[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-80128":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":9,"language":10,"languages":9,"totalLinesOfCode":9,"stars":11,"forks":12,"watchers":13,"openIssues":12,"contributorsCount":14,"subscribersCount":14,"size":14,"stars1d":15,"stars7d":15,"stars30d":16,"stars90d":14,"forks30d":14,"starsTrendScore":17,"compositeScore":17,"rankGlobal":9,"rankLanguage":9,"license":18,"archived":19,"fork":19,"defaultBranch":20,"hasWiki":21,"hasPages":19,"topics":22,"createdAt":9,"pushedAt":9,"updatedAt":23,"readmeContent":24,"aiSummary":25,"trendingCount":14,"starSnapshotCount":14,"syncStatus":26,"lastSyncTime":27,"discoverSource":28},80128,"ComfyUI_SenseNova_U1","smthemex\u002FComfyUI_SenseNova_U1","smthemex","enseNova-U1: Unifying Multimodal Understanding and Generation with NEO-Unify Architecture",null,"Python",60,9,54,0,1,4,3,"Apache License 2.0",false,"main",true,[],"2026-06-12 02:03:58","# ComfyUI_SenseNova_U1\n[SenseNova-U1](https:\u002F\u002Fgithub.com\u002FOpenSenseNova\u002FSenseNova-U1): Unifying Multimodal Understanding and Generation with NEO-Unify Architecture\n\n# Update\n* 暂时以同步卸载模式适配moe模型,moe模式12G显存， 文生图时，count设置为8-9，图生图设置为4左右..\n* support SenseNova-U1-A3B-MoT-SFT and SenseNova-U1-A3B-MoT and SenseNova-U1-A3B-MoT-SFT-gguf ,支持A3B MOE的单体合并模型和gguf模型,快速测试可以修改节点加载repo的分割模型.\n* fix interleave some bugs ,add interleave max images number, 修复bug，交叉模式生成图片数量可以选入参，注意因为kv缓存的原因，越大越占用显存\n* support 8 steps lora now  支持 8步lora\n* Test it use 8G Vram 36G Ram ,确保内存（不是显存）大于36G\n* If Vram >16G make prefetch_count =0, 显存大于16G时，设置swap（prefetch_count）数值为0以关闭层交换（使用Q6 gguf时）\n\n1.Installation  \n-----\n  In the .\u002FComfyUI\u002Fcustom_nodes directory, run the following:   \n```\ngit clone https:\u002F\u002Fgithub.com\u002Fsmthemex\u002FComfyUI_SenseNova_U1\n```\n2.requirements  \n----\n```\npip install -r requirements.txt\n```\n\n3.checkpoints \n----\n[8B-links](https:\u002F\u002Fhuggingface.co\u002Fsmthem\u002FSenseNova-U1-8B-MoT-Merger-gguf)  \n[A3B-links](https:\u002F\u002Fhuggingface.co\u002Fsmthem\u002FSenseNova-U1-A3B-MoT-SFT-gguf)  \n[A3B-links-modelscope](https:\u002F\u002Fwww.modelscope.cn\u002Fmodels\u002Fsmthem\u002FSenseNova-U1-A3B-MoT-SFT) \n[lora](https:\u002F\u002Fhuggingface.co\u002Fsensenova\u002FSenseNova-U1-8B-MoT-LoRAs)\n[夸克网盘](https:\u002F\u002Fpan.quark.cn\u002Fs\u002F8180628d73c5)\n    \n```\n├── ComfyUI\u002Fmodels\u002Fgguf\u002F\n|     ├── SenseNova-U1-8B-MoT-8step-Q6_K.gguf # optional 可选\n|     ├──SenseNova-U1-A3B-MoT-SFT-Q4_K_S.gguf # optional 可选\n├── ComfyUI\u002Fmodels\u002Fdiffusion_models\u002F\n|     ├── SenseNova-U1-8B-MoT-8step-merge_bf16.safetensors # optional 可选\n|     ├── SenseNova-U1-A3B-MoT-SFT-merge_bf16.safetensors  可选\n├── ComfyUI\u002Fmodels\u002Floras\u002F\n|     ├── SenseNova-U1-8B-MoT-LoRA-8step-V1.0.safetensors # optional 可选\n\n```\n\n4. Example\n----\n* A3B MOE\n![](https:\u002F\u002Fgithub.com\u002Fsmthemex\u002FComfyUI_SenseNova_U1\u002Fblob\u002Fmain\u002Fexample_workflows\u002Fexample_a3bedit.png)\n![](https:\u002F\u002Fgithub.com\u002Fsmthemex\u002FComfyUI_SenseNova_U1\u002Fblob\u002Fmain\u002Fexample_workflows\u002Fexample_a3btest.png)\n* 8B\n![](https:\u002F\u002Fgithub.com\u002Fsmthemex\u002FComfyUI_SenseNova_U1\u002Fblob\u002Fmain\u002Fexample_workflows\u002Fexample_in.png)\n![](https:\u002F\u002Fgithub.com\u002Fsmthemex\u002FComfyUI_SenseNova_U1\u002Fblob\u002Fmain\u002Fexample_workflows\u002Fexample_lora.png)\n![](https:\u002F\u002Fgithub.com\u002Fsmthemex\u002FComfyUI_SenseNova_U1\u002Fblob\u002Fmain\u002Fexample_workflows\u002Fexample_edit.png)\n![](https:\u002F\u002Fgithub.com\u002Fsmthemex\u002FComfyUI_SenseNova_U1\u002Fblob\u002Fmain\u002Fexample_workflows\u002Fexample_ti2i.png)\n![](https:\u002F\u002Fgithub.com\u002Fsmthemex\u002FComfyUI_SenseNova_U1\u002Fblob\u002Fmain\u002Fexample_workflows\u002Fexample_t2i.png)\n\nCitation\n-----\n","ComfyUI_SenseNova_U1 是一个基于 NEO-Unify 架构的多模态理解和生成工具。该项目通过统一架构支持文本到图像、图像到图像等多种任务，其核心功能包括对 SenseNova-U1-A3B-MoT-SFT 和 SenseNova-U1-8B-MoT 模型的支持，以及 8 步 Lora 的兼容性，能够在不同显存条件下高效运行。技术上，项目采用 Python 开发，并提供了详细的安装和配置指南，用户可以根据自身硬件条件调整参数以优化性能。适用于需要进行高质量多模态内容生成的研究者或开发者，特别是在资源有限的情况下追求高效模型训练与推理的应用场景。",2,"2026-06-11 03:59:21","CREATED_QUERY"]