Lhaw Rlm
LHAW RLM environment: underspecified prompts, simulated user clarification (ask_user), and LLM judge scoring on the ScaleAI/lhaw dataset.
- Domain
- rl-env
- License
- unknown
- Published
- Mar 2026
Cite
Notes
Only stored in your browser.
Top score 43.8% by GPT-4.1 Mini - 1 model reporting (1 frontier)
Top models
1
Related tools
1Implementations, trainers, datasets and scaffolds linked to this eval.
FAQ
- What is Lhaw Rlm?
- LHAW RLM environment: underspecified prompts, simulated user clarification (ask_user), and LLM judge scoring on the ScaleAI/lhaw dataset.
- What is the current top score on Lhaw Rlm?
- The top reported score is 43.8% by GPT-4.1 Mini, across 1 model reporting (1 from frontier labs).
- How can a model improve its Lhaw Rlm score?
- Tools linked to Lhaw Rlm on Sophon include LHAW RLM RL Env (Community) - RL environments, datasets, and scaffolds that target this eval.
- What license is Lhaw Rlm under?
- Lhaw Rlm is available under unknown.