Gemma 4 Multimodal Fine-Tuner vs Structured Output Benchmark

Gemma 4 Multimodal Fine-Tuner is an open-source toolkit that lets developers fine-tune Google's Gemma 4 and 3n models across all three modalities — text, images, and audio — using only Apple Silicon hardware. It runs natively on PyTorch with Metal Performance Shaders (MPS), bypassing the NVIDIA requirement that has historically blocked Mac users from serious local fine-tuning work. The toolkit handles the full training pipeline including dataset prep, LoRA adapters, and multi-modal data collation. It ships with working example notebooks, a validation suite, and clean abstractions that don't require deep familiarity with the underlying MPS stack. Apple Silicon's unified memory architecture actually helps here — large multimodal batches fit in memory that would otherwise require GPU VRAM splitting on CUDA setups. Posted to Hacker News on April 7 as a Show HN, it pulled 109 upvotes and 165 GitHub stars within hours. The timing is sharp: Gemma 4 just dropped days ago with new multimodal capabilities, and the community immediately wanted local fine-tuning. This fills that gap faster than Google's own tooling.

Interfaze's Structured Output Benchmark (SOB) exposes a gap that has been quietly breaking production AI pipelines: models can produce syntactically valid JSON while getting the actual values wrong. SOB measures value accuracy across 21 models using 5,000 text passages, 209 OCR documents, and 115 meeting transcripts — scoring each on seven metrics including value accuracy, faithfulness (grounding vs. hallucination), type safety, and perfect-response rate. The benchmark reveals some sobering findings. Even top models like GPT-5.4 and Claude Sonnet 4.6 achieve ~83% on text but drop to 67% on images and only 23.7% on audio. No single model dominates all modalities — GPT-5.4, GLM-4.7, Qwen3.5-35B, and Gemini 2.5 Flash cluster within one point of each other on text. Perfect response rates (all seven metrics correct) rarely exceed 50% for even the best performers. For developers building data extraction pipelines, agents that read invoices, or any system where "correct JSON" means more than syntactically valid JSON, this is required reading. The dataset is on Hugging Face, the paper is on arXiv, and the playground lets you test your own model's structured output capability directly.