| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
refactor: make post_norm config-driven, fix fc_type ref, clean qkv_split Co-authored-by: wanggangguo<wanggangguo@huawei.com> # message auto-generated for no-merge-commit merge: !4738 merge refactor into master refactor: make post_norm config-driven, fix fc_type ref, clean qkv_split Created-by: isfrapples Commit-by: wanggangguo Merged-by: ascend-robot Description: ## What this PR does / why we need it? 1. post_norm 由运行时探测改为配置驱动，model_cfg.json 显式声明 2. fc_type 修复为从 load_model 读取，up_down 分支无 gate 3. 移除 qkv_split 冗余配置，ng>1 时始终 per-group 拆分 4. 旧 checkpoint 缺少 post_attn_norm.weight 时容错不崩溃 5. 移除 extra_norm_keys 遍历和 hasattr 死代码 ## Does this PR introduce any user-facing change? 1. 新增 gemma2 配置项 post_norm: true，需同步 model_cfg.json 2. 移除 qwen2-moe、qwen25、gemma2、seed-oss 的 qkv_split 配置 3. 无 API 或 CLI 接口变更，向下兼容旧 checkpoint ## How was this patch tested? 1. 13 个测试用例全部通过，覆盖 6 个模型和 2 个转换方向 2. 每个用例包含完整权重精确对比，gate_proj 参与校验 3. 旧 checkpoint 兼容性验证：gemma2 缺少 post_attn_norm 不崩溃 4. 存量 bug 修复验证：llama2 k_proj 形状错误已修复 fixes#1604 See merge request: Ascend/MindSpeed-LLM!4738 | 2 个月前 | |
update pipeline testcase Co-authored-by: guihaowen666<guihaowen@huawei.com> # message auto-generated for no-merge-commit merge: !4278 merge br_master_pipeline_update into master update pipeline testcase Created-by: guihaowen666 Commit-by: guihaowen666 Merged-by: ascend-robot Description: ## What this PR does / why we need it? Please describe the background and detailed changes of the PR. If it is a bugfix, please attach the related issue. ## 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. ## How was this patch tested? Please explain how to verify the correctness and effectiveness of this feature, as well as its usage constraints and limitations. See merge request: Ascend/MindSpeed-LLM!4278 | 7 个月前 | |
test: add unit tests for parallel weight conversion in WeightLoader Co-authored-by: heyuzhen<heyuzhen5@huawei.com> # message auto-generated for no-merge-commit merge: !4862 merge master into master test: add unit tests for parallel weight conversion in WeightLoader Created-by: heyuzhen Commit-by: heyuzhen Merged-by: ascend-robot Description: ## What this PR does / why we need it? 为 WeightLoader._load_pretrained 中新增的并行权重转换逻辑添加单元测试。 原代码使用顺序循环执行权重转换，修改后改为 ThreadPoolExecutor 并行调度（min(tasks, 4) worker），以提升多 MoE 专家权重转换的加载速度。本 PR 仅针对新增的并行调度代码编写 UT，不涉及已有逻辑的测试。 测试覆盖： - 单任务走 else 顺序分支 - 多任务并行执行，结果完整收集 - worker 数上限为 4 - torch.set_num_threads(1) 在正常/异常情况下均能恢复 - 并行结果与顺序结果完全一致 - 空 tasks 不崩溃 - 并行时各 converter 收到正确的 full_name，无串扰 ## Does this PR introduce any user-facing change? 否。仅新增测试用例，不修改任何业务代码。 ## How was this patch tested? python tests/pipeline/ut/convert_parallel/test_parallel_conversion.py -v See merge request: Ascend/MindSpeed-LLM!4862 | 2 个月前 | |
refactor(megatron):update coverage script Co-authored-by: guihaowen666<guihaowen@huawei.com> # message auto-generated for no-merge-commit merge: !4295 merge br_master_coverage_fix_0313 into master refactor(megatron):update coverage script Created-by: guihaowen666 Commit-by: guihaowen666 Merged-by: ascend-robot Description: # 覆盖率分析脚本更新 ---- ## What this PR does / why we need it? 更新仓库run_coverage.sh脚本，修复覆盖率分析扫描文件不全的问题 ## Does this PR introduce any user-facing change? 不影响仓库基本功能，旨在优化仓库覆盖率分析功能 ## How was this patch tested? 已在蓝区机器上自测通过 See merge request: Ascend/MindSpeed-LLM!4295 | 6 个月前 | |
refactor(megatron):update coverage script Co-authored-by: guihaowen666<guihaowen@huawei.com> # message auto-generated for no-merge-commit merge: !4295 merge br_master_coverage_fix_0313 into master refactor(megatron):update coverage script Created-by: guihaowen666 Commit-by: guihaowen666 Merged-by: ascend-robot Description: # 覆盖率分析脚本更新 ---- ## What this PR does / why we need it? 更新仓库run_coverage.sh脚本，修复覆盖率分析扫描文件不全的问题 ## Does this PR introduce any user-facing change? 不影响仓库基本功能，旨在优化仓库覆盖率分析功能 ## How was this patch tested? 已在蓝区机器上自测通过 See merge request: Ascend/MindSpeed-LLM!4295 | 6 个月前 | |
refactor(megatron):update coverage script Co-authored-by: guihaowen666<guihaowen@huawei.com> # message auto-generated for no-merge-commit merge: !4295 merge br_master_coverage_fix_0313 into master refactor(megatron):update coverage script Created-by: guihaowen666 Commit-by: guihaowen666 Merged-by: ascend-robot Description: # 覆盖率分析脚本更新 ---- ## What this PR does / why we need it? 更新仓库run_coverage.sh脚本，修复覆盖率分析扫描文件不全的问题 ## Does this PR introduce any user-facing change? 不影响仓库基本功能，旨在优化仓库覆盖率分析功能 ## How was this patch tested? 已在蓝区机器上自测通过 See merge request: Ascend/MindSpeed-LLM!4295 | 6 个月前 | |
feat: adapt MLA recompute and swap for TP1 Co-authored-by: Jia_Austin<dengjia6@huawei.com> # message auto-generated for no-merge-commit merge: !4651 merge adapt-tp1-mla-recompute-swap into master feat: adapt MLA recompute and swap for TP1 Created-by: Jia_Austin Commit-by: Jia_Austin Merged-by: ascend-robot Description: [#1649](https://gitcode.com/Ascend/MindSpeed-LLM/issues/1649) ## What this PR does / why we need it? Added support for MLA recalculation and swapping for TP1—which was previously unsupported—and reduced maintenance costs through code refactoring. ## Does this PR introduce any user-facing change? NA ## How was this patch tested? test by ci See merge request: Ascend/MindSpeed-LLM!4651 | 2 个月前 | |
test: pipeline ut testcase fix Co-authored-by: yanzhenghao<yanzhenghao2@huawei.com> Co-authored-by: f00620112<fangminghao@huawei.com> # message auto-generated for no-merge-commit merge: !4592 merge master into master test: pipeline ut testcase fix Created-by: yanzhenghao Commit-by: yanzhenghao;f00620112 Merged-by: ascend-robot Description: fix the error of pipeline ut testcase See merge request: Ascend/MindSpeed-LLM!4592 | 3 个月前 | |
test(megatron):rl testcase remove Co-authored-by: guihaowen666<guihaowen@huawei.com> # message auto-generated for no-merge-commit merge: !4394 merge br_master_pipeline_script_update into master test(megatron):rl testcase remove Created-by: guihaowen666 Commit-by: guihaowen666 Merged-by: ascend-robot Description: ## What this PR does / why we need it? Please describe the background and detailed changes of the PR. If it is a bugfix, please attach the related issue. ## 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. ## How was this patch tested? Please explain how to verify the correctness and effectiveness of this feature, as well as its usage constraints and limitations. See merge request: Ascend/MindSpeed-LLM!4394 | 6 个月前 | |
update pipeline testcase Co-authored-by: guihaowen666<guihaowen@huawei.com> # message auto-generated for no-merge-commit merge: !4278 merge br_master_pipeline_update into master update pipeline testcase Created-by: guihaowen666 Commit-by: guihaowen666 Merged-by: ascend-robot Description: ## What this PR does / why we need it? Please describe the background and detailed changes of the PR. If it is a bugfix, please attach the related issue. ## 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. ## How was this patch tested? Please explain how to verify the correctness and effectiveness of this feature, as well as its usage constraints and limitations. See merge request: Ascend/MindSpeed-LLM!4278 | 7 个月前 |
| 文件 | 最后提交记录 | 最后更新时间 |
|---|---|---|
| 2 个月前 | ||
| 7 个月前 | ||
| 2 个月前 | ||
| 6 个月前 | ||
| 6 个月前 | ||
| 6 个月前 | ||
| 2 个月前 | ||
| 3 个月前 | ||
| 6 个月前 | ||
| 7 个月前 |