krea/Krea-2-Turbo
Full fine-tune vs krea/Krea-2-Raw. Fine-tune: 100% of the weights changed, across 28 of 28 layers.
Tensors changed: Norms 129/129, other 36/36, MLP 105/105, Attention 160/160
207 derivatives checked
- realrebelai/KREA-2_GGUFs: Full fine-tune
- vantagewithai/Krea-2-Turbo-GGUF: Quantization
- Omnico/Krea2_turbo_diff_loras: Different architecture
- Quiho/Krea2_NSFW_Uncensored_Complete: Quantization
- DeepBeepMeep/krea-2: Full fine-tune
- RudySen/Krea2-realism-V2: Adapter
- lvladikov/Krea2-Turbo-Distill-4step-LoRA: Different architecture
- Beinsezii/Krea-2-Turbo-Projector-Scale-LoRA-Diffusers: Adapter
- wikeeyang/Krea2-Turbo-HD-V1: Different architecture
- mflux-community/krea-2-turbo-mflux-q8: Quantization
- ostris/krea2_turbo_training_adapter: Adapter
- thedeoxen/Krea-2-pose-controlnet: Adapter
- Abiray/Krea-2-Turbo-GGUF: Quantization
- uzumix/krea2filterbypass3.safetensors: Different architecture
- Noctaluna/Krea2-Noct-Uncensored: Quantization
- yijunwang2/krea2-anypaint: Adapter
- F16/krea2-turbo-sda: Adapter
- ostris/krea2_turbo_style_reference: Adapter
- byteshape/Krea-2-Turbo-GGUF: Quantization
- uzumix/realism_engine_krea2_v2: Adapter
- lvladikov/Krea2-Turbo-Distill-2step-LoRA: Different architecture
- Raxephion/Krea2-Ronin-V1: LoRA-style
- mflux-community/krea-2-turbo-mflux-q4: Quantization
- QuantFunc/Krea-2-QuantFunc-4bit: Quantization
- AX1Y2JP/Krea-2-Turbo-INT8-ConvRot: Quantization
- uzumix/krea2_nsfw: Adapter
- yijunwang2/krea2-reid: Adapter
- RudySen/Krea2-realism-V1: Adapter
- uzumix/Realism_Engine_Ideogram_4_Krea_2_v3.1: Adapter
- uzumix/krea2_realisticSnapshot: Adapter
10 other repos hold byte-for-byte identical weight files.