LoRA XL 模型 - CivitAI 仓库 🌠

欢迎访问 Hugging Face 与 CivitAI 上的 FFusion LoRA 提取模型仓库！在这里，我们呈现一组使用低秩适应（LoRA）技术提取的模型，旨在为研究和进一步探索提供丰富的数据集。

🌌 FFusion 的 🧪 精选 LoRA 提取宇宙

我们的 LoRA 是从多种模型中精心提取而成，让您能够自由组合不同风格，创造出真正独特的艺术融合。这些提取的 LoRA 并非直接复制，而是捕捉了原始模型的精髓，并增添了“以……风格”或“受……影响”的创意影响。

🧫 以研究为导向的 LoRA

请注意，所有 FFusionAI 提取的 LoRA 仅限于研究用途，不授权用于商业用途。我们鼓励以负责任和合乎道德的方式使用这些 LoRA，以推动AI 艺术创作领域的发展。

⚠️ 许可与使用声明

请在访问或使用模型前仔细阅读完整的许可协议。

正确的许可与权限信息请见：
https://huggingface.co/FFusion/
https://huggingface.co/FFusion/FFXL400/blob/main/LICENSE.md

模型权重：模型/LoRA 所使用的权重“按原样”提供。FFusion AI 和 Source Code Bulgaria 不授予任何商业使用权。这些权重严格用于测试和实验目的。

LoRA 来源：

所提供的 LoRA 和权重是从 SDXL 模型（检查点）中提取的。

必须尊重并遵守原始检查点创建者设定的所有许可、条款和条件。

🔴 本仓库中的模型和权重严格仅限研究与测试用途，除非下文另有说明。它们不用于一般商业用途，且依赖于每个单独的 LoRA。

🔵 商业用途例外：FFusionXL-BASE、FFusion-BaSE、di.FFUSION.ai-v2.1-768-BaSE-alpha 和 di.ffusion.ai.Beta512 模型由 FFusion AI 使用我们拥有授权的图像训练而成。建议用户优先使用这些模型以获得更安全的体验。这些特定模型允许用于商业用途。

🔴 免责声明：FFusion AI 联合 Source Code Bulgaria Ltd 和 BlackswanTechnologies 不认可或担保每个 LoRA 权重所产生的内容。生成 NSFW 或冒犯性内容的可能性存在。我们共同明确声明不对这些权重生成的结果和内容负责。

🔴 致谢：FFusionXL-BASE 模型是 FFusion AI 独特开发的版本。该模型及相关修改的权利归属于 FFusion AI 和 Source Code Bulgaria Ltd。请确保遵守本许可以及 Stability AI Ltd 对所引用模型设定的任何条件。

增强的 LoRA 灵活性

• 动态范围：通过我们灵活的 LoRA 设置，释放图像的全部潜力，提供从 0.2（细微效果）到 2.2（强烈转变）的广泛范围。此扩展范围超越了标准限制，为您提供无与伦比的控制力，以精确微调视觉效果。

无与伦比的定制性

与限制 LoRA 强度在狭窄范围内的传统模型不同，FFusionAI 提供无与伦比的灵活性。您可以将 LoRA 强度从 0.2（微妙效果）调整至 2.2（强烈转变）。这一扩展范围确保您拥有实现完美风格融合所需的工具，无论基础模型或期望结果如何。

🌟 FF100+ 推荐强度设置 🌟

🎨 视觉效果：提升至 2.2，以获得生动鲜明的细节。

🔗 融合 LoRA：保持 0.3 - 1.0，以安全无缝地融合 FF100 以上的最多 6 个 FF LoRA。

📚 主要测试基础模型：

• Hugging Face: FFusionXL-BASE

• CivitAI: FFusionXL-BASE

📢 更新：2023 年 10 月 22 日 📆

🌟 介绍我们即将推出的 LoRA 系列 FF.100 至 FF.176！

📈 优化后的新尺寸：约 200 - 400MB（取决于原始模型的训练和权重）
🏷️ 新命名方式：在 Hugging Face 上实现更快的推理与测试体验。

pipe = DiffusionPipeline.from_pretrained("FFusion/FFusionXL-BASE", torch_dtype=torch.float16).to("cuda")
lora_model_id = "FFusion/400GB-LoraXL"
 lora_filename = "FF.101.safetensors"
pipe.load_lora_weights(lora_model_id, weight_name=lora_filename)

CivitAI 命名格式保持不变。
在 diffusers 中从 CivitAI 加载

待办事项：🔄 同步 CivitAI 仓库：已更新至 FF98

最新 FF60-FF98

**
**

Model: sdxlYamersRealism_version2 - Status: Text encoder is different. 0.0048828125
Model: animeChangefulXL_v10ReleasedCandidate - Status: Text encoder is different. 0.00390625
Model: brixlAMustInYour_v20Banu - Status: Text encoder is different. 0.001434326171875
Model: cinemaxAlphaSDXLCinema_alpha1 - Status: Text encoder is different. 0.00311279296875
Model: copaxTimelessxlSDXL1_v5 - Status: Text encoder is same.
Model: dreamshaperXL10_alpha2Xl10 - Status: Text encoder is same.
Model: endjourneyXL_v11 - Status: Text encoder is different. 0.0029296875
Model: explicitFreedomNSFW_beta - Status: Text encoder is different. 0.001220703125
Model: FinalAnimeCG_mk2a2 - Status: Text encoder is different. 0.00390625
Model: formulaxlXLComfyui_v20Pruned - Status: Text encoder is different. 0.002643585205078125
Model: furtasticxl_BetaEPOCHS3 - Status: Text encoder is different. 0.013824462890625
Model: galaxytimemachinesGTM_xlplusV10 - Status: Text encoder is different. 0.0012865066528320312
Model: hassakuSfwNsfwAlphav_alphaV02 - Status: Text encoder is different. 0.00390625
Model: juggernautXL_version4 - Status: Text encoder is different. 0.0019378662109375
Model: MOHAWK_v10BETA - Status: Text encoder is different. 0.00103759765625
Model: newone_v10 - Status: Text encoder is different. 0.001190185546875
Model: nightvisionXLPhotorealisticPortrait_v0743ReleaseBakedvae - Status: Text encoder is different. 0.009429931640625
Model: pyrosNSFWSDXL_v013e6 - Status: Text encoder is same.
Model: pyrosSDModelsBlowjob_v0122022steps - Status: Text encoder is same.
Model: realisticFreedomSFW_alpha - Status: Text encoder is different. 0.0011749267578125
Model: realisticStockPhoto_v10 - Status: Text encoder is different. 0.0011444091796875
Model: RealitiesEdgeXLANIME_20 - Status: Text encoder is different. 0.0018310546875
Model: RealitiesEdgeXL_30 - Status: Text encoder is different. 0.004150390625
Model: realvisxlV10_v10VAE - Status: Text encoder is different. 0.0029296875
Model: samaritan3dCartoon_v40SDXL - Status: Text encoder is different. 0.00390625
Model: sdvn6Realxl_detailface - Status: Text encoder is same.
Model: sdxlNuclearGeneralPurposeSemi_v10 - Status: Text encoder is different. 0.003021240234375
Model: sdxlUnstableDiffusers_v6StabilityEater - Status: Text encoder is different. 0.0029296875
Model: sdxlYamersRealism_version2 - Status: Text encoder is different. 0.0048828125
Model: unsafexl_v20 - Status: Text encoder is different. 0.068359375
Model: venusxl_v11 - Status: Text encoder is different. 0.0013863444328308105
Model: xlYamersCartoonArcadia_v1 - Status: Text encoder is different. 0.0029296875
Model: FFusionXL-BASE-v1 - Status: Text encoder is different. 0.032245635986328125
Model: FFXL-400-v2 - Status: Text encoder is different. 0.023212432861328125
Model: FFXL400-LoRA-XL-FFusion-v1 - Status: Text encoder is different. 0.020404815673828125
Model: FFXXL-400-v2 - Status: Text encoder is different. 0.00948333740234375
Model: realcartoonXL_v2 - Status: Text encoder is different. 0.0015802383422851562

Model: sdxlYamersRealism_version2.FFai.lora64.safetensors
UNet weight average magnitude: 4.154722048359913
UNet weight average strength: 0.010771295011342323
UNet Conv weight average magnitude: 4.015763928139778
UNet Conv weight average strength: 0.004715556773610134
Text Encoder (1) weight average magnitude: 3.958945306529754
Text Encoder (1) weight average strength: 0.013064685133728026
Text Encoder (2) weight average magnitude: 3.9970537933453656
Text Encoder (2) weight average strength: 0.01012922219208529
----------------------------
Model: FF.66.hassakuSfwNsfwAlphav_alphaV02.lora.safetensors
UNet weight average magnitude: 4.6113617624162275
UNet weight average strength: 0.011981260592954776
UNet Conv weight average magnitude: 6.686307668617343
UNet Conv weight average strength: 0.006950538604713883
Text Encoder (1) weight average magnitude: 3.807746602732888
Text Encoder (1) weight average strength: 0.012745779610859834
Text Encoder (2) weight average magnitude: 3.729743715233202
Text Encoder (2) weight average strength: 0.009551327927254742
----------------------------
Model: FF.67.galaxytimemachinesGTM_xlplusV10.lora.safetensors
UNet weight average magnitude: 5.2081857497500135
UNet weight average strength: 0.012861152998866098
UNet Conv weight average magnitude: 6.477215331015863
UNet Conv weight average strength: 0.005731545812523109
Text Encoder (1) weight average magnitude: 3.865321475649114
Text Encoder (1) weight average strength: 0.012968309181164591
Text Encoder (2) weight average magnitude: 3.791585137796209
Text Encoder (2) weight average strength: 0.009739622211064131
----------------------------
Model: FF.68.furtasticxl_BetaEPOCHS3.lora.safetensors
UNet weight average magnitude: 4.82028448554389
UNet weight average strength: 0.012252009690673311
UNet Conv weight average magnitude: 6.774379998733585
UNet Conv weight average strength: 0.007177153983462227
Text Encoder (1) weight average magnitude: 4.20241893596518
Text Encoder (1) weight average strength: 0.01346020465857439
Text Encoder (2) weight average magnitude: 4.260738640446866
Text Encoder (2) weight average strength: 0.010471828656006711
----------------------------
Model: FF.69.formulaxlXLComfyui_v20Pruned.lora.safetensors
UNet weight average magnitude: 4.194797467480407
UNet weight average strength: 0.010794051441520451
UNet Conv weight average magnitude: 5.658129971781666
UNet Conv weight average strength: 0.004699672960547711
Text Encoder (1) weight average magnitude: 3.9974802957054556
Text Encoder (1) weight average strength: 0.013097433444426298
Text Encoder (2) weight average magnitude: 4.090353610501367
Text Encoder (2) weight average strength: 0.010226978548569817
----------------------------
Model: FF.70.FinalAnimeCG_mk2a2.lora.safetensors
UNet weight average magnitude: 5.832734982003316
UNet weight average strength: 0.013620979564593433
UNet Conv weight average magnitude: 6.588312134998715
UNet Conv weight average strength: 0.006310420276329548
Text Encoder (1) weight average magnitude: 3.856879807170544
Text Encoder (1) weight average strength: 0.012947154068967848
Text Encoder (2) weight average magnitude: 3.7769155501438316
Text Encoder (2) weight average strength: 0.009654614341923677
----------------------------
Model: FF.71.explicitFreedomNSFW_beta.lora.safetensors
UNet weight average magnitude: 4.501298830893416
UNet weight average strength: 0.01109003259855744
UNet Conv weight average magnitude: 6.204555848757276
UNet Conv weight average strength: 0.005750268214362425
Text Encoder (1) weight average magnitude: 3.85944453350698
Text Encoder (1) weight average strength: 0.012919606802022875
Text Encoder (2) weight average magnitude: 3.9375385889629477
Text Encoder (2) weight average strength: 0.010088601556714144
----------------------------
Model: FF.72.endjourneyXL_v11.lora.safetensors
UNet weight average magnitude: 4.202640614034873
UNet weight average strength: 0.010788684869548844
UNet Conv weight average magnitude: 5.80301284455635
UNet Conv weight average strength: 0.005029451652697187
Text Encoder (1) weight average magnitude: 3.835258093635928
Text Encoder (1) weight average strength: 0.012878727225694529
Text Encoder (2) weight average magnitude: 3.7550355683040344
Text Encoder (2) weight average strength: 0.009627099200498888
----------------------------
Model: FF.73.dreamshaperXL10_alpha2Xl10.lora.safetensors
UNet weight average magnitude: 3.859263254032285
UNet weight average strength: 0.010177448403109668
UNet Conv weight average magnitude: 0.0
UNet Conv weight average strength: 0.0
Text Encoder: Not Found
----------------------------
Model: FF.74.copaxTimelessxlSDXL1_v5.lora.safetensors
UNet weight average magnitude: 4.006565464438231
UNet weight average strength: 0.010389718183037322
UNet Conv weight average magnitude: 5.738000089710234
UNet Conv weight average strength: 0.0048703539869873365
Text Encoder: Not Found
----------------------------
Model: FF.75.cinemaxAlphaSDXLCinema_alpha1.lora.safetensors
UNet weight average magnitude: 4.466204403397648
UNet weight average strength: 0.011222293042751443
UNet Conv weight average magnitude: 5.684097723570108
UNet Conv weight average strength: 0.004689726735887235
Text Encoder (1) weight average magnitude: 3.9233677697347935
Text Encoder (1) weight average strength: 0.013047985608868315
Text Encoder (2) weight average magnitude: 3.967672834668905
Text Encoder (2) weight average strength: 0.010161683571519127
----------------------------
Model: FF.76.brixlAMustInYour_v20Banu.lora.safetensors
UNet weight average magnitude: 5.201652157233597
UNet weight average strength: 0.012340885235722432
UNet Conv weight average magnitude: 6.246570986909302
UNet Conv weight average strength: 0.005628776318139394
Text Encoder (1) weight average magnitude: 3.7901131354041215
Text Encoder (1) weight average strength: 0.012251635754363702
Text Encoder (2) weight average magnitude: 3.9011343266469787
Text Encoder (2) weight average strength: 0.009675557128661683
----------------------------
Model: FF.77.animeChangefulXL_v10ReleasedCandidate.lora.safetensors
UNet weight average magnitude: 4.8712592588918255
UNet weight average strength: 0.011882757534620026
UNet Conv weight average magnitude: 6.307265147238472
UNet Conv weight average strength: 0.005707653219309981
Text Encoder (1) weight average magnitude: 3.806143895360976
Text Encoder (1) weight average strength: 0.012739821013629662
Text Encoder (2) weight average magnitude: 3.7378093050117975
Text Encoder (2) weight average strength: 0.009586058803350757
----------------------------
Model: FF.78.xlYamersCartoonArcadia_v1.lora.safetensors
UNet weight average magnitude: 4.353353198959002
UNet weight average strength: 0.010753757289463425
UNet Conv weight average magnitude: 5.9177157902332835
UNet Conv weight average strength: 0.0051653985959496315
Text Encoder (1) weight average magnitude: 3.8127760281067853
Text Encoder (1) weight average strength: 0.012772330040804636
Text Encoder (2) weight average magnitude: 3.764581932297466
Text Encoder (2) weight average strength: 0.009682294095990565
----------------------------
Model: FF.79.venusxl_v11.lora.safetensors
UNet weight average magnitude: 4.0781163529498725
UNet weight average strength: 0.01056802143213069
UNet Conv weight average magnitude: 5.725042873950945
UNet Conv weight average strength: 0.004766753768581111
Text Encoder (1) weight average magnitude: 3.8819661703272876
Text Encoder (1) weight average strength: 0.01297504551077796
Text Encoder (2) weight average magnitude: 3.8989897630581978
Text Encoder (2) weight average strength: 0.00999233670699671
----------------------------
Model: FF.80.unsafexl_v20.lora.safetensors
UNet weight average magnitude: 4.433128703574937
UNet weight average strength: 0.01126235056722307
UNet Conv weight average magnitude: 5.6776551531768105
UNet Conv weight average strength: 0.004711627911345002
Text Encoder (1) weight average magnitude: 3.9928442365475028
Text Encoder (1) weight average strength: 0.013100078304973888
Text Encoder (2) weight average magnitude: 3.945462724939238
Text Encoder (2) weight average strength: 0.010062376848996262
----------------------------
Model: FF.81.sdxlYamersRealism_version2.lora.safetensors
UNet weight average magnitude: 4.229406260655774
UNet weight average strength: 0.01076863108078825
UNet Conv weight average magnitude: 5.653783535189452
UNet Conv weight average strength: 0.004649401315378378
Text Encoder (1) weight average magnitude: 3.958945306529754
Text Encoder (1) weight average strength: 0.013064685133728026
Text Encoder (2) weight average magnitude: 3.9970537933453656
Text Encoder (2) weight average strength: 0.01012922219208529
----------------------------
Model: FF.82.sdxlUnstableDiffusers_v6StabilityEater.lora.safetensors
UNet weight average magnitude: 4.387654105095919
UNet weight average strength: 0.010840575656477952
UNet Conv weight average magnitude: 5.859291158408854
UNet Conv weight average strength: 0.004964447160293478
Text Encoder (1) weight average magnitude: 3.8646596391683863
Text Encoder (1) weight average strength: 0.012911755181541458
Text Encoder (2) weight average magnitude: 3.840901404987889
Text Encoder (2) weight average strength: 0.009815472265736007
----------------------------
Model: FF.83.sdxlNuclearGeneralPurposeSemi_v10.lora.safetensors
UNet weight average magnitude: 4.329690552630377
UNet weight average strength: 0.01081156604611163
UNet Conv weight average magnitude: 5.754435529197304
UNet Conv weight average strength: 0.004791491470688117
Text Encoder (1) weight average magnitude: 3.908995280978119
Text Encoder (1) weight average strength: 0.012984716052686607
Text Encoder (2) weight average magnitude: 3.8730233638208733
Text Encoder (2) weight average strength: 0.009816295838443996
----------------------------
Model: FF.84.sdvn6Realxl_detailface.lora.safetensors
UNet weight average magnitude: 3.9204966894076203
UNet weight average strength: 0.010152018695796424
UNet Conv weight average magnitude: 5.609827023476847
UNet Conv weight average strength: 0.004578104347668462
Text Encoder: Not Found
----------------------------
Model: FF.85.samaritan3dCartoon_v40SDXL.lora.safetensors
UNet weight average magnitude: 4.1930053871423265
UNet weight average strength: 0.010823639858269587
UNet Conv weight average magnitude: 6.242507300692357
UNet Conv weight average strength: 0.006012499761466946
Text Encoder (1) weight average magnitude: 3.807746602732888
Text Encoder (1) weight average strength: 0.012745779610859834
Text Encoder (2) weight average magnitude: 3.729743715233202
Text Encoder (2) weight average strength: 0.009551327927254742
----------------------------
Model: FF.86.realvisxlV10_v10VAE.lora.safetensors
UNet weight average magnitude: 4.035726046516959
UNet weight average strength: 0.01043685083171328
UNet Conv weight average magnitude: 5.780022388037139
UNet Conv weight average strength: 0.0049551385295671935
Text Encoder (1) weight average magnitude: 3.862534960968426
Text Encoder (1) weight average strength: 0.01291815120168007
Text Encoder (2) weight average magnitude: 3.8792245692334855
Text Encoder (2) weight average strength: 0.010027987691388776
----------------------------
Model: FF.87.RealitiesEdgeXLANIME_20.lora.safetensors
UNet weight average magnitude: 4.322741449452443
UNet weight average strength: 0.011017050541178184
UNet Conv weight average magnitude: 5.957632120776351
UNet Conv weight average strength: 0.005321540223768453
Text Encoder (1) weight average magnitude: 3.9027693617053862
Text Encoder (1) weight average strength: 0.013066310297084008
Text Encoder (2) weight average magnitude: 3.941240896860996
Text Encoder (2) weight average strength: 0.010187814902599733
----------------------------
Model: FF.88.RealitiesEdgeXL_30.lora.safetensors
UNet weight average magnitude: 4.527436449035657
UNet weight average strength: 0.011438576163998578
UNet Conv weight average magnitude: 6.042128532601058
UNet Conv weight average strength: 0.0053643976503331536
Text Encoder (1) weight average magnitude: 3.96435868300754
Text Encoder (1) weight average strength: 0.013183793628117942
Text Encoder (2) weight average magnitude: 4.03501811478197
Text Encoder (2) weight average strength: 0.01033219734045475
----------------------------
Model: FF.89.realisticStockPhoto_v10.lora.safetensors
UNet weight average magnitude: 4.178010046544553
UNet weight average strength: 0.01060077238986419
UNet Conv weight average magnitude: 5.832883513120958
UNet Conv weight average strength: 0.005094057992644391
Text Encoder (1) weight average magnitude: 3.838598740372775
Text Encoder (1) weight average strength: 0.012775584451815206
Text Encoder (2) weight average magnitude: 3.8534473782218375
Text Encoder (2) weight average strength: 0.009703626948148766
----------------------------
Model: FF.90.realisticFreedomSFW_alpha.lora.safetensors
UNet weight average magnitude: 4.570225351823505
UNet weight average strength: 0.011338880456799554
UNet Conv weight average magnitude: 6.107921122775599
UNet Conv weight average strength: 0.005313926393612039
Text Encoder (1) weight average magnitude: 3.9145800451769137
Text Encoder (1) weight average strength: 0.012987243885510853
Text Encoder (2) weight average magnitude: 3.9456476675702756
Text Encoder (2) weight average strength: 0.010086475486504298
----------------------------
Model: FF.91.realcartoonXL_v2.lora.safetensors
UNet weight average magnitude: 4.264556294830096
UNet weight average strength: 0.010837268212782766
UNet Conv weight average magnitude: 5.775273580445967
UNet Conv weight average strength: 0.004823115907624419
Text Encoder (1) weight average magnitude: 3.868685000881062
Text Encoder (1) weight average strength: 0.012967535154814412
Text Encoder (2) weight average magnitude: 3.8942008722126786
Text Encoder (2) weight average strength: 0.009956078788817995
----------------------------
Model: FF.92.pyrosSDModelsBlowjob_v0122022steps.lora.safetensors
UNet weight average magnitude: 4.29299465986103
UNet weight average strength: 0.011065152509191439
UNet Conv weight average magnitude: 6.148179389228268
UNet Conv weight average strength: 0.005785365500822891
Text Encoder: Not Found
----------------------------
Model: FF.93.pyrosNSFWSDXL_v013e6.lora.safetensors
UNet weight average magnitude: 4.462978487594761
UNet weight average strength: 0.011458003048327881
UNet Conv weight average magnitude: 6.365678967519903
UNet Conv weight average strength: 0.006252718402740558
Text Encoder: Not Found
----------------------------
Model: FF.94.nightvisionXLPhotorealisticPortrait_v0743ReleaseBakedvae.lora.safetensors
UNet weight average magnitude: 4.30821859959078
UNet weight average strength: 0.01092674471500856
UNet Conv weight average magnitude: 5.760595716272804
UNet Conv weight average strength: 0.0047913433799900915
Text Encoder (1) weight average magnitude: 4.082814836813033
Text Encoder (1) weight average strength: 0.013277437149876429
Text Encoder (2) weight average magnitude: 4.269554751742187
Text Encoder (2) weight average strength: 0.0104525629385582
----------------------------
Model: FF.95.newone_v10.lora.safetensors
UNet weight average magnitude: 3.9863974933790827
UNet weight average strength: 0.010221166935769414
UNet Conv weight average magnitude: 5.591587011383119
UNet Conv weight average strength: 0.004544408523927106
Text Encoder (1) weight average magnitude: 3.826913276992613
Text Encoder (1) weight average strength: 0.012515731668562081
Text Encoder (2) weight average magnitude: 3.7789877235680827
Text Encoder (2) weight average strength: 0.008847150427050579
----------------------------
Model: FF.96.MOHAWK_v10BETA.lora.safetensors
UNet weight average magnitude: 4.13427196290026
UNet weight average strength: 0.010604709463386349
UNet Conv weight average magnitude: 5.906059771550209
UNet Conv weight average strength: 0.005266774851315859
Text Encoder (1) weight average magnitude: 3.8816106810049615
Text Encoder (1) weight average strength: 0.013007851116722372
Text Encoder (2) weight average magnitude: 3.795246249757246
Text Encoder (2) weight average strength: 0.009741588405668723
----------------------------
Model: FF.97.juggernautXL_version4.lora.safetensors
UNet weight average magnitude: 4.351658373013424
UNet weight average strength: 0.01097575598820061
UNet Conv weight average magnitude: 5.7254163997882515
UNet Conv weight average strength: 0.0048427100518286656
Text Encoder (1) weight average magnitude: 3.98009165065858
Text Encoder (1) weight average strength: 0.013189073899460014
Text Encoder (2) weight average magnitude: 4.452439746998783
Text Encoder (2) weight average strength: 0.010877184808674183
----------------------------
Model: FF.98.sdxlYamersRealism_version2.lora.safetensors
UNet weight average magnitude: 4.229406260655774
UNet weight average strength: 0.01076863108078825
UNet Conv weight average magnitude: 5.653783535189452
UNet Conv weight average strength: 0.004649401315378378
Text Encoder (1) weight average magnitude: 3.958945306529754
Text Encoder (1) weight average strength: 0.013064685133728026
Text Encoder (2) weight average magnitude: 3.9970537933453656
Text Encoder (2) weight average strength: 0.01012922219208529
----------------------------

📦 基础模型

以下模型构成了我们提取工作的基础：

• stabilityai/stable-diffusion-xl-base-1.0

• stabilityai/stable-diffusion-xl-base-0.9

• FFusion/FFusionXL-BASE

🌟 推荐用于推理的模型

对于正在寻找理想模型以驱动其推理任务的用户，我们特别推荐：

• FFusionXL-BASE —— 我们的标志性基础模型，使用授权图像精心训练而成。

• FFXL400 组合 LoRA 模型 🚀 —— LoRA 模型世界中力量与精度的星际融合。

请放心，我们的 LoRA 即使在权重为 1.0 时，也与大多数当前的 SDXL 模型保持兼容。

🔍 提取细节

• 变体：每个基础模型被提取为 4-5 个不同的变体。

• 提取深度：此处上传的模型包含约 70% 的提取数据。这些提取生成的数据集大小约为 400 GB。

• 精度：我们尝试了 float32 和 float64 以获得最优的提取结果。

• 差异测量：使用奇异值分解（SVD）测量原始模型与微调模型之间的差异。通常使用 1e-3 的阈值，但在某些情况下也测试了 1e-5 和 1e-2。

• 演示参数：在我们的演示中，我们使用了 "conv_dim": 256 和 "conv_alpha": 256。

⚙️ 技术说明

• 本集合中的大多数 SDXL 模型并非传统意义上的“训练”所得。相反，它们是通过 Comfy UI 辅助，从之前的 SDXL 0.9 版本合并或使用其他方法创建的。

• 用户请注意：所有使用 Comfy 保存的模型都会额外添加一个键 text_model.encoder.text_model.embeddings.position_ids。我们已进行必要调整，以确保与 Kohya 的当前脚本兼容。

📈 使用场景

这些提取的模型旨在用于研究和测试。它们尤其适用于：

• 探索合并多个 LoRA 的潜力。

• 同时对 1-5 个 LoRA 进行权重实验。

• 探究从不同基础模型中提取出的 LoRA 之间的差异与相似性。

FFusion LoRA 提取模型 —— 使用指南 🧠

欢迎使用 FFusion LoRA 提取模型的技术指南。本文将引导您完成融合 LoRA 参数、加载检查点及执行推理的步骤。

融合 LoRA 参数 🔗

为将 LoRA 参数与底层模型的原始参数合并，从而可能加快推理延迟：

pipe.fuse_lora()

取消融合 LoRA 参数 ⛓️

要撤销 fuse_lora() 的效果：

pipe.unfuse_lora()

使用不同 LoRA 缩放比例 🎚️

为控制 LoRA 参数对输出的影响：

pipe.fuse_lora(lora_scale=0.5)

使用 FFusion 模型 🔍

以下是加载和使用我们 FFusion 模型的方法：

from diffusers import DiffusionPipeline
import torch

pipeline_id = "FFusion/FFusionXL-BASE"

pipe = DiffusionPipeline.from_pretrained(pipeline_id, torch_dtype=torch.float16)
pipe.enable_model_cpu_offload()

lora_model_id = "FFusion/400GB-LoraXL"
lora_filename = "FFai.0038.Realitycheckxl_Alpha11.lora.safetensors"
pipe.load_lora_weights(lora_model_id, weight_name=lora_filename)

prompt = "papercut sonic"
image = pipe(prompt=prompt, num_inference_steps=20, generator=torch.manual_seed(0)).images[0]
image

执行推理 🖼️

加载所需模型后，您可以按如下方式执行推理：

generator = torch.manual_seed(0)
images_fusion = pipe(
    "masterpiece, best quality, mountain", output_type="np", generator=generator, num_inference_steps=25
).images

可用 LoRA 模型库 📚

您可以从我们在 Hugging Face 的仓库或即将在 CivitAI 上发布的仓库中选择任何模型。以下是使用 lora_model_id = "FFusion/400GB-LoraXL" 的可用模型列表：

lora_filename = 
    - FFai.0001.4Guofeng4xl_V1125d.lora_Dim64.safetensors
    - FFai.0002.4Guofeng4xl_V1125d.lora_Dim8.safetensors
    - FFai.0003.4Guofeng4xl_V1125d.loraa.safetensors
    - FFai.0004.Ambiencesdxl_A1.lora.safetensors
    - FFai.0005.Ambiencesdxl_A1.lora_8.safetensors
    - FFai.0006.Angrasdxl10_V22.lora.safetensors
    - FFai.0007.Animaginexl_V10.lora.safetensors
    - FFai.0008.Animeartdiffusionxl_Alpha3.lora.safetensors
    - FFai.0009.Astreapixiexlanime_V16.lora.safetensors
    - FFai.0010.Bluepencilxl_V010.lora.safetensors
    - FFai.0011.Bluepencilxl_V021.lora.safetensors
    - FFai.0012.Breakdomainxl_V03d.lora.safetensors
    - FFai.0013.Canvasxl_Bfloat16v002.lora.safetensors
    - FFai.0014.Cherrypickerxl_V20.lora.safetensors
    - FFai.0015.Copaxtimelessxlsdxl1_V44.lora.safetensors
    - FFai.0016.Counterfeitxl-Ffusionai-Alpha-Vae.lora.safetensors
    - FFai.0017.Counterfeitxl_V10.lora.safetensors
    - FFai.0018.Crystalclearxl_Ccxl.lora.safetensors
    - FFai.0019.Deepbluexl_V006.lora.safetensors
    - FFai.0020.Dream-Ffusion-Shaper.lora.safetensors
    - FFai.0021.Dreamshaperxl10_Alpha2xl10.lora.safetensors
    - FFai.0022.Duchaitenaiartsdxl_V10.lora.safetensors
    - FFai.0023.Dynavisionxlallinonestylized_Beta0371bakedvae.lora.safetensors
    - FFai.0024.Dynavisionxlallinonestylized_Beta0411bakedvae.lora.safetensors
    - FFai.0025.Fantasticcharacters_V55.lora.safetensors
    - FFai.0026.Fenrisxl_V55.lora.safetensors
    - FFai.0027.Fudukimix_V10.lora.safetensors
    - FFai.0028.Infinianimexl_V16.lora.safetensors
    - FFai.0029.Juggernautxl_Version1.lora_1.safetensors
    - FFai.0030.Lahmysterioussdxl_V330.lora.safetensors
    - FFai.0031.Mbbxlultimate_V10rc.lora.safetensors
    - FFai.0032.Miamodelsfwnsfwsdxl_V30.lora.safetensors
    - FFai.0033.Morphxl_V10.lora.safetensors
    - FFai.0034.Nightvisionxlphotorealisticportrait_Beta0681bakedvae.lora_1.safetensors
    - FFai.0035.Osorubeshialphaxl_Z.lora.safetensors
    - FFai.0036.Physiogenxl_V04.lora.safetensors
    - FFai.0037.Protovisionxlhighfidelity3d_Beta0520bakedvae.lora.safetensors
    - FFai.0038.Realitycheckxl_Alpha11.lora.safetensors
    - FFai.0039.Realmixxl_V10.lora.safetensors
    - FFai.0040.Reproductionsdxl_V31.lora.safetensors
    - FFai.0041.Rundiffusionxl_Beta.lora.safetensors
    - FFai.0042.Samaritan3dcartoon_V40sdxl.lora.safetensors
    - FFai.0043.Sdvn6realxl_Detailface.lora.safetensors
    - FFai.0044.Sdvn7realartxl_Beta2.lora.safetensors
    - FFai.0045.Sdxl10arienmixxlasian_V10.lora.safetensors
    - FFai.0046.Sdxlbasensfwfaces_Sdxlnsfwfaces03.lora.safetensors
    - FFai.0047.Sdxlfaetastic_V10.lora.safetensors
    - FFai.0048.Sdxlfixedvaefp16remove_Basefxiedvaev2fp16.lora.safetensors
    - FFai.0049.Sdxlnijiv4_Sdxlnijiv4.lora.safetensors
    - FFai.0050.Sdxlronghua_V11.lora.safetensors
    - FFai.0051.Sdxlunstablediffusers_V5unchainedslayer.lora.safetensors
    - FFai.0052.Sdxlyamersanimeultra_Yamersanimev2.lora.safetensors
    - FFai.0053.Shikianimexl_V10.lora.safetensors
    - FFai.0054.Spectrumblendx_V10.lora.safetensors
    - FFai.0055.Stablediffusionxl_V30.lora.safetensors
    - FFai.0056.Talmendoxlsdxl_V11beta.lora.safetensors
    - FFai.0057.Wizard_V10.lora.safetensors
    - FFai.0058.Wyvernmix15xl_Xlv11.lora.safetensors
    - FFai.0059.Xl13asmodeussfwnsfw_V17bakedvae.lora.safetensors
    - FFai.0060.Xl3experimentalsd10xl_V10.lora.safetensors
    - FFai.0061.Xl6hephaistossd10xlsfw_V21bakedvaefp16fix.lora.safetensors
    - FFai.0062.Xlperfectdesign_V2ultimateartwork.lora.safetensors
    - FFai.0063.Xlyamersrealistic_V3.lora.safetensors
    - FFai.0064.Xxmix9realisticsdxl_Testv20.lora.safetensors
    - FFai.0065.Zavychromaxl_B2.lora.safetensors

📊 文本编码器差异概览

根据提取过程，我们观察到不同模型的文本编码器存在以下差异：

• bluePencilXL_v021 ✅ 文本编码器可用。差异为 0.00140380859375

• sdvn7Realartxl_beta2 ✅ 文本编码器可用。差异为 0.00362396240234375

• 4Guofeng4XL_v1125D 🚫 文本编码器不可用。与 SDXL 1.0 基础模型相同

• ambienceSDXL_a1 ✅ 文本编码器可用。差异为 0.003082275390625

• angraSDXL10_v22 ✅ 文本编码器可用。差异为 0.001953125

• animagineXL_v10 🚫 文本编码器不可用。与 SDXL 1.0 基础模型相同

• animeArtDiffusionXL_alpha3 🚫 文本编码器不可用。与 SDXL 1.0 基础模型相同

• astreapixieXLAnime_v16 ✅ 文本编码器可用。差异为 0.0029296875

• bluePencilXL_v010 ✅ 文本编码器可用。差异为 0.00177001953125

• breakdomainxl_v03d ✅ 文本编码器可用。差异为 0.0013427734375

• canvasxl_Bfloat16V002 ✅ 文本编码器可用。差异为 0.00390625

• cherryPickerXL_v20 ✅ 文本编码器可用。差异为 0.0016450881958007812

• copaxTimelessxlSDXL1_v44 🚫 文本编码器不可用。与 SDXL 1.0 基础模型相同

• counterfeitxl_v10 ✅ 文本编码器可用。差异为 0.001708984375

• crystalClearXL_ccxl ✅ 文本编码器可用。差异为 0.0012865066528320312

• deepblueXL_v006 ✅ 文本编码器可用。差异为 0.00200653076171875

• dreamshaperXL10_alpha2Xl10 🚫 文本编码器不可用。与 SDXL 1.0 基础模型相同

• duchaitenAiartSDXL_v10 🚫 文本编码器不可用。与 SDXL 1.0 基础模型相同

• dynavisionXLAllInOneStylized_beta0371Bakedvae ✅ 文本编码器可用。差异为 0.00321197509765625

• dynavisionXLAllInOneStylized_beta0411Bakedvae ✅ 文本编码器可用。差异为 0.0037841796875

• envyoverdrivexl_v11 🚫 文本编码器不可用。与 SDXL 1.0 基础模型相同

• envypoodaxl01_v10 ✅ 文本编码器可用。差异为 0.0011358261108398438

• fantasticCharacters_v55 ✅ 文本编码器可用。差异为 0.00390625

• fenrisxl_V55 ✅ 文本编码器可用。差异为 0.0086822509765625

• fudukiMix_v10 ✅ 文本编码器可用。差异为 0.0011138916015625

• infinianimexl_v16 ✅ 文本编码器可用。差异为 0.0048828125

• juggernautXL_version1 ✅ 文本编码器可用。差异为 0.001953125

• LahMysteriousSDXL_v330 🚫 文本编码器不可用。与 SDXL 1.0 基础模型相同

• mbbxlUltimate_v10RC 🚫 文本编码器不可用。与 SDXL 1.0 基础模型相同

• miamodelSFWNSFWSDXL_v30 ✅ 文本编码器可用。差异为 0.0047607421875

• morphxl_v10 ✅ 文本编码器可用。差异为 0.001861572265625

• nightvisionXLPhotorealisticPortrait_beta0681Bakedvae ✅ 文本编码器可用。差异为 0.013885498046875

• osorubeshiAlphaXL_z ✅ 文本编码器可用。差异为 0.005615234375

• physiogenXL_v04 ✅ 文本编码器可用。差异为 0.00390625

• protovisionXLHighFidelity3D_beta0520Bakedvae ✅ 文本编码器可用。差异为 0.007568359375

• realitycheckXL_alpha11 ✅ 文本编码器可用。差异为 0.0015010833740234375

• realmixXL_v10 ✅ 文本编码器可用。差异为 0.0023899078369140625

• reproductionSDXL_v31 ✅ 文本编码器可用。差异为 0.00146484375

• rundiffusionXL_beta ✅ 文本编码器可用。差异为 0.00196075439453125

• samaritan3dCartoon_v40SDXL ✅ 文本编码器可用。差异为 0.0009765625

• sdvn6Realxl_detailface 🚫 文本编码器不可用。与 SDXL 1.0 基础模型相同

• sdxl10ArienmixxlAsian_v10 ✅ 文本编码器可用。差异为 0.00048828125

• sdxlbaseNsfwFaces_sdxlNsfwFaces03 ✅ 文本编码器可用。差异为 0.008056640625

• sdxlFaetastic_v10 ✅ 文本编码器可用。差异为 0.0029296875

• sdxlFixedvaeFp16Remove_baseFxiedVaeV2Fp16 🚫 文本编码器不可用。与 SDXL 1.0 基础模型相同

• sdxlNijiV4_sdxlNijiV4 ✅ 文本编码器可用。差异为 0.0009765625

• SDXLRonghua_v11 ✅ 文本编码器可用。差异为 0.0009765625

• sdxlUnstableDiffusers_v5UnchainedSlayer ✅ 文本编码器可用。差异为 0.001251220703125

• sdxlYamersAnimeUltra_yamersAnimeV2 ✅ 文本编码器可用。差异为 0.000732421875

• sdXL_v10VAEFix 🚫 文本编码器不可用。与 SDXL 1.0 基础模型相同

• shikianimexl_v10 ✅ 文本编码器可用。差异为 0.0009765625

• spectrumblendx_v10 ✅ 文本编码器可用。差异为 0.0013065338134765625

• stableDiffusionXL_v30 🚫 文本编码器不可用。与 SDXL 1.0 Base 相同

• talmendoxlSDXL_v11Beta 🚫 文本编码器不可用。与 SDXL 1.0 Base 相同

• wizard_v10 ✅ 文本编码器可用。差异为 0.000244140625

🎉 致谢与引用

衷心感谢社区的持续支持与反馈！我们正共同推动机器学习潜力的边界！

同时，我们感谢并致谢以下项目及作者：

• ComfyUI：我们使用并修改了 ComfyUI 的部分代码用于本项目。

• kohya-ss/sd-scripts 和 bmaltais：我们的工作还整合了来自 kohya-ss/sd-scripts 的修改。

• lora-inspector：我们受益于 lora-inspector 项目。

• KohakuBlueleaf：特别感谢 KohakuBlueleaf 的宝贵贡献。

HowMuch ???

你是否曾经自问：“我浪费了多少空间在 *.ckpt 和 *.safetensors 检查点上？” 🤔 介绍 HowMuch：从现在起，开始检查检查点所浪费的空间吧！

😄 享受这款虽有些多余，但“全家都能乐在其中”的磁盘空间分析工具吧。😄

概述

HowMuch 是一个用于扫描驱动器（或指定目录）并报告特定扩展名文件（主要是 .ckpt 和 .safetensors）所占用总空间的 Python 工具。

它输出：

• 每个被扫描驱动器或目录的总存储容量。

• .ckpt 和 .safetensors 文件所占用的空间。

• 剩余可用空间。

• 一个清晰的条形图，可视化上述数据。

安装

GitHub

通过 PyPI 安装

你可以通过 pip 轻松安装 HowMuch：

pip install howmuch

从源码安装

1. 克隆仓库：

   git clone https://github.com/1e-2/HowMuch.git
   

2. 进入克隆目录并安装：

   cd HowMuch
   pip install .
   

使用

不带参数运行工具以扫描所有驱动器：

howmuch

或指定特定目录或驱动器进行扫描：

howmuch --scan C:

🌐 FFusion.ai 联系方式

由 Source Code Bulgaria Ltd 与 Black Swan Technologies 联合维护。

• 📧 联系我们：[email protected] — 用于咨询或技术支持。

• 🌍 办公地点：索菲亚 | 伊斯坦布尔 | 伦敦

关注我们：

• 🐙 GitHub

• 😊 Hugging Face

• 💡 Civitai

我们的网站：

• 🌐 FFusion.ai

• 🌌 FFAI.eu

• 🔗 1e-2.com