List your company

Embodied text agents on ALFWorld 134 (test)

77.4Success RateCVT-RL
70.74472.47274.275.928Jun 3, 2026

Evaluation results

MethodLinks
CVT-RLDescription=per-seed CVT-RL success, Split (train/dev/test)=3553/140/134, Individual Seeds=76.0/78.1/77.5/79.0/76.42026.0677.43.4
IMDescription=information-matched counterfactual-process RL, Split (train/dev/test)=3553/140/1342026.0674—
CMDescription=compute-matched non-causal RL, Split (train/dev/test)=3553/140/1342026.0671—