Guardado en:
| Autores principales: | , , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | https://arxiv.org/abs/2603.20433 |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866918400427556864 |
|---|---|
| author | Piao, Yen-Ting Liao, Jay Chiehen Chien, Wei-Tang Ogimoto, Toshiki Chen, Shang-Tse Chen, Yun-Nung Lee, Chun-Yi Lo, Shao-Yuan |
| author_facet | Piao, Yen-Ting Liao, Jay Chiehen Chien, Wei-Tang Ogimoto, Toshiki Chen, Shang-Tse Chen, Yun-Nung Lee, Chun-Yi Lo, Shao-Yuan |
| contents | While Large Audio-Language Models (LALMs) have been shown to exhibit degraded instruction-following capabilities, their ability to infer task patterns from in-context examples under audio conditioning remains unstudied. To address this gap, we present ALICE, a three-stage framework that progressively reduces textual guidance to systematically evaluate LALMs' in-context learning ability under audio conditioning. Evaluating six LALMs across four audio understanding tasks under two output constraint categories, we uncover a consistent asymmetry across all stages and LALMs: in-context demonstrations reliably improve format compliance but fail to improve, and often degrade, the core task performance. This suggests that LALMs can glean surface-level formatting patterns from demonstrations but may struggle to leverage cross-modal semantic grounding to reliably infer task objectives from audio-conditioned examples, highlighting potential limitations in current cross-modal integration. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_20433 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | ALICE: A Multifaceted Evaluation Framework of Large Audio-Language Models' In-Context Learning Ability Piao, Yen-Ting Liao, Jay Chiehen Chien, Wei-Tang Ogimoto, Toshiki Chen, Shang-Tse Chen, Yun-Nung Lee, Chun-Yi Lo, Shao-Yuan Sound Artificial Intelligence Computation and Language Audio and Speech Processing While Large Audio-Language Models (LALMs) have been shown to exhibit degraded instruction-following capabilities, their ability to infer task patterns from in-context examples under audio conditioning remains unstudied. To address this gap, we present ALICE, a three-stage framework that progressively reduces textual guidance to systematically evaluate LALMs' in-context learning ability under audio conditioning. Evaluating six LALMs across four audio understanding tasks under two output constraint categories, we uncover a consistent asymmetry across all stages and LALMs: in-context demonstrations reliably improve format compliance but fail to improve, and often degrade, the core task performance. This suggests that LALMs can glean surface-level formatting patterns from demonstrations but may struggle to leverage cross-modal semantic grounding to reliably infer task objectives from audio-conditioned examples, highlighting potential limitations in current cross-modal integration. |
| title | ALICE: A Multifaceted Evaluation Framework of Large Audio-Language Models' In-Context Learning Ability |
| topic | Sound Artificial Intelligence Computation and Language Audio and Speech Processing |
| url | https://arxiv.org/abs/2603.20433 |