music3lab / checkpoints /README.md
coolpoodle's picture
code and training scripts
90884df verified
|
Raw
History Blame Contribute Delete
1.57 kB

Checkpoints (pointers only β€” no weights in Git)

This directory intentionally contains no model weights. Trained adapters/encoders are distributed separately (Hugging Face) and are pulled in by you. See ../MODEL_CARD.md for what each one is and its gate result.

⚠️ Before any of these are uploaded/redistributed: confirm the MiniMax-Music3 license permits derivative-weight redistribution (see ../THIRD_PARTY.md). These checkpoints were trained through the frozen Music3 decoder / on captured Music3 state and may be governed by it.

Expected layout once downloaded

Place downloaded checkpoints here (git-ignored) or point configs at them:

checkpoints/
β”œβ”€β”€ flow-encoder/encoder.safetensors                 # base continuous encoder (~1.34 ms)
β”œβ”€β”€ external-finetune/encoder.safetensors            # rejected real-music specialist
β”œβ”€β”€ masked-flow-inpaint/adapter.safetensors          # captured-condition inpaint pilot
β”œβ”€β”€ native-state-stage0-v3/native_state_stage0.safetensors
β”œβ”€β”€ native-state-distill/native_state_distilled.safetensors
└── ...                                              # failed-experiment adapters (see MODEL_CARD)

Downloading (once the HF repo exists and license is confirmed)

hf download <your-org>/music3lab-adapters --local-dir ./checkpoints

<your-org>/music3lab-adapters is a placeholder β€” replace with the real Hugging Face repo id after the license review. Until then, this directory is a pointer only.