# Puzzle Solving on Sokoban (In Distribution) Source: [Puzzle Solving on Sokoban (In Distribution)]() Collection index: [llms.txt]() Data updated: 2026-03-23T11:44:27.059Z Evaluation of model performance on solving puzzles within the in-distribution split of the Sokoban environment, measured using Pass@K and Average metrics. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Puzzle Solving | | Dataset (including any stated split) | Sokoban (In Distribution) | | Metric | Average Score @128 | | Unit | SCALAR | | Direction | Higher is better | | Results loaded | 12 of 12 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Average Score @128](): Higher is better; unit: SCALAR. - [Pass@128](): Higher is better; unit: PERCENT. ## Recorded results | Method | Average Score @128 | Result date | Report paper | | --- | --- | --- | --- | | Markov | 89.7 | Not recorded | 69c0c8ed0636e5bf4afc3aaa | | Markov | 76.1 | Not recorded | 69c0c8ed0636e5bf4afc3aaa | | State-action-sequence | 57.4 | Not recorded | 69c0c8ed0636e5bf4afc3aaa | | State-action-sequence | 43.6 | Not recorded | 69c0c8ed0636e5bf4afc3aaa | | Action-sequence | 2.3 | Not recorded | 69c0c8ed0636e5bf4afc3aaa | | State-action-sequence | 1.4 | Not recorded | 69c0c8ed0636e5bf4afc3aaa | | Action-sequence | 1 | Not recorded | 69c0c8ed0636e5bf4afc3aaa | | State-action-sequence | 0.6 | Not recorded | 69c0c8ed0636e5bf4afc3aaa | | Action-sequence | 0.5 | Not recorded | 69c0c8ed0636e5bf4afc3aaa | | Markov | 0.4 | Not recorded | 69c0c8ed0636e5bf4afc3aaa | | Action-sequence | 0.2 | Not recorded | 69c0c8ed0636e5bf4afc3aaa | | Markov | 0.2 | Not recorded | 69c0c8ed0636e5bf4afc3aaa | ### Markov Average Score @128: 89.7. Result ID: 69c1272247f34ce924a9b3e5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-3B-It","Training Stage":"RL post-training"}. This record does not assert independent reproduction. ### Markov Average Score @128: 76.1. Result ID: 69c1272247f34ce924a9b3df. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen3-4B","Training Stage":"RL post-training"}. This record does not assert independent reproduction. ### State-action-sequence Average Score @128: 57.4. Result ID: 69c1272247f34ce924a9b3e1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen3-4B","Training Stage":"RL post-training"}. This record does not assert independent reproduction. ### State-action-sequence Average Score @128: 43.6. Result ID: 69c1272247f34ce924a9b3e7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-3B-It","Training Stage":"RL post-training"}. This record does not assert independent reproduction. ### Action-sequence Average Score @128: 2.3. Result ID: 69c1272247f34ce924a9b3dd. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen3-4B","Training Stage":"RL post-training"}. This record does not assert independent reproduction. ### State-action-sequence Average Score @128: 1.4. Result ID: 69c1272247f34ce924a9b3e0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen3-4B","Training Stage":"SFT Warm-up"}. This record does not assert independent reproduction. ### Action-sequence Average Score @128: 1. Result ID: 69c1272247f34ce924a9b3e3. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-3B-It","Training Stage":"RL post-training"}. This record does not assert independent reproduction. ### State-action-sequence Average Score @128: 0.6. Result ID: 69c1272247f34ce924a9b3e6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-3B-It","Training Stage":"SFT Warm-up"}. This record does not assert independent reproduction. ### Action-sequence Average Score @128: 0.5. Result ID: 69c1272247f34ce924a9b3e2. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-3B-It","Training Stage":"SFT Warm-up"}. This record does not assert independent reproduction. ### Markov Average Score @128: 0.4. Result ID: 69c1272247f34ce924a9b3de. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen3-4B","Training Stage":"SFT Warm-up"}. This record does not assert independent reproduction. ### Action-sequence Average Score @128: 0.2. Result ID: 69c1272247f34ce924a9b3dc. [Reporting paper]() Recorded evaluation configuration: {"Model":"Qwen3-4B","Training Stage":"SFT Warm-up"}. This record does not assert independent reproduction. ### Markov Average Score @128: 0.2. Result ID: 69c1272247f34ce924a9b3e4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model":"Qwen2.5-3B-It","Training Stage":"SFT Warm-up"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.