Neurosymbolic Program Induction Tracker Content Archive

14 posts · Page 1 of 1

1. A DreamCoder-style test of semantic primitives2026-07-09
2. DreamCoder learns photonic decomposition rules that scale beyond the training dimension2026-07-20
3. Neural Language Interpreter learns a program language and searches it with gradients2026-07-27
4. Solver-aware decomposition trades ground-truth subgoals for tractable program search2026-08-10
5. Hypothesis Frontier 把验证器变成搜索状态，却没有学习 DreamCoder 式库2026-08-17
6. aDSL makes 3D spatial relations executable, then lets the critic repair them2026-08-24
7. Narcissus keeps LLM proposals in context—and leaves every rule reachable2026-08-31
8. REFACTOR-VLA finds a library signal in robot trajectories, then loses the motor programs2026-09-07
9. SimSkill accumulates 150 simulator skills in plain text, but trades inference cost for task coverage2026-09-14
10. A pretrained transformer's equation sketches recover a third of exact formulas; repair search reaches 85%2026-09-21
11. Dozens of banned expression schemas and a laptop enumerator beat a coding agent on 14 continuous games2026-09-28
12. A robot writes the physics its simulator is missing, then infers it from a few noisy experiments2026-10-05
13. An LLM proposes one logic rule per round; gradient boosting decides which survive2026-10-12
14. Certified pattern edits train a search ranker: 63 of 80 cases, and 0 of 57 with an unseen action2026-10-19

Explore more channels on Discover