Human Data
Human experience is what turns information into intelligence.
License curated datasets or work with us to build expert data for a capability your model needs to learn.
Off-the-shelf data
Curated collections, ready to license.
Our catalogue provides licensed training and evaluation data for difficult model capabilities. Every release states its provenance, permitted use, composition, quality process and limits.
Clinical Triage Reasoning - UK
Doctor-authored pathways transformed into auditable graphs, decision turns, interactive trajectories and correction data for sequential UK clinical triage.
- Modalities
- Clinical graphs · dialogue · structured actions
- Designed for
- Action tuning, interactive reasoning, safety evaluation and RL
Why human data
Scale alone does not make a model useful.
Models do not become useful through scale alone. They improve when training captures how people reason, decide, create, correct mistakes and recognise when an answer is not good enough.
Logarithms Labs turns expert work into training data, feedback and evaluations. We begin with the capability a model must learn, recruit people qualified to judge it and build a production system that makes their judgement consistent and measurable.
The result is data with a purpose: stronger reasoning, better decisions and model behaviour that survives contact with the real world.
Bespoke programmes
Start with the capability. Build the data it needs.
We design end-to-end human data programmes for post-training and evaluation. The work can include demonstrations, preference data, rubric-scored outputs, tool-use trajectories and private test sets.
Data strategy
Define the model behaviour, evidence and evaluation plan before collection begins.
Expert recruitment
Find and qualify people whose judgement represents the work the model must learn.
Task production
Turn real workflows into examples, demonstrations, comparisons and adversarial cases.
Quality systems
Use clear rubrics, calibration, review and disagreement analysis to protect signal quality.
Dataset standard
Useful data needs a clear record of how it was made.
- Documented provenance, consent and permitted use
- Contributors selected for relevant expertise
- Human verification against explicit rubrics
- Train, validation and private evaluation splits
- Versioned collection and quality-control methods
- Known limitations recorded in a data card