bdsqlsz

/

qinglong_DetailedEyes_Z-Image

like 22

Just "lora key not loaded:" spam and doesn't enhance eyes

I haven't encountered this issue myself. Perhaps you can update comfyui and then use a lora loading node without clip.

using latest comfyui version (git pull done before trying) and got same error. Which lora node do you use ?

[ZImageTextEncoder] Loaded 4 templates
lora key not loaded: lora_unet_context_refiner_0_attention_to_k.alpha
lora key not loaded: lora_unet_context_refiner_0_attention_to_k.lora_down.weight
lora key not loaded: lora_unet_context_refiner_0_attention_to_k.lora_up.weight
lora key not loaded: lora_unet_context_refiner_0_attention_to_out_0.alpha
lora key not loaded: lora_unet_context_refiner_0_attention_to_out_0.lora_down.weight
lora key not loaded: lora_unet_context_refiner_0_attention_to_out_0.lora_up.weight
lora key not loaded: lora_unet_context_refiner_0_attention_to_q.alpha
lora key not loaded: lora_unet_context_refiner_0_attention_to_q.lora_down.weight
lora key not loaded: lora_unet_context_refiner_0_attention_to_q.lora_up.weight
lora key not loaded: lora_unet_context_refiner_0_attention_to_v.alpha

Thanks

using latest comfyui version (git pull done before trying) and got same error. Which lora node do you use ?

[ZImageTextEncoder] Loaded 4 templates
lora key not loaded: lora_unet_context_refiner_0_attention_to_k.alpha
lora key not loaded: lora_unet_context_refiner_0_attention_to_k.lora_down.weight
lora key not loaded: lora_unet_context_refiner_0_attention_to_k.lora_up.weight
lora key not loaded: lora_unet_context_refiner_0_attention_to_out_0.alpha
lora key not loaded: lora_unet_context_refiner_0_attention_to_out_0.lora_down.weight
lora key not loaded: lora_unet_context_refiner_0_attention_to_out_0.lora_up.weight
lora key not loaded: lora_unet_context_refiner_0_attention_to_q.alpha
lora key not loaded: lora_unet_context_refiner_0_attention_to_q.lora_down.weight
lora key not loaded: lora_unet_context_refiner_0_attention_to_q.lora_up.weight
lora key not loaded: lora_unet_context_refiner_0_attention_to_v.alpha

Thanks

I guess this is because the key for the FP8 gguf version has changed; I'm using the official original bf16 key structure.

how did you train the lora???? VERY interested in the process!

https://huggingface.co/Kijai/Z-Image_comfy_fp8_scaled/blob/main/qinglong_detailedeye_lora_z-image_comfy.safetensors

using latest comfyui version (git pull done before trying) and got same error. Which lora node do you use ?

[ZImageTextEncoder] Loaded 4 templates
lora key not loaded: lora_unet_context_refiner_0_attention_to_k.alpha
lora key not loaded: lora_unet_context_refiner_0_attention_to_k.lora_down.weight
lora key not loaded: lora_unet_context_refiner_0_attention_to_k.lora_up.weight
lora key not loaded: lora_unet_context_refiner_0_attention_to_out_0.alpha
lora key not loaded: lora_unet_context_refiner_0_attention_to_out_0.lora_down.weight
lora key not loaded: lora_unet_context_refiner_0_attention_to_out_0.lora_up.weight
lora key not loaded: lora_unet_context_refiner_0_attention_to_q.alpha
lora key not loaded: lora_unet_context_refiner_0_attention_to_q.lora_down.weight
lora key not loaded: lora_unet_context_refiner_0_attention_to_q.lora_up.weight
lora key not loaded: lora_unet_context_refiner_0_attention_to_v.alpha

Thanks

I guess this is because the key for the FP8 gguf version has changed; I'm using the official original bf16 key structure.

i tried using also the bf16 model in my wf to load your lora.
still this error.

using latest comfyui version (git pull done before trying) and got same error. Which lora node do you use ?

[ZImageTextEncoder] Loaded 4 templates
lora key not loaded: lora_unet_context_refiner_0_attention_to_k.alpha
lora key not loaded: lora_unet_context_refiner_0_attention_to_k.lora_down.weight
lora key not loaded: lora_unet_context_refiner_0_attention_to_k.lora_up.weight
lora key not loaded: lora_unet_context_refiner_0_attention_to_out_0.alpha
lora key not loaded: lora_unet_context_refiner_0_attention_to_out_0.lora_down.weight
lora key not loaded: lora_unet_context_refiner_0_attention_to_out_0.lora_up.weight
lora key not loaded: lora_unet_context_refiner_0_attention_to_q.alpha
lora key not loaded: lora_unet_context_refiner_0_attention_to_q.lora_down.weight
lora key not loaded: lora_unet_context_refiner_0_attention_to_q.lora_up.weight
lora key not loaded: lora_unet_context_refiner_0_attention_to_v.alpha

Thanks

I guess this is because the key for the FP8 gguf version has changed; I'm using the official original bf16 key structure.

i tried using also the bf16 model in my wf to load your lora.
still this error.

It looks like comfyui doesn't support the lora format. After the update, you can use this converted version
https://huggingface.co/Kijai/Z-Image_comfy_fp8_scaled/blob/main/qinglong_detailedeye_lora_z-image_comfy.safetensors

Oops, it has been removed
https://huggingface.co/Kijai/Z-Image_comfy_fp8_scaled/commit/952c72
Why?
Download link: https://huggingface.co/Kijai/Z-Image_comfy_fp8_scaled/blob/9719d6/qinglong_detailedeye_lora_z-image_comfy.safetensors
New, official: https://huggingface.co/bdsqlsz/qinglong_DetailedEyes_Z-Image/blob/main/qinglong_detailedeye_z-imageV2(comfy).safetensors