Papers for

audio software teams

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

Gemma 4 edge models fine-tuned for better MATLAB signal code on laptops

Fine-Tuning Gemma 4 Edge Models for MATLAB Signal-Processing Code with SFT and QLoRA on a Laptop iGPU

Abstract: Small language models that run on a laptop are attractive for writing signal-processing code offline, but they make systematic errors on digital signal processing (DSP) conventions. We apply supervised fine-tuning (SFT) to the Gemma 4 edge models E2B and E4B with 66 checked examples covering 11 signal-processing MATLAB subtasks, using low-rank adaptation (LoRA) and quantized LoRA (QLoRA) on the integrated GPU (iGPU) of a laptop, and deploy the 8-bit integer (int8) models with the adapters through OpenVINO. On held-out variants scored by deterministic checkers, the pass rate rises from 18/66 to 37/66 (LoRA) and 37/66 (QLoRA) for E2B and from 32/66 to 47/66 for E4B (QLoRA). Reading all 159 failing programs shows that the adapters remove most errors of convention, such as the sign of a lag, units and array shapes, but few errors in building signals and noise. Applied at run time, the adapter costs 17-19% of the generation speed.

Tue 17 NovMachine LearningArtificial IntelligenceSoftware Engineering
The gist
Small language models that run locally on laptops can help write MATLAB code for signal processing, but they often make specific technical mistakes. The authors improved two such models by teaching them with examples of correct signal-processing tasks using a special fine-tuning method on a laptop’s integrated GPU. After this training, the models made fewer errors in following digital signal processing rules while generating code. The improvements come at a small cost to the speed of code generation.
Open → 2611.23205v1