Training epochs: 416
Training steps: 10000
Learning rate: 1.0
- Learning rate schedule: constant
- Warmup steps: 0
Max grad value: 1.0
Effective batch size: 1
- Micro-batch size: 1
- Gradient accumulation steps: 1
- Number of GPUs: 1
Gradient checkpointing: True
Prediction type: autoregressive_next_token
Optimizer: prodigy
Trainable parameter precision: Pure BF16
Base model precision: no_change
Caption dropout probability: 0.0%
LoRA Rank: 64
LoRA Alpha: None
LoRA Dropout: 0.1
LoRA initialisation style: default
LoRA mode: Standard

Datasets

soad-mm3-xm-20260822-10k-prodigy-nocapdrop-audio

Repeats: 0
Total number of audio files: 24
Duration buckets: audio (24)
Used for regularisation data: No

Inference

python

import torch
from diffusers import DiffusionPipeline

model_id = 'MiniMaxAI/MiniMax-Music3'
adapter_id = 'RareConcepts/soad-mm3-xm-20260822-10k-prodigy-nocapdrop'
pipeline = DiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.bfloat16) # loading directly in bf16
pipeline.load_lora_weights(adapter_id)

prompt = "An astronaut is riding a horse through the jungles of Thailand."
negative_prompt = 'blurry, cropped, ugly'

## Optional: quantise the model to save on vram.
## Note: The model was not quantised during training, so it is not necessary to quantise it during inference time.
#from optimum.quanto import quantize, freeze, qint8
#quantize(pipeline.transformer, weights=qint8)
#freeze(pipeline.transformer)
    
pipeline.to('cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu') # the pipeline is already in its target precision level
model_output = pipeline(
    prompt=prompt,
    negative_prompt=negative_prompt,
    num_inference_steps=30,
    generator=torch.Generator(device='cuda' if torch.cuda.is_available() else 'mps' if torch.backends.mps.is_available() else 'cpu').manual_seed(42),
    width=256,
    height=256,
    guidance_scale=7.5,
).images[0]

model_output.save("output.png", format="PNG")

Controlled checkpoint and strength tournament

This gallery compares Explorative Modeling, Prodigy, no caption dropout checkpoints under four caption conditions. Every render uses the same unseen lyrics and random stream, 40 seconds maximum duration, 30 flow steps, flow CFG 1.7, and seed 20260822. Within each batch, only LoRA strength changes.

The four caption controls separate trigger response from descriptive style learning and unrelated-style leakage:

Off-genre, no trigger: Alternative pop-jazz, 102 BPM, brushed drums, upright bass, clean electric piano, airy male vocals, smooth verse-chorus songwriting, restrained dynamics, polished intimate mix.
Trigger only: system of a down, soad style music
Descriptors, no trigger: Lilting hypnotic arpeggios, a gentle melodic mid-tempo groove with layered harmonies, sudden explosive distorted riffs, unstable meter changes, theatrical male vocals, tightly syncopated bass and drums, abrupt shifts between fragile verses and frantic heavy choruses.
Trigger + descriptors: system of a down, soad style music, lilting hypnotic arpeggios, a gentle melodic mid-tempo groove with layered harmonies, sudden explosive distorted riffs, unstable meter changes, theatrical male vocals, tightly syncopated bass and drums, abrupt shifts between fragile verses and frantic heavy choruses.

Machine-readable manifest

checkpoint-1000

strength	Off-genre, no trigger	Trigger only	Descriptors, no trigger	Trigger + descriptors
0.25
0.50
0.75
1.00

checkpoint-2500

strength	Off-genre, no trigger	Trigger only	Descriptors, no trigger	Trigger + descriptors
0.25
0.50
0.75
1.00

checkpoint-5000

strength	Off-genre, no trigger	Trigger only	Descriptors, no trigger	Trigger + descriptors
0.25
0.50
0.75
1.00

checkpoint-7500

strength	Off-genre, no trigger	Trigger only	Descriptors, no trigger	Trigger + descriptors
0.25
0.50
0.75
1.00

checkpoint-10000

strength	Off-genre, no trigger	Trigger only	Descriptors, no trigger	Trigger + descriptors
0.25
0.50
0.75
1.00

Unseen lyrics used for every render

text

[Intro]
Count the teeth in the traffic light
Paint the silence ultraviolet

[Verse 1]
Paper crowns on the factory floor
Everybody wants a little more
Wind-up prophets in a checkout line
Selling borrowed seconds back as time

[Pre-Chorus]
Turn the dial, divide the sky
Ask the wires to tell us why

[Chorus]
Wake up the walls, let the ceiling shake
We built a feast from a plastic lake
Hands in the air while the engines stall
Who signed the order to swallow it all?

[Verse 2]
Velvet sirens in the avenue
Paint every warning a pleasant blue
I fed my shadow to the voting machine
It came back polished, quiet, and clean

[Bridge]
Softly, softly, count to four
Kick down the arithmetic door
One for the hunger, two for the show
Three for the ones who already know

[Final Chorus]
Wake up the walls, let the ceiling shake
We built a feast from a plastic lake
Hands in the air while the engines stall
Who signed the order to swallow it all?

[Outro]
Count the teeth
Kill the light
Carry the silence
Into the night

NextLat predictor and XM route-embedding sidecars are training-only and are intentionally not used during inference.