ALFWorld
A widely used evaluation benchmark in embodied agent research. Its tasks involve completing a series of object manipulation operations in a simulated household environment, used to assess agent planning and execution capabilities.
Timeline (last 90 days)
The paper evaluated GEPA-Curator on four agent benchmarks: GoBrowse, ALFWorld, AppWorld, and τ-bench.
ALFWorld is a text-based embodied decision-making benchmark in which an agent must complete multi-step operations in a simulated household environment according to natural-language instructions, and the success rate is the proportion of completed tasks.
J-Comm reaches 94.80% on MATH500, 54.55% on GPQA-Diamond, 73.33% on AIME2024+2025, and a 62.93% success rate on ALFWorld.
ALFWorld-based coordinates can support performance prediction and identify behaviorally similar models.
The behavior map method has so far been validated mainly in environments such as ALFWorld, and transfer to more open environments still requires further research.
In ALFWorld, behavioral similarity persists after controlling for success rate and model size, and even spans different model families.
SAPD was experimented on ALFWorld, WebShop, and search-based question answering tasks.
This reproduction study covers four agent task environments—ALFWorld, ScienceWorld, Search-QA, and WebShop—spans three model scales, and compares 10 representative self-distillation methods against a GRPO baseline.
CoPE was validated on five benchmarks: ALFWorld, ScienceWorld, WebShop, τ²-bench, and AppWorld.
SkillScaffold was evaluated on four benchmarks—ALFWorld, WebShop, ScienceWorld, and τ²-Bench—and surpassed strong agentic RL baselines and skill-driven RL baselines on both short-horizon and long-horizon tasks.
13 more timeline events
All Facts (20)
The paper evaluated GEPA-Curator on four agent benchmarks: GoBrowse, ALFWorld, AppWorld, and τ-bench.
50%UnverifiedALFWorld is a text-based embodied decision-making benchmark in which an agent must complete multi-step operations in a simulated household environment according to natural-language instructions, and the success rate is the proportion of completed tasks.
50%UnverifiedJ-Comm reaches 94.80% on MATH500, 54.55% on GPQA-Diamond, 73.33% on AIME2024+2025, and a 62.93% success rate on ALFWorld.
50%UnverifiedALFWorld-based coordinates can support performance prediction and identify behaviorally similar models.
50%UnverifiedThe behavior map method has so far been validated mainly in environments such as ALFWorld, and transfer to more open environments still requires further research.
50%UnverifiedIn ALFWorld, behavioral similarity persists after controlling for success rate and model size, and even spans different model families.
50%UnverifiedSAPD was experimented on ALFWorld, WebShop, and search-based question answering tasks.
50%UnverifiedThis reproduction study covers four agent task environments—ALFWorld, ScienceWorld, Search-QA, and WebShop—spans three model scales, and compares 10 representative self-distillation methods against a GRPO baseline.
50%UnverifiedCoPE was validated on five benchmarks: ALFWorld, ScienceWorld, WebShop, τ²-bench, and AppWorld.
50%UnverifiedSkillScaffold was evaluated on four benchmarks—ALFWorld, WebShop, ScienceWorld, and τ²-Bench—and surpassed strong agentic RL baselines and skill-driven RL baselines on both short-horizon and long-horizon tasks.
50%UnverifiedCompared to a scheme with seven independent trajectories, shared prefix reuse reduces student-generated token costs by approximately 19% on ALFWorld.
50%UnverifiedOn the ID/OOD average success rate on ALFWorld, BranchOPD improves over Vanilla OPD by 13.8 percentage points.
50%UnverifiedAfter targeted training using the ALFWorld and TextCraft benchmarks, the two types of planners performed similarly on plan generation success rate.
50%UnverifiedALFWorld combines text instructions with embodied navigation, requiring agents to perform everyday tasks in a virtual household environment.
50%UnverifiedThe research team validated SAGA on two interactive environment benchmarks, ScienceWorld and ALFWorld, achieving improved task completion performance.
50%UnverifiedALFWorld is a text-based household interaction benchmark used to evaluate language models' planning and tool-use capabilities.
50%UnverifiedOn ALFWorld, CTWM achieved consistent token savings under identical paired comparisons.
50%UnverifiedThe research team validated MATE's effectiveness on 134 ALFWorld tasks.
50%UnverifiedMATE's current validation is primarily focused on the ALFWorld benchmark, and its generalization capability in more complex and open-ended real-world environments remains to be further examined.
50%UnverifiedExperiments were conducted in the ALFWorld environment using a frozen 7B model as the backbone.
50%