Credit

Author : spadira272

Origin Model 1: Kleinfetti260322

Origin Model 2: Kleinfetti260315

Look at the civitai models description for prompt infomation and preview images for workflow.

Overall a really cool finetune, caught my eye and want more people to use it.

I dont have time to test all quants , fp8 sacled versions might not be the best it could be. I based them on comfyUI's docs. Feedback is appreciated as always.

Models

  • version 260315

Kleinfetti260315_4b_base_nvfp4.safetensors

Kleinfetti260315_4b_base_nvfp4_txt_attn_BF16.safetensors txt_attn layers kept at bf16, might help with prompt following

kleinfetti_260315_fp8_scaled_e4m3fn_comfyui.safetensors

kleinfetti_260315_fp8_scaled_e5m2_comfyui.safetensors

  • version 260322

Kleinfetti260322_4b_base_nvfp4.safetensors

Kleinfetti260322_4b_base_nvfp4_txt_attn_BF16.safetensors txt_attn layers kept at bf16, might help with prompt following

kleinfetti_260322_fp8_scaled_e4m3fn_comfyui.safetensors

kleinfetti_260322_fp8_scaled_e5m2_comfyui.safetensors

Downloads last month
-
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for ApacheOne/FluxKleinfetti-4b_base-nvfp4_mixed

Quantized
(7)
this model

Collection including ApacheOne/FluxKleinfetti-4b_base-nvfp4_mixed