Using the dataset and script from <a target="_blank" rel="ugc" href="https://civarchive.com/user/seruva19">https://civarchive.com/user/seruva19</a> (Please give him all the credit)In this Lora, <a target="_blank" rel="ugc" href="https://civarchive.com/models/1753118/anime-denser-wan21-t2v-14b">https://civarchive.com/models/1753118/anime-denser-wan21-t2v-14b</a>This is Hidream version of that same concept lora -3 strength less density +3 strength more density.<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/e3c02106-f214-4405-83af-f2ab02a546a1/width=525/e3c02106-f214-4405-83af-f2ab02a546a1.jpeg" /><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/fddf05a7-df35-4a88-9060-254f47aedfb2/width=525/fddf05a7-df35-4a88-9060-254f47aedfb2.jpeg" />I used ai-toolkit to train both the positive and negative lora's and used the same difference_lora.py script to make the final lora.* The only change I had to make for the difference_lora.py script was change the ".lora_down.weight" references into ".lora_A.weight" references (different names but same concepts).

Files with SHA256: f830698ce3468bc34c4077ec9ac95c287dbf91f782c7db773053719df303d47a