Metraipolis › archive › dc_cd8a7113ca35

X-AuT: Progressive Audio-Encoder Compression for Speech LLMs with Cross-Scale Distillation — arXiv:2609.11412

Source: arXiv (en)
Original: https://arxiv.org/abs/2609.11412
License (as archived): CC0-1.0
Archived: 2026-10-16T05:27:02.188Z
SHA-256 of the archived text: a65631797ee716dd3a56331fe1fe351d4de9ab3843021443d1646665986d66d9
Stable URL of this page: https://metraipolis.com/documento/dc_cd8a7113ca35

How to cite

X-AuT: Progressive Audio-Encoder Compression for Speech LLMs with Cross-Scale Distillation — arXiv:2609.11412 — arXiv (en), https://arxiv.org/abs/2609.11412 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_cd8a7113ca35 (Metraipolis archive dc_cd8a7113ca35, sha256 a65631797ee716dd3a56331fe1fe351d4de9ab3843021443d1646665986d66d9)

BibTeX

@misc{dccd8a7113ca35,
  title = {X-AuT: Progressive Audio-Encoder Compression for Speech LLMs with Cross-Scale Distillation — arXiv:2609.11412},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.11412},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_cd8a7113ca35; sha256 a65631797ee716dd3a56331fe1fe351d4de9ab3843021443d1646665986d66d9},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.11412. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Reducing audio-encoder depth lowers the inference cost of speech large language models, but removing complete blocks perturbs the embeddings consumed by the decoder and can cause deletion and premature end-of-sequence errors. We introduce X-AuT, a progressive framework that selects layer combinations through short behavioral probes and restores the pruned model through representation alignment, cross-scale distillation, scheduled student-policy supervision, and LoRA finetuning. The language-model backbone remains frozen, while attention LoRA adapters and the tied output embedding adapt during distillation. Training uses the highest-agreement tier from a transcript-consistency pipeline, followed by source reweighting during finetuning. On ten public Chinese--English benchmarks, compressing Qwen3-ASR-0.6B from 18 to 16 audio-encoder layers reduces macro-average error from 5.61% to 5.27%. The 14-layer model reaches 5.75% with 20.7% fewer audio-tower parameters. Under the matched recipe, the 1.7B teacher yields 5.55% mean error, compared with 8.45% for self-distillation, and progressive 18$\rightarrow$14 pruning outperforms direct pruning (5.75% vs. 6.73%). These single-run results establish two practical operating points and show that the accuracy effects vary across benchmarks. Project website: this https URL

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-11 in cs.AI).
Categorie arXiv: Sound (cs.SD) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.11412 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-16T05:17:52.163Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-11?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).