 天前
__init__.py	feat: add attn ST case	1 个月前
megatron_adaptor.py	refactor: local branch refactor	2 个月前