Thematic Generalization
This benchmark measures how effectively various LLMs can infer a narrow or specific "theme" (category/rule) from a small set of examples and anti-examples, then detect which item truly fits that theme among a collection of misleading candidates.
- Domain
- rl-env
- License
- unknown
- Published
- Sep 2025
Cite
Notes
Only stored in your browser.
Top score 65.6% by gpt-oss-120b - 1 model reporting (1 frontier)
Top models
1
Related tools
1Implementations, trainers, datasets and scaffolds linked to this eval.
FAQ
- What is Thematic Generalization?
- This benchmark measures how effectively various LLMs can infer a narrow or specific "theme" (category/rule) from a small set of examples and anti-examples, then detect which item truly fits that theme among a collection of misleading candidates.
- What is the current top score on Thematic Generalization?
- The top reported score is 65.6% by gpt-oss-120b, across 1 model reporting (1 from frontier labs).
- How can a model improve its Thematic Generalization score?
- Tools linked to Thematic Generalization on Sophon include Thematic Generalization RL Env (Prime Intellect) - RL environments, datasets, and scaffolds that target this eval.
- What license is Thematic Generalization under?
- Thematic Generalization is available under unknown.