GPBam: New German Public Law Benchmark Finds Retrieval Gains Limited, Essay Quality Tracks Statutory Knowledge
aaire · hf · 2026-11-18
GPBam is a new benchmark built from German public-law exam cases and statutory provisions, jointly scoring models on legal retrieval and long-form opinion generation.
Across many models, two key findings emerge:
- Retrieval augmentation yields only limited gains;
- Opinion quality strongly tracks the model's prior statutory knowledge.
In short, a model's pre-existing legal knowledge predicts writing performance better than bolted-on retrieval.
More from Research
- Stanford team predicts viral evolution and designs antibodies that neutralize future escape variants — BrianHie · 2026-11-18
- AI agents uncover first known genome-wide frameshifting in bacteria — GaryMarcus · 2026-11-18
- Google launches Antigravity Science plugin, gives 10,000 scientists free top-tier access — DynamicWebPaige · 2026-11-18
- AI predicts viral evolution and designs antibodies that neutralize forecasted escape variants — BrianHie · 2026-11-18
- Atomistic foundation models hit data-quality wall: fit error now below DFT label error — matt_slotnick · 2026-11-18
- New ApexW optimizer claims lower loss than AdamW with half the VRAM of bitsandbytes — omendUUde · 2026-11-18