The auditable paper sequence is pinned at revision a9324d03830d58fca87eac010162982efa106b22. It contains four fp16 active-interface checkpoints:

Each directory contains TASU, the acoustic projector, and the instruction router: 84,655,552 parameters, or 5.4839% of the main LLM's 1,543,714,304 independent parameters. The tied token embedding and language-model head are counted once rather than as separate state-dict aliases. The accompanying verification records exact tensor equality for Qwen2.5-1.5B-Instruct, SenseVoice, and ArCap across the sequence.

The claim is deliberately narrower than “training without audio”: real audio is used for grounding before the two zero-audio post-alignment stages. Current evidence is strongest for semantic and speech-paralinguistic transfer; general non-speech signal/event transfer still needs a matched causal experiment.

AIR-Bench metrics in the manifest distinguish historical valid-only accuracy from coverage-aware end-to-end accuracy; unparsable responses are incorrect in the latter.

See the release manifest, frozen-state verification, and GitHub documentation.

The repository root also stores the original full non-main-LLM export for phase2_3_chat/model-5000.pt. It includes the frozen SenseVoice and ArCap experts as well as the Phase 2.3 interfaces:

These frozen expert tensors can reconstruct every checkpoint in the paper timeline because the exact comparison proves that they never change. Load the full-branch state first, then overwrite the three active interfaces with the selected timeline directory. Qwen2.5-1.5B-Instruct is loaded separately from its upstream release.

For an unwrapped PyTorch model, strip the training-time module. prefix:

The historical Phase 2.3 checkpoint reports ASR test-clean/test-other WER 3.50%/7.22%, AIR-Bench Chat 2.365/10 with a Qwen3.5-27B judge, and VoiceBench sd-qa 21.8%. Metrics, failure cases, and claim boundaries are documented in the GitHub repository rather than inferred from a single aggregate score.