文件最后提交记录最后更新时间
【feature】【refactor】优化敏感层分析的实现 Co-authored-by: zhangz200102<zhangzheng183@huawei.com> # message auto-generated for no-merge-commit merge: !206 merge refactor/analysis_processor_contents into master 【feature】【refactor】优化敏感层分析的实现 Created-by: zhangz200102 Commit-by: zhangz200102 Merged-by: ascend-robot Description: 本PR对敏感层分析的processor模块进行一定程度的重构,主要调整目录结构,优化相关实现,包含以下内容: 总览: 修改前目录结构 ![修改前目录.PNG](https://raw.gitcode.com/user-images/assets/8444818/3a363e93-7572-4dae-8b93-ab44c73f9224/修改前目录.PNG '修改前目录.PNG') 修改后目录结构 ![修改后目录.PNG](https://raw.gitcode.com/user-images/assets/8444818/98427fa7-4c9f-470f-bdd0-b900a845a9a1/修改后目录.PNG '修改后目录.PNG') - 重命名 AnalysisPipelineLoader -> YamlAnalysisPipelineLoader 基础设施体现第三方依赖,原AnalysisPipelineLoader依赖于yaml模板文件读取 - 敏感层分析结果存储位置变更:context.state -> context.debug 算法间或模块间的需要传递的信息存储到debug中;便于通过上下文机制在debug模式下自动落盘数据 - attention_mse新建独立目录 原attention_mse.py和interface.py变更为attention_mse/impl.py和attention_mse/interface.py - 一元/二元方法基类从method_base.py中分别拆分移动到各自子目录下 method_base中只保留公共逻辑,一元/二元基类(UnaryAnalysisMethod/BinaryAnalysisMethod)从method_base.py中提取分别放到新建的unary_operator/metrics/base.py和binary_operator/metrics/base.py中 - 实现泛型工厂,分别创建一元/二元方法的工厂,各自独立注册、管理。 新建unary_operator/metrics/factory.py和binary_operator/metrics/factory.py - 其他内容:相应调整import和ut文件 See merge request: Ascend/msmodelslim!2062 个月前
【feature】mse_layer_wise敏感层分析优化;摸高算法支持layer级回退 Co-authored-by: xyxin_006<xyxin_hit@163.com> # message auto-generated for no-merge-commit merge: !374 merge feature/analysis_based_mse into master 【feature】mse_layer_wise敏感层分析优化;摸高算法支持layer级回退 Created-by: xyxin_006 Commit-by: xyxin_006 Merged-by: ascend-robot Description: 感谢您贡献的Pull Request! 在提交之前,请务必阅读 [CONTRIBUTING.md](https://gitcode.com/Ascend/msmodelslim/blob/master/CONTRIBUTING.md)。 Thanks for sending a pull request! BEFORE SUBMITTING, PLEASE READ [CONTRIBUTING.md](https://gitcode.com/Ascend/msmodelslim/blob/master/CONTRIBUTING.md). ## PR描述 (What this PR does / why we need it?) 1、mse_layer_wise 实现优化:原实现计算单个layer层的敏感度针对专家模型实现不准确,修改后直接统计单个Decoder量化前后的MSELoss 2、摸高算法支持layer级回退:原实现仅支持linear级回退,针对专家模型的自动调优流程有较大局限性,优化后可以实现整层回退,缩小搜索空间,加快自动调优迭代效率 3、敏感层分析命令行优化,可以根据分析粒度选择算法,不同粒度可选范围不同。 4、资料同步修改 ## 面向用户的变更 (Does this PR introduce _any_ user-facing change)? - 请注意,这里指的是**任何**面向用户的变更,包括但不限于API、用户界面或其他使用方式上的变更。 - Note that it means *any* user-facing change including all aspects such as API, interface or other behavior changes. ## 功能验证 (How was this patch tested?) 请确认CI已通过增量及存量的单元测试用例。 如果本次测试方式与常规单元测试不同,请详细说明您的测试步骤(最好提供完整的可复现的操作路径及关键截图),以便Committer能够快速复现验证,也便于后续的维护。 如果未添加测试,请说明未添加的原因,以及为何难添加测试。 - [_] 功能自验 - [_] 本地自验截图(涉及个人标识符等敏感信息请注意脱敏) - [_] 新增/变更内容是否已新增/适配UT测试用例看护 CI passed with new added/existing test. If it was tested in a way different from regular unit tests, please clarify how you tested step by step, ideally copy and paste-able, so that other reviewers can test and check, and descendants can verify in the future. If tests were not added, please describe why they were not added and/or why it was difficult to add. - [_] Self-verification of the feature. - [_] Screenshot of local self-verification (please anonymize any sensitive information such as personal identifiers) - [_] Have new or modified unit test (UT) cases been added or adapted to cover the newly added or changed content? See merge request: Ascend/msmodelslim!37430 天前
【feature】mse_layer_wise敏感层分析优化;摸高算法支持layer级回退 Co-authored-by: xyxin_006<xyxin_hit@163.com> # message auto-generated for no-merge-commit merge: !374 merge feature/analysis_based_mse into master 【feature】mse_layer_wise敏感层分析优化;摸高算法支持layer级回退 Created-by: xyxin_006 Commit-by: xyxin_006 Merged-by: ascend-robot Description: 感谢您贡献的Pull Request! 在提交之前,请务必阅读 [CONTRIBUTING.md](https://gitcode.com/Ascend/msmodelslim/blob/master/CONTRIBUTING.md)。 Thanks for sending a pull request! BEFORE SUBMITTING, PLEASE READ [CONTRIBUTING.md](https://gitcode.com/Ascend/msmodelslim/blob/master/CONTRIBUTING.md). ## PR描述 (What this PR does / why we need it?) 1、mse_layer_wise 实现优化:原实现计算单个layer层的敏感度针对专家模型实现不准确,修改后直接统计单个Decoder量化前后的MSELoss 2、摸高算法支持layer级回退:原实现仅支持linear级回退,针对专家模型的自动调优流程有较大局限性,优化后可以实现整层回退,缩小搜索空间,加快自动调优迭代效率 3、敏感层分析命令行优化,可以根据分析粒度选择算法,不同粒度可选范围不同。 4、资料同步修改 ## 面向用户的变更 (Does this PR introduce _any_ user-facing change)? - 请注意,这里指的是**任何**面向用户的变更,包括但不限于API、用户界面或其他使用方式上的变更。 - Note that it means *any* user-facing change including all aspects such as API, interface or other behavior changes. ## 功能验证 (How was this patch tested?) 请确认CI已通过增量及存量的单元测试用例。 如果本次测试方式与常规单元测试不同,请详细说明您的测试步骤(最好提供完整的可复现的操作路径及关键截图),以便Committer能够快速复现验证,也便于后续的维护。 如果未添加测试,请说明未添加的原因,以及为何难添加测试。 - [_] 功能自验 - [_] 本地自验截图(涉及个人标识符等敏感信息请注意脱敏) - [_] 新增/变更内容是否已新增/适配UT测试用例看护 CI passed with new added/existing test. If it was tested in a way different from regular unit tests, please clarify how you tested step by step, ideally copy and paste-able, so that other reviewers can test and check, and descendants can verify in the future. If tests were not added, please describe why they were not added and/or why it was difficult to add. - [_] Self-verification of the feature. - [_] Screenshot of local self-verification (please anonymize any sensitive information such as personal identifiers) - [_] Have new or modified unit test (UT) cases been added or adapted to cover the newly added or changed content? See merge request: Ascend/msmodelslim!37430 天前
【feature】【refactor】优化敏感层分析的实现 Co-authored-by: zhangz200102<zhangzheng183@huawei.com> # message auto-generated for no-merge-commit merge: !206 merge refactor/analysis_processor_contents into master 【feature】【refactor】优化敏感层分析的实现 Created-by: zhangz200102 Commit-by: zhangz200102 Merged-by: ascend-robot Description: 本PR对敏感层分析的processor模块进行一定程度的重构,主要调整目录结构,优化相关实现,包含以下内容: 总览: 修改前目录结构 ![修改前目录.PNG](https://raw.gitcode.com/user-images/assets/8444818/3a363e93-7572-4dae-8b93-ab44c73f9224/修改前目录.PNG '修改前目录.PNG') 修改后目录结构 ![修改后目录.PNG](https://raw.gitcode.com/user-images/assets/8444818/98427fa7-4c9f-470f-bdd0-b900a845a9a1/修改后目录.PNG '修改后目录.PNG') - 重命名 AnalysisPipelineLoader -> YamlAnalysisPipelineLoader 基础设施体现第三方依赖,原AnalysisPipelineLoader依赖于yaml模板文件读取 - 敏感层分析结果存储位置变更:context.state -> context.debug 算法间或模块间的需要传递的信息存储到debug中;便于通过上下文机制在debug模式下自动落盘数据 - attention_mse新建独立目录 原attention_mse.py和interface.py变更为attention_mse/impl.py和attention_mse/interface.py - 一元/二元方法基类从method_base.py中分别拆分移动到各自子目录下 method_base中只保留公共逻辑,一元/二元基类(UnaryAnalysisMethod/BinaryAnalysisMethod)从method_base.py中提取分别放到新建的unary_operator/metrics/base.py和binary_operator/metrics/base.py中 - 实现泛型工厂,分别创建一元/二元方法的工厂,各自独立注册、管理。 新建unary_operator/metrics/factory.py和binary_operator/metrics/factory.py - 其他内容:相应调整import和ut文件 See merge request: Ascend/msmodelslim!2062 个月前