芯动开源-openMind专场活动第2期来袭！11月7日-11月27日，连续3周，每周31个获奖名额，每位获奖者100-1000元不等，快来挑战吧~>>> openMind开源实习正在进行中！高额奖励，难得的实习机会，快报名参加吧！>>> 关于云脑任务中统一路径访问方式的公告>>>
deng
Public Activity Repositories Datasets Starred Repositories
83
 Following
18
 Followers
4
 Achievement Badge
1
Loading Heatmap…

deng synced commits to dev at deng/mindformers from mirror

  • ee4b12ae69 !4962 【dev】【config精简】优化模板逻辑，兼容None输入 Merge pull request !4962 from husichao/dev_fix
  • e06b100d18 template config: don't raise error when input is none
  • Compare 2 commits »

12 hours ago

deng synced commits to r1.3.1 at deng/mindformers from mirror

  • d677e85852 !4946 同步r1.3.1分支漏合的PR引入的权重加载报错 Merge pull request !4946 from wja/wja_r1.3.1
  • 2a38281b9f 修改原因：同步r1.3.0分支漏合的PR引入的权重加载报错 修改人：w30008384
  • Compare 2 commits »

1 day ago

deng synced commits to r1.3.0 at deng/mindformers from mirror

  • d25ee86578 !4937 add en file 1.3.0 Merge pull request !4937 from 宦晓玲/code_docs_124.3
  • 19e67190a6 add en file 1.3.0
  • a272f57064 !4960 【r1.3.0】【Bugfix】修复ulysses与gqa特性组合场景下的精度问题 Merge pull request !4960 from 杨承翰/r1.3.0_cp
  • c14f1c2f27 【r1.3.0】【Bugfix】修复ulysses与gqa特性组合场景下的精度问题
  • Compare 4 commits »

1 day ago

deng synced commits to dev at deng/mindformers from mirror

  • eeb4c5586c !4910 【dev】【feature】：动静统一api适配LoRA低参微调 Merge pull request !4910 from wangjialin/dev_lora
  • 5f40f11ebc !4969 Fix docs error Merge pull request !4969 from Jingwei Huang/docs_error
  • 4b392fc232 !4957 【dev】【Bugfix】修复ulysses与gqa特性组合场景下的精度问题 Merge pull request !4957 from 杨承翰/dev_ulysses_cp
  • cff237b8d5 Fix docs error
  • 235357fa90 【dev】【Bugfix】修复ulysses与gqa特性组合场景下的精度问题
  • Compare 10 commits »

1 day ago

deng synced commits to r1.3.0 at deng/mindformers from mirror

  • 6d66269774 !4914 【R1.3.0】【bugfix】 修复部分使用msrun启动MF_LOG_PATH失效的场景 Merge pull request !4914 from Yule100/bugfix_r1.3.0
  • 635be0e15f !4903 1.3.0 llama下架文档修改 Merge pull request !4903 from qsc/code_docs_llama
  • 038f13e6be bugfix_r1.3.0_log_mf_path
  • 49c9236e18 modify_doc
  • Compare 4 commits »

2 days ago

deng synced commits to dev at deng/mindformers from mirror

  • 48a28a7a65 !4939 【dev】DeepSeek1.5模型推理适配 Merge pull request !4939 from baiyulu/code_docs_DeepSeekCoder
  • 334d37ab58 !4947 【dev】【ci】update ms 20241130 & 下架动态图相关用例 Merge pull request !4947 from niujunhao/update_ms
  • 2d965d4072 !4942 Add is_exists function for Context Merge pull request !4942 from Jingwei Huang/Context_is_exists
  • a3211058f2 !4825 【dev】【Feature】统一API静态图支持cp Merge pull request !4825 from 杨承翰/dev_graph_ulysses
  • 2b33f848e7 !4878 【dev】一招推理上库 Merge pull request !4878 from 吴昊天/code_docs_yizhao_dev_predict
  • Compare 20 commits »

2 days ago

deng synced commits to dev at deng/mindformers from mirror

  • 143f228698 !4772 [dev] [Bugfix] glm推理logits适配mindie服务化推理做reshape对齐 Merge pull request !4772 from 何泽泉/glm-bug-fix
  • 549823fd26 !4021 quantization_config参数新增 Merge pull request !4021 from yyyyrf/dev
  • f00e99b637 add ptq_config new parameters
  • 63524184fd 修改推理logits的维度
  • Compare 4 commits »

3 days ago

deng synced commits to r1.3.0 at deng/mindformers from mirror

  • 2bd4cb76ea !4895 【r1.3.0】【文档修复】修复api文档中example执行失败问题 Merge pull request !4895 from niujunhao/bugfix/fix_examples_1.3.0
  • 45375f99d3 !4893 【1.3.0】【bugfix】修正llm_boost的文档和示例中的错误 Merge pull request !4893 from zhangminli/atb_mf_1.3.0_bugfix
  • 02750663c7 fix llm_boost doc error
  • 64e0a5a927 fix auto class example
  • Compare 4 commits »

3 days ago

deng synced commits to dev at deng/mindformers from mirror

  • 85bd65bf46 !4891 Bugfix of ParallelLlamaForCausalLM in pynative mode Merge pull request !4891 from zhanzhan/bugfix
  • 90f002bab3 !4758 slice to stridedslice in mla scope Merge pull request !4758 from 黄勇/update_deepseekv2
  • 413d9a4011 !4883 [dev] [feature] update pipline auto balance 2 Merge pull request !4883 from czrz/ppb_2
  • afcab2eb71 !4911 Qwen2.5_readme修改 Merge pull request !4911 from gaojie/code_docs_qwen2_5_readme
  • 22cc14cac1 !4869 [dev] [feature] update pipline auto balance 1 Merge pull request !4869 from czrz/ppb_1
  • Compare 12 commits »

3 days ago

deng synced commits to dev at deng/mindformers from mirror

  • 31a92de4ab !4904 增加对SLoRA配置target_modules、adapter_id错误情况的错误拦截 Merge pull request !4904 from guoguopot/IB7500
  • 26597d94fb !4897 Add get_attr and set_attr for MindformerConfig Merge pull request !4897 from Jingwei Huang/mindformconfig
  • ed6011b2db !4870 【dev】【bugfix】修复 MoE + Bfloat16 初始化场景下，精度未对齐的问题 Merge pull request !4870 from YingLai Lin/dev_fix_moe_loss
  • 6820f287de target_modules，adapter_id配置错误拦截
  • 823f0fd71f Add get_attr and set_attr for MindformerConfig
  • Compare 6 commits »

4 days ago

deng synced commits to dev at deng/mindformers from mirror

6 days ago

deng synced commits to dev at deng/mindformers from mirror

1 week ago

deng synced commits to r1.3.0 at deng/mindformers from mirror

  • 8776099eec !4806 【r1.3.0】简化首页README，相关内容链接至官网文档 Merge pull request !4806 from Xinrui Chen/code_docs_r1.3.0_readme
  • 15dc3f403f !4873 【r1.3.0】【docs】qwen 7b配置文件mode拼写错误 Merge pull request !4873 from niujunhao/code_docs_qwen_yaml_r1.3.0
  • dcd2076536 Simplified home page README, contents link to corresponding official documentation
  • da8dfc36a1 fix mode word in yaml
  • Compare 4 commits »

1 week ago

deng synced commits to dev at deng/mindformers from mirror

  • 77269f0477 !4882 【轻量级 PR】：[bugfix][dev][toolkits]修复pp模拟器不同chunk内存计算问题 Merge pull request !4882 from xwkgch/N/A
  • 0e82c3f56f !4865 LLM boost support for llama ascendNative fused model Merge pull request !4865 from ehaleva/dev-export
  • 855a92330b !4856 【dev】【bugfix】修复 calculate_per_token_loss 默认值中英文文档不一致问题 Merge pull request !4856 from YingLai Lin/code_docs_dev_fix_doc
  • f51d35064a !4881 Adapt the case of no parallel in config Merge pull request !4881 from Jingwei Huang/parallel_bugfix
  • 2a5abaa05b !4885 【dev】【config精简】【bugfix】回滚pretrain_llama2_7b.yaml Merge pull request !4885 from husichao/fix_maoyan
  • Compare 18 commits »

1 week ago

deng synced commits to dev at deng/mindformers from mirror

  • 763b622ce8 !4863 Fix error of getting use_past and parallel context Merge pull request !4863 from Jingwei Huang/context_bugfix
  • ab4e033674 Fix error of getting use_past and parallel context
  • Compare 2 commits »

1 week ago

deng synced commits to r1.4.0-beta1 at deng/mindformers from mirror

1 week ago

deng synced commits to r1.3.0 at deng/mindformers from mirror

  • 209f483295 !4847 【r1.3.0】【bugfix】修复 moe + bfloat16 初始化场景下 loss 未对齐问题 Merge pull request !4847 from YingLai Lin/r1.3.0_fix_moe_loss
  • 437cf9f562 "修复 Moe + bfloat16 场景下，loss 未对齐的场景"
  • Compare 2 commits »

1 week ago

deng synced commits to dev at deng/mindformers from mirror

  • a0cf27c2f2 !4860 【bugfix】统一FA mask接口 Merge pull request !4860 from tan-wei-cheng/develop-twc-dev4
  • 048e5f5f6b 统一FA mask接口
  • 45040efb16 !4854 Bugfix of LlamaForCausalLM and ParallelLlamaForCausalLM for pynative mode Merge pull request !4854 from zhanzhan/bugfix
  • 911da6e009 Bugfix of LlamaForCausalLM and ParallelLlamaForCausalLM for pynative mode
  • Compare 4 commits »

1 week ago

deng synced commits to dev at deng/mindformers from mirror

1 week ago

deng synced commits to dev at deng/mindformers from mirror

  • c13f97bc7b !4861 【dev】【bugfix】qwen finetune yaml中mode错误修复 Merge pull request !4861 from niujunhao/code_docs_qwen_yaml
  • c5240a81ee fix qwen finetune yaml
  • 2afa2cd7ac !4858 Fix the error caused by the type of parallel_config being TransformerOpParallelConfig Merge pull request !4858 from Jingwei Huang/context_bugfix
  • 1992d77cfa !4842 【dev】【bugfix】qkv concat校验修改为仅作用于推理任务 Merge pull request !4842 from pengjingyou/bugfix_qkv_check_resume
  • b8133da404 Fix the error caused by the type of parallel_config being TransformerOpParallelConfig
  • Compare 6 commits »

1 week ago

Please read the following content carefully:

Dear OpenI User

Thank you for your continuous support to the Openl Qizhi Community AI Collaboration Platform. In order to protect your usage rights and ensure network security, we updated the Openl Qizhi Community AI Collaboration Platform Usage Agreement in January 2024. The updated agreement specifies that users are prohibited from using intranet penetration tools. After you click "Agree and continue", you can continue to use our services. Thank you for your cooperation and understanding.

For more agreement content, please refer to the《Openl Qizhi Community AI Collaboration Platform Usage Agreement》

Agree and continue
Disagree, exit