StanfordVL/BEHAVIOR-1K
BDDL: Refactor goal state generation to support lazy evaluation with iterators
开放
#1,845 创建于 2025年9月30日
good first issue
仓库指标
- 星标
- (1,604 个星标)
- PR 合并指标
- (PR 指标待抓取)
描述
The current implementation in bddl/bddl/condition_evaluation.py eagerly converts iterators to lists in multiple places during goal state generation, which prevents users from being able to stop generation early and causes unnecessary memory usage for large combinatorial spaces.