Trained for 12 epochs and 1000 steps.

Trained with datasets ['text-embeds', 'suno-reggae-audio']
Learning rate 5e-05 and batch size 1.
Training objective: flow matching.
Distillation method: h3_drift.
Distillation config: {"h3_drift": {"audio_weight": 1.0, "balance": "token", "loss_weight": 0.5, "sft_loss_weight": 1.0, "video_weight": 1.0}}
Base model: MiniMaxAI/MiniMax-H3
VAE: MiniMaxAI/MiniMax-H3

pytorch_lora_weights.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:3d2874fae092ed6029dd3c43e075ce90a67fb98d78083974cb9e158f73038108
+oid sha256:64b7a7fb0298477d25068dbcdf8bc387242da0997a42f24423ba395fb0635425
 size 670770080