Instructions to use DeadSoulInside/aworldofhate_adarkerworld with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use DeadSoulInside/aworldofhate_adarkerworld with PEFT:
Task type is invalid.
- Notebooks
- Google Colab
- Kaggle
aworldofhate_adarkerworld
A LoRA for ACE-Step 1.5, trained on dark, instrumental electronic music โ a fusion of a personal solo archive spanning nearly three decades with newer productions.
Preview Track:
Model Details
Model Description
This LoRA biases ACE-Step 1.5 toward a dark-electronic instrumental sound. It was trained on 35 original tracks: the bulk written between 1997 and 2008, with additional material from 2025โ2026 blended in. The name combines the artist's earlier solo project ("A World of Hate") with the newer studio works used with the hybrid human/AI project name ("A Darker World") the works used were all human written and produced using FL Studio 2.0 and up with hardware synths (Roland Juno, Waldorf Microwave for exmaples) and midi controllers.
- Developed by: DeadSoulInside
- Shared by: DeadSoulInside
- Model type: LoRA adapter (PEFT) for music generation
- Genre / Style: Dark electronic, instrumental
- License: CC-BY 4.0 (attribution required; commercial use permitted)
- Finetuned from model: ACE-Step 1.5 (
acestep-v15-turbo), MIT licensed
Model Sources
Uses
Direct Use
Load as a LoRA adapter alongside ACE-Step 1.5 to generate instrumental dark-electronic tracks, using the custom tag dark_style to invoke the trained style.
Out-of-Scope Use
Not trained for vocals/lyrics generation, and may not generalize well to genres far outside dark electronic (e.g. orchestral, acoustic, hip-hop).
How to Get Started with the Model
ACE-Step loads LoRAs by folder name, so place the adapter in a folder named aworldofhate_adarkerworld (containing adapter_model.safetensors and its config) inside ACE-Step's LoRA directory, then reference the dark_style tag in your prompt to invoke the trained style. Since this was trained on pure instrumental works, you will need to set the strength between .4-.5 if you are trying to use lyrics.
Training Details
Training Data
35 original, instrumental dark-electronic tracks written by the author. The bulk of the material is from 1997โ2008 (the artist's early solo era), with additional tracks added from 2025โ2026.
Training Procedure
- Custom tag:
dark_style - Tag/Style weighting: 50% tag / 50% style
- Tags Used: dark-electro, aggrotech, Industrial, EBM, darkwave, synthwave, dark-pop, dark-techno, dark-ambient, dark-ebm
- Epochs: 1000
Speeds, Sizes, Times
- Total training time: 9h 52m
Bias, Risks, and Limitations
Trained on a narrow, personal catalog within a single genre โ outputs will skew strongly toward dark/atmospheric electronic textures and may not perform well for brighter or more conventional styles.
License
This LoRA is released under CC-BY 4.0. You're free to use, remix, and redistribute (including commercially), provided you credit DeadSoulInside. Note this is separate from the ACE-Step 1.5 base model, which is MIT licensed.
Citation
BibTeX:
@misc{deadsoulinside_aworldofhate_adarkerworld,
author = {DeadSoulInside},
title = {aworldofhate_adarkerworld: A Dark Electronic LoRA for ACE-Step 1.5},
year = {2026},
publisher = {Hugging Face},
url = {https://huggingface.co/DeadSoulInside/aworldofhate_adarkerworld}
}
APA:
DeadSoulInside. (2026). aworldofhate_adarkerworld: A Dark Electronic LoRA for ACE-Step 1.5. Hugging Face. https://huggingface.co/DeadSoulInside/aworldofhate_adarkerworld
Model Card Authors
DeadSoulInside
Model Card Contact
https://huggingface.co/DeadSoulInside
Framework versions
- PEFT 0.18.1
- Downloads last month
- -