I quantized the latest versions of Lodestones Chroma for use in lower VRAM machines. Enjoy. For full versions <a target="_blank" rel="ugc" href="https://huggingface.co/lodestones/Chroma/tree/main">https://huggingface.co/lodestones/Chroma/tree/main</a>For me it works best with Euler/Beta or DPM++2M/ SGM Uniform16-20 steps or Restart/SGM Uniform 6-10 steps. CFG 2 in Forge and 2.7 in Comfy with res_2s/bong_tangent. Download and use the Hyper_LowStep (str.1) or Chroma2Schnell (str. 0.125) LoRA from <a target="_blank" rel="ugc" href="https://huggingface.co/silveroxides/Chroma-LoRA-Experiments/tree/main">https://huggingface.co/silveroxides/Chroma-LoRA-Experiments/tree/main</a> for fewer steps and faster generation times.

chroma-unlocked-Q4_K_S

<ol><li>'Annealing' = letting the V50 model train on the entire database of training images ( 5 million or maybe the full 20 million ) one final time at a very low learning rate</li></ol>

Files with SHA256: e2df4f83e9ab3457742275807023d96bed9d6d2b1a168bad610a9db9cd262ad9