Gemma

2026-10-21

RESEARCH · 1 SOURCE · Hugging Face

VocabX tokenizer specialization cuts inference cost 77% for Qwen3-4B essay grader

Researchers/engineers report applying a method called VocabX to rebuild the tokenizer for a Qwen3-4B-based automated Portuguese essay grader while keeping the same base model, data, and training recipe; they measure a 77% reduction in inference cost by optimizing the vocabulary around per-token decode cost rather than token frequency. The numbers come from a paper currently under review and may change before publication.

7.0

2026-10-20

RESEARCH · 1 SOURCE · Hugging Face

FairStress benchmark released to evaluate demographic fairness and decision stability in LLMs

The team introduced FairStress, a 300-scenario benchmark across 29 domains (hiring, healthcare, lending, education, housing, criminal justice, etc.) that measures demographic fairness beyond final model choices by testing how models respond to matched persuasive "adversarial prefixes." FairStress is available for download and will be integrated into Olmo-eval; initial results show asymmetric effects of identical persuasive arguments across models including Olmo, Llama, Qwen, Gemma, Claude, and the GPT family, and differences in explanations and internal activations across training stages.

7.0

2026-10-15

RESEARCH · 1 SOURCE · arXiv cs.AI

MOSAIC: sparse multimodal latent attribution for vision–language models (arXiv:2610.14190v1)

The paper introduces MOSAIC (Multimodal Orthogonalized Sparse Autoencoders for Influence Characterization), which learns a shared sparse image–text latent space with contiguous, phrase-aligned visual concepts and applies EK-FAC to decompose training-example influence through those concepts to specific phrases and image regions. The authors evaluate MOSAIC on Qwen and Gemma for captioning, chest X‑ray understanding, and visual question answering, reporting more localized and coherent features and that ablating influence-ranked latents alters outputs more than ablating latents ranked by activation or frequency.

7.0

2026-09-23

MODELS · 1 SOURCE · Google

Google's MedGemma open-weight medical models deployed in global healthcare pilots

Google has released MedGemma, a family of open-weight AI models for medical text and imaging built on its Gemma models; the models (downloaded over 10 million times) are being adapted by organizations including Crane AI, Dawa Health, Visilant, AIIMS Delhi and Indonesia’s Ministry of Health for offline clinical decision support, screening apps, dermatology triage and national TB screening pilots. MedGemma can be run on-device or on local servers and is distributed with code and documentation to support privacy-preserving, locally adapted deployments.

7.0