Lisanbench
Single-turn evaluation where the model is tasked to generate the longest valid chain of 1-word edits from a given starting word. The final score is the sum of the longest valid chains across all starting words.
- Domain
- rl-env
- License
- unknown
- Published
- Sep 2025
Cite
Notes
Only stored in your browser.
Top score 13.22 by Qwen3 4B - 1 model reporting
Top models
1
Related tools
1Implementations, trainers, datasets and scaffolds linked to this eval.
FAQ
- What is Lisanbench?
- Single-turn evaluation where the model is tasked to generate the longest valid chain of 1-word edits from a given starting word. The final score is the sum of the longest valid chains across all starting words.
- What is the current top score on Lisanbench?
- The top reported score is 13.22 by Qwen3 4B, across 1 model reporting.
- How can a model improve its Lisanbench score?
- Tools linked to Lisanbench on Sophon include Lisanbench RL Env (Prime Intellect) - RL environments, datasets, and scaffolds that target this eval.
- What license is Lisanbench under?
- Lisanbench is available under unknown.