Ecosyste.ms: Issues

An open API service for providing issue and pull request metadata for open source projects.

GitHub / jingyaogong/minimind-v issues and pull requests

#34 - 关于loss不下降的问题

Issue - State: open - Opened by LiuCharming 4 days ago - 1 comment

#32 - loss

Issue - State: closed - Opened by hxqiu 10 days ago - 1 comment

#32 - loss

Issue - State: closed - Opened by hxqiu 10 days ago - 1 comment

#31 - 关于DDP的问题

Issue - State: open - Opened by hlh2023214 about 1 month ago - 1 comment

#30 - 关于mapping层的设计

Issue - State: open - Opened by ZKCCZ about 1 month ago - 2 comments

#29 - 请问多图微调数据集中提到的output.json文件应该去哪里下载呢？

Issue - State: closed - Opened by why135724 about 2 months ago - 3 comments

#28 - 语言模型能替换成其他模型吗

Issue - State: closed - Opened by zhaowanjin about 2 months ago - 1 comment

#27 - 微调的时候报错了是什么问题呢

Issue - State: closed - Opened by LiuCharming about 2 months ago - 5 comments

#26 - 微调的时候报错了是什么问题呢

Issue - State: closed - Opened by LiuCharming about 2 months ago

#25 - CLIP对应的词表与语言模型是相同的吗

Issue - State: closed - Opened by lightning0016 2 months ago - 1 comment

#24 - AMP训练-数据类型选择问题

Issue - State: closed - Opened by chongkuiqi 2 months ago - 1 comment

#23 - 模型结构中dropout层位置的问题

Issue - State: closed - Opened by chongkuiqi 2 months ago - 1 comment

#22 - More modality

Issue - State: open - Opened by Yu-xm 2 months ago - 1 comment

#20 - 讨论、对之前代码优化以及添加Siglip编码器

Pull Request - State: closed - Opened by xinyanghuang7 3 months ago - 5 comments

#19 - 添加多图训练和推理过程

Pull Request - State: closed - Opened by xinyanghuang7 3 months ago - 3 comments

#18 - 显存需求

Issue - State: open - Opened by AHPUymhd 3 months ago - 1 comment

#17 - no issue

Issue - State: closed - Opened by why135724 3 months ago

#16 - sft 阶段loss异常增高

Issue - State: closed - Opened by WangRongsheng 3 months ago - 3 comments

#15 - 如何训练自己的数据集呢？

Issue - State: open - Opened by cripsgreen 3 months ago - 2 comments

#14 - Fix default max_seq_len from 200 to 512

Pull Request - State: closed - Opened by MAOJIASONG 3 months ago - 1 comment

#13 - sft-vlm 报错

Issue - State: closed - Opened by MAOJIASONG 3 months ago - 1 comment

#10 - 关于在autodl 训练遇到的 lfs 问题

Issue - State: closed - Opened by pengqianhan 3 months ago - 2 comments

#8 - 关于visual encoder部分的小问题

Issue - State: open - Opened by wrsnice 4 months ago - 9 comments

#7 - 大佬之後會有音頻模型(或是附加音頻的三模態)嗎?

Issue - State: open - Opened by win10ogod 4 months ago - 4 comments

#5 - 关于效果预期

Issue - State: open - Opened by jingyaogong 4 months ago - 2 comments

#3 - new_h_i = torch.cat((before, vision_proj[i], after), dim=0)[:seqlen]

Issue - State: open - Opened by pengqianhan 4 months ago - 1 comment

#2 - 用GPU和用CPU,训练损失差别好大

Issue - State: open - Opened by cqcracked 4 months ago - 1 comment