LEAD: Breaking the No-Recovery Bottleneck in Long-Horizon Reasoning
Fuente:
arXiv
Salvato in:
| Autori principali: | , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866918461651812352 |
|---|---|
| author | Pushkin, Denys Abbe, Emmanuel |
| author_facet | Pushkin, Denys Abbe, Emmanuel |
| contents | Long-horizon execution in Large Language Models (LLMs) remains unstable even when high-level strategies are provided. Evaluating on controlled algorithmic puzzles, we demonstrate that while decomposition is essential for stability, extreme decomposition creates a "no-recovery bottleneck". We show that this bottleneck becomes critical due to highly non-uniform error distribution, where consistent errors on a few "hard" steps become irreversible.
To address this, we propose Lookahead-Enhanced Atomic Decomposition (LEAD). By incorporating short-horizon future validation and aggregating overlapping rollouts, LEAD provides enough isolation to maintain stability while retaining enough local context to correct errors. This enables the o4-mini model to solve Checkers Jumping up to complexity $n=13$, whereas extreme decomposition fails beyond $n=11$. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_06870 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | LEAD: Breaking the No-Recovery Bottleneck in Long-Horizon Reasoning Pushkin, Denys Abbe, Emmanuel Artificial Intelligence Long-horizon execution in Large Language Models (LLMs) remains unstable even when high-level strategies are provided. Evaluating on controlled algorithmic puzzles, we demonstrate that while decomposition is essential for stability, extreme decomposition creates a "no-recovery bottleneck". We show that this bottleneck becomes critical due to highly non-uniform error distribution, where consistent errors on a few "hard" steps become irreversible. To address this, we propose Lookahead-Enhanced Atomic Decomposition (LEAD). By incorporating short-horizon future validation and aggregating overlapping rollouts, LEAD provides enough isolation to maintain stability while retaining enough local context to correct errors. This enables the o4-mini model to solve Checkers Jumping up to complexity $n=13$, whereas extreme decomposition fails beyond $n=11$. |
| title | LEAD: Breaking the No-Recovery Bottleneck in Long-Horizon Reasoning |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2603.06870 |