This lora was made using 17 drum'n'bass tracks composed by myself.

It's an attempt to fix the model's inability to produce fast breakbeats. While it doesn't work 100% of the time, the result usually at least vaguely resembles d'n'b

  • The "drum solo" version was additionally trained on looping drum'n'bass breaks from the same songs. Use "drum solo" to activate. It can also be used to generate full songs, but they tend to have overwhelmingly loud drums.

  • V2 is made with rank 128 and gradient accumulation 4.

I think that's the limit of what I can do with my limited resources. Sadly, ACEStep really fights backs against any attempts to teach it this style and I'd need a much bigger dataset and more compute to properly hammer it in.

Example output

V1: V2:

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for worstplayer/acestep15_drumnbass_lora

Finetuned
(2)
this model