minimaxh3-suno-reggae-rank128

bghira/minimaxh3-suno-reggae-rank128

This is a PEFT LoRA derived from MiniMaxAI/MiniMax-H3.

The main validation prompt used during training was:

<STYLE>
roots reggae, warm analog dub production, laid-back one-drop drum groove, deep round bassline, skanking guitar upstrokes, organ bubble, spacious mix with tape echo and spring reverb, soulful male lead vocal, mid-tempo

<LYRICS>
[verse]
Morning sun a rise pon di mountain top
River run easy and di worries stop
We carry good vibes from di country road
Every likkle burden turn a lighter load

[chorus]
Lift up yuh heart now, sing it loud and clear
One love a di message and di roots right here
Drum and di bass dem a guide di way
Sweet reggae music till di break of day

Validation settings

• CFG: 1.0
• CFG Rescale: 0.0
• Steps: 40
• Sampler: MiniMaxH3Scheduler
• Seed: 42
• Resolution: 256

Note: The validation settings are not necessarily the same as the training settings.

You can find some example images and videos in the following gallery:

<Gallery />

The text encoder was not trained. You may reuse the base model text encoder for inference.

Training settings

• Training epochs: 12

• Training steps: 1000

• Learning rate: 5e-05

  • Learning rate schedule: constant_with_warmup
  • Warmup steps: 500

• Max grad norm: 0.5

• Effective batch size: 8

  • Micro-batch size: 1
  • Gradient accumulation steps: 1
  • Number of GPUs: 8

• Gradient checkpointing: True

• Prediction type: flow_matching[]

• Optimizer: adamw_bf16 (config=weight_decay=0.0,eps=1e-8)

• Trainable parameter precision: Pure BF16

• Base model precision: no_change

• Caption dropout probability: 0.0%

• LoRA Rank: 128

• LoRA Alpha: 128.0

• LoRA Dropout: 0.0

• LoRA initialisation style: default

• LoRA mode: Standard

Datasets

suno-reggae-audio

• Repeats: 0
• Total number of images: 123
• Total number of aspect buckets: 15
• Resolution: 480 px
• Cropped: False
• Crop style: None
• Crop aspect: None
• Used for regularisation data: No

Inference

import torch
from diffusers import DiffusionPipeline

model_id = 'MiniMaxAI/MiniMax-H3'
adapter_id = 'bghira/minimaxh3-suno-reggae-rank128'
pipeline = DiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.bfloat16) # loading directly in bf16
pipeline.load_lora_weights(adapter_id)

prompt = "<STYLE>
roots reggae, warm analog dub production, laid-back one-drop drum groove, deep round bassline, skanking guitar upstrokes, organ bubble, spacious mix with tape echo and spring reverb, soulful male lead vocal, mid-tempo

<LYRICS>
[verse]
Morning sun a rise pon di mountain top
River run easy and di worries stop
We carry good vibes from di country road
Every likkle burden turn a lighter load

[chorus]
Lift up yuh heart now, sing it loud and clear
One love a di message and di roots right here
Drum and di bass dem a guide di way
Sweet reggae music till di break of day"
negative_prompt = ''

## 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=40,
    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=1.0,
).images[0]

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