Search

Any time Past 24 hours Past week Past monthSort:Relevance Newest

Benchmark

ALFWorld

Reported in 65 papers on Zotpaper. The latest, each as the paper states it:

AI agents complete more multistep tasks by reusing related experience94.01% In-Success
AI agents can reuse successful decisions to learn with fewer tokens88.3% success rate

All benchmarks →

2 results for “ALFWorld”

Research · Robotics & embodied AI
Symbolic checks make household robot plans safer and cheaper to repair
SAGE: Symbolic Action-Gating and Editing for LLM Task Planners
SAGE blocks actions that violate environmental preconditions and regenerates only the affected portion of an LLM-generated plan.
AI2-THOR: 0.71 step_success (as reported)
29 Sept 2026Korea Electronics Technology Institute (KETI)
Research · Agents
Visual skill representations reduce exposure-driven action bias in language model agents
Procedural Exposure Bias in Skill-Augmented Language Model Agents
The same procedural content, shown as text versus as an image, leads to measurable differences in action preference and downstream agent success.
ALFWorld, ScienceWorld, WebShop: 56.34% macro-average success rate (as reported)
20 Oct 2026Northeastern University, Alibaba Group