Gemma Team: original lightweight open-model training resource

What changed

Released pretrained and instruction-tuned lightweight language-model checkpoints with a documented training, filtering and evaluation recipe adapted from Gemini research.

Scientific scope

Released pretrained and instruction-tuned lightweight language-model checkpoints with a documented training, filtering and evaluation recipe adapted from Gemini research. A reusable model-resource/system contribution, not a claim to invent its inherited MQA, RoPE, GeGLU, normalization, sharding or alignment methods. The two original sizes form one record; later Gemma2 distillation remains distinct.

Evidence

The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. A reusable model-resource/system contribution, not a claim to invent its inherited MQA, RoPE, GeGLU, normalization, sharding or alignment methods. The two original sizes form one record; later Gemma2 distillation remains distinct.

Contributors

• Source-listed authors and technical contributors · SCIENCE: Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.

• Gemma: Open Models Based on Gemini Research and Technology
• Gemma Team: original lightweight open-model training resource — official research record

Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.

Full scientific record · Browse contribution grades · Request a correction