| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
Fix Qwen3 FIA shape grid coverage Co-authored-by: Secluded_Ocean<tangchuxiao0709@qq.com> # message auto-generated for no-merge-commit merge: !429 merge codex/qwen3-fia-shape-coverage into master Fix Qwen3 FIA shape grid coverage Created-by: Secluded_Ocean Commit-by: Secluded_Ocean Merged-by: ascend-robot Description: ## Summary - enumerate dense FIA rows by model TP local head variants - add Qwen3 3.5k prefill/decode grid points so TP4 attention shape is generated - add regression coverage for Qwen3 TP4 3593 decode FIA and TP head variants ## Verification - py -3.10 -m pytest tests/regression/cli/test_generate_shape_grid.py tests/regression/cli/test_perf_tooling_ci_map.py -q - parsed 3.5k_10_data.zip, generated shape grid, backfilled FusedInferAttentionScore missing-only on 2-card A3, final Qwen3-32B 3593 decode run hits attention duration 27.760us with no shape miss See merge request: Ascend/msmodeling!429 | 4 天前 | |
补充ut Co-authored-by: tt0cool<xujintao8@h-partners.com> # message auto-generated for no-merge-commit merge: !401 merge master into master 补充ut Created-by: tt0cool Commit-by: tt0cool Merged-by: ascend-robot Description: # PR Template Thanks for your contribution; we appreciate it a lot. The following instructions will make your pull request healthier and help you get feedback more easily. If you do not understand some items, don't worry, just make the pull request and seek help from maintainers. 感谢您的贡献，我们非常重视。以下说明将使您的拉取请求更健康，更易于获得反馈。如果您不理解某些项目，请不要担心，只需提交拉取请求并从维护人员那里寻求帮助即可。 **PR Type / PR类型** - [ ] Feature（功能新增） - [ ] Bugfix（Bug 修复） - [ ] Docs（文档更新） - [ ] CI/CD（持续集成/持续部署） - [ ] Refactor（代码重构） - [ ] Perf（性能优化） - [x] Test-Cases（测试用例更新） - [ ] Other（其他） ## 🔍 Motivation / 变更动机 **Please describe the motivation of this PR and the goal you want to achieve through this PR.** 补充测试用例 ------ ## 📝 Modification / 修改内容 **Please briefly describe what modification is made in this PR.** 修改ut补充测试用例 ------ ## 📐 Associated Test Results / 关联测试结果 **Please provide the related test results, such as test reports, etc.** **请提供相关测试结果，例如测试报告等。**  ------ ## 🌟 Use cases (Optional) / 使用案例（可选） **If this PR introduces a new feature, it is better to list some use cases here and update the documentation.** **如果此拉取请求引入了新功能，最好在此处列出一些用例并更新文档。** ------ ## ✅ Checklist / 检查列表 **Before PR**: - [x] Bug fixes are fully covered by unit tests, the case that causes the bug should be added in the unit tests. / 修复的 Bug 已完全由单元测试覆盖，导致 Bug 的情况应在单元测试中添加。 - [x] The modification is covered by complete unit tests. If not, please add more unit tests to ensure the correctness. / 此拉取请求中的修改已完全由单元测试覆盖。如果不是，请添加更多单元测试以确保正确性。 - [x] All relevant documentation (API docs, docstrings, example tutorials) has been updated to reflect these changes. / 所有相关文档（API 文档、文档字符串、示例教程）已更新以反映这些更改。 - [x] Please ensure code files contain no Chinese comments. / 请保证代码文件中不含中文注释。 ------ See merge request: Ascend/msmodeling!401 | 5 天前 | |
fix: align shape grid model ids and database Co-authored-by: Secluded_Ocean<tangchuxiao0709@qq.com> # message auto-generated for no-merge-commit merge: !348 merge codex/fix-shape-grid-profile-db into master fix: align shape grid model ids and database Created-by: Secluded_Ocean Commit-by: Secluded_Ocean Merged-by: ascend-robot Description: ## Summary - align generate_shape_grid.py --target-models with text_generate model_id naming and reject legacy short names such as dsv3 - keep --rows effective when sampling is capped but rng/seed is None - replace vllm0.18.0_torch2.9.0_cann8.5 with the shape_generated database after validating it has more effective data ## Validation - Database comparison before replacement: old=68 CSV / 823 valid shape rows / 823 positive metric rows; shape_generated=104 CSV / 36198 valid shape rows / 15008 positive metric rows - Final database path: 104 CSV / 36198 shape rows / 15008 positive metric rows; shape_generated path removed - python -m py_compile tools/perf_data_collection/generate_shape_grid.py tools/perf_data_collection/grid_generator/model_configs.py tools/perf_data_collection/grid_generator/theory_router.py tools/perf_data_collection/grid_generator/generators/fused_attention.py - pytest tests/regression/cli/test_shape_grid_model_configs.py tests/regression/cli/test_model_configs.py tests/regression/cli/test_runner.py tests/regression/cli/test_theory_router_pure.py tests/regression/cli/test_generate_shape_grid.py -q See merge request: Ascend/msmodeling!348 | 4 天前 | |
fix(security): add model source safety checks Co-authored-by: jia_ya_nan<jiayanan3@h-partners.com> # message auto-generated for no-merge-commit merge: !385 merge fix/trust-remote-code-safety into master fix(security): add model source safety checks Created-by: jia_ya_nan Commit-by: jia_ya_nan Merged-by: ascend-robot Description: # PR Template Thanks for your contribution; we appreciate it a lot. The following instructions will make your pull request healthier and help you get feedback more easily. If you do not understand some items, don't worry, just make the pull request and seek help from maintainers. 感谢您的贡献，我们非常重视。以下说明将使您的拉取请求更健康，更易于获得反馈。如果您不理解某些项目，请不要担心，只需提交拉取请求并从维护人员那里寻求帮助即可。 **PR Type / PR类型** - [ ] Feature（功能新增） - [x] Bugfix（Bug 修复） - [ ] Docs（文档更新） - [ ] CI/CD（持续集成/持续部署） - [ ] Refactor（代码重构） - [ ] Perf（性能优化） - [ ] Test-Cases（测试用例更新） - [x] Other（其他） ## 🔍 Motivation / 变更动机 **Please describe the motivation of this PR and the goal you want to achieve through this PR.** **请描述您的拉取请求的动机和您希望通过此拉取请求实现的目标。** 安全加固 ------ ## 📝 Modification / 修改内容 **Please briefly describe what modification is made in this PR.** **请简要描述此拉取请求中进行的修改。** 增加本地路径权限校验；增加日志风险提示 去掉不维护的老接口 ------ ## 📐 Associated Test Results / 关联测试结果 **Please provide the related test results, such as test reports, etc.** **请提供相关测试结果，例如测试报告等。**  ------ ## 🌟 Use cases (Optional) / 使用案例（可选） **If this PR introduces a new feature, it is better to list some use cases here and update the documentation.** **如果此拉取请求引入了新功能，最好在此处列出一些用例并更新文档。** ------ ## ✅ Checklist / 检查列表 **Before PR**: - [x] Bug fixes are fully covered by unit tests, the case that causes the bug should be added in the unit tests. / 修复的 Bug 已完全由单元测试覆盖，导致 Bug 的情况应在单元测试中添加。 - [x] The modification is covered by complete unit tests. If not, please add more unit tests to ensure the correctness. / 此拉取请求中的修改已完全由单元测试覆盖。如果不是，请添加更多单元测试以确保正确性。 - [x] All relevant documentation (API docs, docstrings, example tutorials) has been updated to reflect these changes. / 所有相关文档（API 文档、文档字符串、示例教程）已更新以反映这些更改。 - [x] Please ensure code files contain no Chinese comments. / 请保证代码文件中不含中文注释。 ------ See merge request: Ascend/msmodeling!385 | 4 天前 | |
fix(test): align swiglu fusion test with PR !362 selected_token_indices semantics Co-authored-by: Horacehxw<horacehxw@gmail.com> # message auto-generated for no-merge-commit merge: !459 merge bug-fix-mtp into master fix(test): align swiglu fusion test with PR !362 selected_token_indices semantics Created-by: Horacehxw Commit-by: Horacehxw Merged-by: ascend-robot Description: **PR Type / PR类型** - [x] Bugfix（Bug 修复） - [x] Test-Cases（测试用例更新） ## 🔍 Motivation / 变更动机 PR !362 (commit 31d9de33) 合入 master 后，以下两个 nightly 回归测试失败： FAILED test_swiglu_fused_op_present_deepseek_0_deepseek_ai_DeepSeek_V3_1 FAILED test_swiglu_fused_op_present_deepseek_1_deepseek_ai_DeepSeek_V3_1 AssertionError: torch.Size([1, 100, 129280]) != (1, 1, 129280) ### 根因分析 PR !362 将 CausalLmWrapper.forward() 中 lm_head 前的 hidden states 行选择逻辑从： python # 旧代码 if sampling_metadata and sampling_metadata.selected_token_indices is not None: hidden_states = hidden_states.index_select(1, sampling_metadata.selected_token_indices) 替换为统一的 select_lm_head_hidden_states() 函数，内部通过 _has_explicit_selected_token_indices() 判断： python def _has_explicit_selected_token_indices(indices): return indices is not None and indices.ndim > 0 SamplingMetadata.selected_token_indices 的默认值是 torch.tensor(-1)（标量，ndim==0）。 | | 旧行为 | 新行为 | |---|---|---| | tensor(-1) 判定 | is not None → True → 执行 index_select | ndim == 0 → False → 不做选择 | | lm_head 输入 | 选最后 1 个 token → (1, 1, hidden) | 全部 100 token → (1, 100, hidden) | PR !362 对默认值的语义修改**本身是合理的**（哨兵值与有效值分离、默认"不选择"是 fail-safe 行为、生产代码 generate_inputs 从未依赖旧的隐式行为）。但遗漏了更新 test_swiglu_fused_op_present_deepseek 测试——该测试手动构造 SamplingMetadata 时依赖了旧的 tensor(-1) 隐式选择行为。 此外，旧测试的 (1, 1, vocab_size) 期望本身是**偶然正确**的：标量 -1 通过 index_select 只选了全局最后一个 token，而对于 2 序列 packed batch（query_start_loc=[0, 55, 100]），正确的 prefill 行为应选每序列最后一个 token，产出 2 行。 ------ ## 📝 Modification / 修改内容 修改 test_swiglu_fused_op_present_deepseek 测试： 1. **显式传入 selected_token_indices**：使用 attn_meta.query_start_loc[1:] - 1 选取每个序列的最后一个 token（对齐 generate_inputs_varlen 的 prefill 路径逻辑） 2. **更新期望形状**：从硬编码 (1, 1, vocab_size) 改为动态计算 (1, num_sequences, vocab_size) 修复后的测试比旧测试**语义更正确**：旧测试偶然选了 1 个全局最后 token，新测试显式选每序列最后 token。 ------ ## 📐 Associated Test Results / 关联测试结果 pytest: 10 passed, 2 warnings in 60.32s 覆盖 test_swiglu_fusion_pass.py 全部 10 个测试用例（含原先失败的 2 个 nightly 用例）。 ------ ## ✅ Checklist / 检查列表 - [x] Linting tools used / 使用 lintrunner 工具 - [x] Bug fixes covered by unit tests / 修复的 Bug 已由单元测试覆盖 - [x] Modification covered by unit tests / 修改已由单元测试覆盖 - [ ] Documentation updated / 文档已更新 - [x] No Chinese comments in code files / 代码文件中不含中文注释 See merge request: Ascend/msmodeling!459 | 1 小时前 | |
feat(diffusers): support remote repo config autoload - add shared config-only Hugging Face and ModelScope snapshot helpers - resolve remote Diffusers repo ids and explicit snapshot subfolders before loading configs - expose video remote-source in CLI and Web UI - update RFC and offline regression coverage Signed-off-by: minghang_c <chiminghang@h-partners.com> Co-authored-by: minghang_c<chiminghang@h-partners.com> # message auto-generated for no-merge-commit merge: !356 merge diffusers-hf-autoload-master-impl into master feat(diffusers): support remote repo config autoload Created-by: minghang_c Commit-by: minghang_c Merged-by: ascend-robot Description: Thanks for your contribution; we appreciate it a lot. The following instructions will make your pull request healthier and help you get feedback more easily. If you do not understand some items, don't worry, just make the pull request and seek help from maintainers. 感谢您的贡献，我们非常重视。以下说明将使您的拉取请求更健康，更易于获得反馈。如果您不理解某些项目，请不要担心，只需提交拉取请求并从维护人员那里寻求帮助即可。 **PR Type / PR类型** - [x] Feature（功能新增） - [ ] Bugfix（Bug 修复） - [x] Docs（文档更新） - [ ] CI/CD（持续集成/持续部署） - [ ] Refactor（代码重构） - [ ] Perf（性能优化） - [x] Test-Cases（测试用例更新） - [ ] Other（其他） ## 🔍 Motivation / 变更动机 video_generate 原先只支持本地 Diffusers 模型目录，使用公开 Hugging Face / ModelScope Diffusers repo 时需要用户手动下载并整理配置目录。 本 PR 增加 Diffusers 远端 repo id 自动加载能力，使 video_generate 可以直接传入远端 repo id，并且只下载 config 文件，不下载权重文件。同时补充 ModelScope remote source、aggregate repo 子目录寻址和 Web UI 支持，使体验与 text_generate 的 remote source 设计保持一致。 ------ ## 📝 Modification / 修改内容 - 新增共享 Hub helper： - Hugging Face config-only snapshot 下载 - ModelScope config-only snapshot 下载 - ModelScope 参数兼容 allow_patterns / allow_file_pattern - snapshot 下载期间隐藏进度输出和噪音日志 - 新增 Diffusers model resolver： - 本地目录保持原行为，不访问网络 - 非本地输入按 remote_source 解析 Hugging Face / ModelScope repo id - 支持 <namespace>/<repo>/<subfolder> 格式，例如：tencent/HunyuanVideo-1.5/transformer/720p_i2v_distilled_sparse - Diffusers builder 接入 resolver，并将解析后的本地路径交给现有 load_config_from_file - video_generate 两个 CLI 入口新增： - --remote-source {huggingface,modelscope} - 更新 model_id help 文案 - Web UI video_generate 表单新增 remote-source 下拉框，并纳入 task params/hash - 更新 RFC：覆盖 Hugging Face、ModelScope、子目录寻址、日志行为与测试策略 - 新增/更新离线回归测试，覆盖 Hub helper、resolver、builder、CLI help、Web UI command builder/callback/frontend workflow ------ ## 📐 Associated Test Results / 关联测试结果 Focused regression: bash .venv/bin/python -m pytest \ tests/regression/tensor_cast/test_model_hub.py \ tests/regression/tensor_cast/test_diffusers_model_resolver.py \ tests/regression/tensor_cast/test_diffusers_remote_builder.py \ tests/regression/tensor_cast/test_auto_model_config_loader.py::test_modelscope_snapshot_config_only_uses_allowlist \ tests/regression/cli/test_video_generate_remote_source.py \ tests/regression/web_ui/test_command_builder.py \ tests/regression/web_ui/test_callbacks.py \ tests/regression/web_ui/test_frontend_workflows.py -q Result: text 598 passed, 67 warnings Whitespace check: bash git diff --check Result: no output. ------ ## 🌟 Use cases (Optional) / 使用案例（可选） Hugging Face 默认来源： bash python -m cli.inference.video_generate Wan-AI/Wan2.2-T2V-A14B-Diffusers \ --device TEST_DEVICE \ --batch-size 1 \ --seq-len 128 \ --frame-num 81 \ --sample-step 1 ModelScope 来源： bash python -m cli.inference.video_generate Wan-AI/Wan2.2-T2V-A14B-Diffusers \ --remote-source modelscope \ --device TEST_DEVICE \ --batch-size 1 \ --seq-len 128 \ --frame-num 81 \ --sample-step 1 Aggregate repo 子目录： bash python -m cli.inference.video_generate tencent/HunyuanVideo-1.5/transformer/720p_i2v_distilled_sparse \ --device TEST_DEVICE \ --batch-size 1 \ --seq-len 128 \ --frame-num 121 \ --sample-step 1 ------ ## ✅ Checklist / 检查列表 **Before PR**: - [ ] Bug fixes are fully covered by unit tests, the case that causes the bug should be added in the unit tests. / 修复的 Bug 已完全由单元测试覆盖，导致 Bug 的情况应在单元测试中添加。 - [x] The modification is covered by complete unit tests. If not, please add more unit tests to ensure the correctness. / 此拉取请求中的修改已完全由单元测试覆盖。如果不是，请添加更多单元测试以确保正确性。 - [x] All relevant documentation (API docs, docstrings, example tutorials) has been updated to reflect these changes. / 所有相关文档（API 文档、文档字符串、示例教程）已更新以反映这些更改。 - [x] Please ensure code files contain no Chinese comments. / 请保证代码文件中不含中文注释。 ------ 🤖 Generated with [Claude Code](https://claude.com/claude-code) See merge request: Ascend/msmodeling!356 | 12 天前 | |
【同步】【非开发代码】代码从 develop 同步到 master Co-authored-by: yydyzr<liuyuncong1@huawei.com> Co-authored-by: gcw_61YBRfIt<chuzhenxing@huawei.com> Co-authored-by: 孔炳翔<1120200577@qq.com> Co-authored-by: zhengxinqian<qianzhengxin@huawei.com> Co-authored-by: hw_whx<wanghexiang7@huawei.com> Co-authored-by: jgong5<steven.gong@gmail.com> Co-authored-by: hw_whx<2952154980@qq.com> # message auto-generated for no-merge-commit merge: !330 merge master into master 【同步】【非开发代码】代码从 develop 同步到 master Created-by: AvadaKedavrua Commit-by: liujiawang;ascend-robot;AvadaKedavrua;lutean;Horacehxw;eveyin1;minghang_c;zwt__;tt0cool;elrond-g;jia_ya_nan;zhenyu_zhang;ChenHuiwen;wangshen001;Hudingyi;wendellX;Secluded_Ocean;jhon-117;yaohan404;jiangruitao;zhenghaojie;stormchasingg;panyj1993;cmh1056291129;yuyinkai1;sunguozhong;genius52;liu_jiaxu;HongMaoShuiGuai;zhengxinqian;weixin_43368449;jsez-li-bin;jgong5;wqh17101;w00609794;yydyzr;JieZhang679;sppedforcy;gcw_61YBRfIt;Jiong Gong;hw_whx;gongjiong;孔炳翔 Merged-by: ascend-robot Description: 代码从 develop 同步到 master，后续基于 master 演进，并支持打包 See merge request: Ascend/msmodeling!330 | 16 天前 |
| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
| 4 天前 | ||
| 5 天前 | ||
| 4 天前 | ||
| 4 天前 | ||
| 1 小时前 | ||
| 12 天前 | ||
| 16 天前 |