0
代码
代码
Issues
Pull Requests
讨论
Wiki
分析
0
ascend-robotascend-robot[Feature] qwen3.5 support new dcp convert and fused rmsnorm
2a44e12f创建于 3月7日历史提交
文件最后提交记录最后更新时间
__init__.py
!1121 [Refactor] understanding model weight convert code restructure Merge pull request !1121 from pjgao/master 11 个月前
deepseekvl2.py
!1445 [Bugfix]Modify the file permission control in the weight conversion scenario Merge pull request !1445 from zhangxubin/master 9 个月前
glm.py
!1445 [Bugfix]Modify the file permission control in the weight conversion scenario Merge pull request !1445 from zhangxubin/master 9 个月前
internvl.py
!1445 [Bugfix]Modify the file permission control in the weight conversion scenario Merge pull request !1445 from zhangxubin/master 9 个月前
mistral3.py
[Feature] Add Magistral3 lora readme and script Co-authored-by: wanghao<wanghao773@h-partners.com> # message auto-generated for no-merge-commit merge: !2075 merge lora_sh_mistral into master [Feature] Add Magistral3 lora readme and script Created-by: vectorwhh Commit-by: wanghao Merged-by: ascend-robot Description: [Feature] Add Magistral3 lora readme and script 此PR依赖于前置PR：https://gitcode.com/Ascend/MindSpeed-MM/pull/2064和https://gitcode.com/Ascend/MindSpeed-MM/pull/2061 检视时可只看fsdp2_lora_config.yaml、finetune_magistral_2509_lora.sh两个文件以及readme的lora微调部分 See merge request: Ascend/MindSpeed-MM!20754 个月前
moe_expert.py
[Modify]Performance Optimization for Qwen3-Omni Thinker MoE Expert Weight Conversion Co-authored-by: yaoyaoxu<xuyaoyao.824404@huawei.com> # message auto-generated for no-merge-commit merge: !1857 merge better_perf_qwen3omni into master [Modify]Performance Optimization for Qwen3-Omni Thinker MoE Expert Weight Conversion Created-by: yaoyaoxu Commit-by: yaoyaoxu Merged-by: ascend-robot Description: ## Motivation Performance Optimization for Qwen3-Omni Thinker MoE Expert Weight Conversion ## Modification Performance Optimization for Qwen3-Omni Thinker MoE Expert Weight Conversion ## Self-test (Optional) If modifications to this PR may cause/fix function/accuracy/performance DTSs/issues, a self-inspection record needs to be attached. ## BC-breaking (Optional) If there are compatibility issues, such as dependencies on cann/torch_npu versions, they need to be explained in the PR. ## Checklist **Before PR**: - [ ] The new code needs to comply with the Clean Code specification. - [ ] The PR content is self-checked, and the expression can be clear and the writing standardized **After PR**: - [ ] CLA has been signed and all committers have signed the CLA in this PR. - [ ] The ci-pipeline is passed, Code Check is passed. See merge request: Ascend/MindSpeed-MM!18575 个月前
qwen2_5omni.py
[feature] hetero pp ckpt converter for vl and omni model Co-authored-by: xiaoyue994<xiaoyuanhang@huawei.com> # message auto-generated for no-merge-commit merge: !2214 merge mm_ckpt_convert into master [feature] hetero pp ckpt converter for vl and omni model Created-by: xiaoyue994 Commit-by: xiaoyue994 Merged-by: ascend-robot Description: ## Motivation Add hetero-pp-ckpt-converter for hetero parallel when powering pipeline parallel. ## Modification Modify args check and pp converter func. ## Self-test (Optional) If modifications to this PR may cause/fix function/accuracy/performance DTSs/issues, a self-inspection record needs to be attached. ## BC-breaking (Optional) If there are compatibility issues, such as dependencies on cann/torch_npu versions, they need to be explained in the PR. ## Checklist **Before PR**: - [ ] The new code needs to comply with the Clean Code specification. - [ ] The PR content is self-checked, and the expression can be clear and the writing standardized **After PR**: - [ ] CLA has been signed and all committers have signed the CLA in this PR. - [ ] The ci-pipeline is passed, Code Check is passed. See merge request: Ascend/MindSpeed-MM!22143 个月前
qwen2_5vl.py
[Feature] Qwen2.5vl compatible with LoRA Co-authored-by: LKONE<wanglikai4@huawei.com> # message auto-generated for no-merge-commit merge: !1683 merge master into master [Feature] Qwen2.5vl compatible with LoRA Created-by: LKONE Commit-by: LKONE Merged-by: ascend-robot Description: ## Motivation Qwen2.5VL adapted for LoRA weight conversion. ## Modification Added weight conversion code for LoRA between mm model and huggingface model. ## Self-test (Optional) If modifications to this PR may cause/fix function/accuracy/performance DTSs/issues, a self-inspection record needs to be attached. ## BC-breaking (Optional) If there are compatibility issues, such as dependencies on cann/torch_npu versions, they need to be explained in the PR. ## Checklist **Before PR**: - [x] The new code needs to comply with the Clean Code specification. - [x] The PR content is self-checked, and the expression can be clear and the writing standardized **After PR**: - [x] CLA has been signed and all committers have signed the CLA in this PR. - [x] The ci-pipeline is passed, Code Check is passed. See merge request: Ascend/MindSpeed-MM!16836 个月前
qwen2vl.py
[Bugfix] Fix weight conversion bug Co-authored-by: ningmengliu<liuhao438@huawei.com> # message auto-generated for no-merge-commit merge: !2025 merge master into master [Bugfix] Fix weight conversion bug Created-by: ningmenglh Commit-by: ningmengliu Merged-by: ascend-robot Description: ## Motivation Fix the inconsistency between mm-hf weight conversion and hf-mm weight conversion. ## Modification Fix the inconsistency between mm-hf weight conversion and hf-mm weight conversion. ## Self-test (Optional) If modifications to this PR may cause/fix function/accuracy/performance DTSs/issues, a self-inspection record needs to be attached. ## BC-breaking (Optional) If there are compatibility issues, such as dependencies on cann/torch_npu versions, they need to be explained in the PR. ## Checklist **Before PR**: - [ ] The new code needs to comply with the Clean Code specification. - [ ] The PR content is self-checked, and the expression can be clear and the writing standardized **After PR**: - [ ] CLA has been signed and all committers have signed the CLA in this PR. - [ ] The ci-pipeline is passed, Code Check is passed. See merge request: Ascend/MindSpeed-MM!20254 个月前
qwen3_5.py
[Feature] qwen3.5 support new dcp convert and fused rmsnorm Co-authored-by: young256<liumingyang16@huawei.com> # message auto-generated for no-merge-commit merge: !2246 merge fsdp2_dev_qwen35_publish into master [Feature] qwen3.5 support new dcp convert and fused rmsnorm Created-by: young256 Commit-by: young256 Merged-by: ascend-robot Description: ## What this PR does / why we need it? Qwen3.5 needs to support new DCP weight conversion to reduce the vector operator overhead caused by weight transpose in the program. The mul vector operator caused by RMSNorm and RMSNormGate needs to be eliminated. ## Does this PR introduce any user-facing change? DCP weight conversion instructions have changed. ## How was this patch tested? The weight verification after weight conversion is valid. The fused operator replacement has completed precision alignment. See merge request: Ascend/MindSpeed-MM!22462 个月前
qwen3vl.py
[feature] Support qwen3vl lora training based on fsdp2 Co-authored-by: suis-moi<zhushixiang@huawei.com> Co-authored-by: zhushixiang<1318709014@qq.com> # message auto-generated for no-merge-commit merge: !2061 merge master into master [feature] Support qwen3vl lora training based on fsdp2 Created-by: suis-moi Commit-by: suis-moi;zhushixiang Merged-by: ascend-robot Description: Support qwen3vl lora training based on fsdp2 and torch dcp format a. supoport qwen3vl lora training and lora weights convert, must enable fsdp2 and using torch_dcp fomat. b. support MoE module lora training and reuires Pert>=18.0.1 See merge request: Ascend/MindSpeed-MM!20614 个月前
qwen3vl_megatron.py
[mindspore] ckpt convert for qwen3vl between hf and megatron format Co-authored-by: chengpeng25<chengpeng9@huawei.com> # message auto-generated for no-merge-commit merge: !1765 merge dev_convert_ckpt into master [mindspore] ckpt convert for qwen3vl between hf and megatron format Created-by: chengpeng25 Commit-by: chengpeng25 Merged-by: ascend-robot Description: ## Motivation Ckpt converter between huggingface and megatron format for qwen3vl. ## Modification qwen3vl ckpt converter between hf and megatron, including Dense and moe models. ## Self-test (Optional) If modifications to this PR may cause/fix function/accuracy/performance DTSs/issues, a self-inspection record needs to be attached. ## BC-breaking (Optional) If there are compatibility issues, such as dependencies on cann/torch_npu versions, they need to be explained in the PR. ## Checklist **Before PR**: - [x] The new code needs to comply with the Clean Code specification. - [x] The PR content is self-checked, and the expression can be clear and the writing standardized **After PR**: - [x] CLA has been signed and all committers have signed the CLA in this PR. - [x] The ci-pipeline is passed, Code Check is passed. See merge request: Ascend/MindSpeed-MM!17656 个月前
videoalign.py
[Bugfix]Fix some encoding issues Co-authored-by: zhangxubin<1656631289@qq.com> # message auto-generated for no-merge-commit merge: !1715 merge master into master [Bugfix]Fix some encoding issues Created-by: MoCuishle-M Commit-by: zhangxubin Merged-by: ascend-robot Description: ## Motivation 根据扫描结果修复一些编码问题. ## Modification 存在的问题： 1. 函数重复定义 2. 函数使用错误 3. 变量名拼写错误 4. 使用未定义变量 ## Self-test (Optional) If modifications to this PR may cause/fix function/accuracy/performance DTSs/issues, a self-inspection record needs to be attached. ## BC-breaking (Optional) If there are compatibility issues, such as dependencies on cann/torch_npu versions, they need to be explained in the PR. ## Checklist **Before PR**: - [ ] The new code needs to comply with the Clean Code specification. - [ ] The PR content is self-checked, and the expression can be clear and the writing standardized **After PR**: - [ ] CLA has been signed and all committers have signed the CLA in this PR. - [ ] The ci-pipeline is passed, Code Check is passed. See merge request: Ascend/MindSpeed-MM!17156 个月前