Views
No views yet
krea/Krea-2-LoRA-darkbrush
laid out for direct loading by the vLLM-Omni diffusion
LoRA manager.darkbrush.safetensors with no adapter_config.json. vLLM-Omni's
diffusion LoRA loader (via vLLM's PEFTHelper / LoRAModel.from_local_checkpoint) requires a directory
containing the exact PEFT filenames adapter_model.safetensors + adapter_config.json, with rank/alpha
read from the config. This repo provides exactly that; the tensors are unchanged apart from the standard
base_model.model. key prefix.krea/Krea-2-Raw, krea/Krea-2-Turbo)img_in, time_mod_proj, text-fusion + transformer-block attention/FF projections, final_layer.linear)r = 32, lora_alpha = 32 (scaling alpha / r = 1.0)1from vllm_omni.entrypoints.omni import Omni
2from vllm_omni.lora.request import LoRARequest
3from vllm_omni.lora.utils import stable_lora_int_id
4from vllm_omni.inputs.data import OmniDiffusionSamplingParams
5
6REPO = "NagaSaiAbhinay/Krea-2-vllm-darkbrush-LoRA"
7omni = Omni(model="krea/Krea-2-Turbo", lora_path=REPO)
8sp = OmniDiffusionSamplingParams(height=1024, width=1024, num_inference_steps=8, guidance_scale=0.0, seed=42)
9sp.lora_request = LoRARequest("darkbrush", stable_lora_int_id(REPO), REPO)
10sp.lora_scale = 1.0/v1/images/generations)1curl -X POST http://localhost:8091/v1/images/generations \
2 -H "Content-Type: application/json" \
3 -d '{"prompt":"a fox in the snow, photorealistic","size":"1024x1024",
4 "seed":42,"num_inference_steps":8,"guidance_scale":0.0,
5 "lora":{"name":"darkbrush","path":"NagaSaiAbhinay/Krea-2-vllm-darkbrush-LoRA","scale":1.0}}'krea/Krea-2-LoRA-darkbrush; all credit for the trained weights to Krea.