News Company news Data Development & Processing · 1 min read

Agreement: Legal-domain LLM pre-training and instruction-tuning data

This agreement covers legal-domain LLM pre-training and instruction-tuning data for a government digital & AI agency.

Contract date
Service area
Data Development & Processing
Client type
A government digital & AI agency
Role
Consortium member
Branching knowledge tree of terminology and language assets

Agreement overview

This agreement covers legal-domain LLM pre-training and instruction-tuning data for a government digital & AI agency.

Scope of the agreement

  • Contributed to building pre-training data for legal-domain LLMs
  • Contributed to building instruction-response data for instruction tuning
  • Processed data per annotation guidelines with cross-checks by two or more reviewers
  • Managed quality against diversity, accuracy and validity metrics

Technologies & capabilities

LLM training data · Instruction tuning · Data quality management

Contract scope at a glance

Diagram of the main scope of the agreement

← Back to contracts & partnerships