SPIN Unprocessed October 22, 2026 ai_technology research
How Far Can the Gemma 4 Speech Encoder Go Without Changing Its Weights? Saudi and Gulf Arabic Speech Recognition with a Frozen and an Adapter-Augmented Encoder
View original on arxiv.orgOverview
arXiv:2610.28673v1 Announce Type: new Abstract: We study the 305M-parameter audio encoder shared by Gemma 4 and EmbeddingGemma 2 (identical except for its final projection) for Saudi and Gulf Arabic speech recognition, without changing any of its weights. A frozen system trains only a 22.6M-parameter CTC reader over a mix of encoder layers; an adapted system adds 3.1M low-rank adapter parameters, a 41.5M-parameter hybrid CTC/attention reader and joint decoding with a sub-word language model. Aga
SpinGraph analysis pending — check back after processing.
Ask AI about this story
Opens with the SpinGraph .md URL and structured context — one click, prompt included.
More from arXiv Computation and Language
View all →- A Large-Scale Evaluation of Role-Enhanced Zero-Shot Chain-of-Thought Prompting
- Our Own Jev: Typed Decisions in One Forward Pass, from an In-House Judge to Open LoRA Models
- Target Regeneration Shrinks the Gains of Draft-Training Objectives in Speculative Decoding
- Language Impairment as a Lens onto Large Language Models
- Scan or Index? large Evidence Coverage, Answer Quality, and Cost in Long-Document RAG
- Guilt by Association: Semantic Leakage Drives Associative Gender Bias
Markdown (.md) · JSON-LD schema (.json) · Machine-readable for AI & GEO