Interactive Embodied Instruction Following on ALFWorld In-domain
97ScoreTask-Perturbed NLL Optimization
Evaluation Results
| Method | Links | |
|---|---|---|
| Task-Perturbed NLL OptimizationBackbone=Qwen3-8B2026.06 | 97 | |
| GRPOBackbone=Qwen3-8B2026.06 | 95 | |
| Task-Perturbed NLL OptimizationBackbone=Qwen3-4B2026.06 | 86 | |
| GRPOBackbone=Qwen3-4B2026.06 | 82 |