# Reasoning failure prediction and recovery on SAFIM L3

Source: [Reasoning failure prediction and recovery on SAFIM L3](<https://www.sota2.com/research/sota/reasoning-failure-prediction-and-recovery-on-safim-l3>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-04-21T09:25:19.194Z

Evaluation of model performance on reasoning failure prediction and recovery tasks using the SAFIM L3 dataset, measured by Accuracy and Intervention Pass@1.

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 | Reasoning failure prediction and recovery |
| Dataset (including any stated split) | SAFIM L3 |
| Metric | Accuracy |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 2 of 2 |
| 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

- [Accuracy](<https://www.sota2.com/research/sota/reasoning-failure-prediction-and-recovery-on-safim-l3.md?metric=Accuracy>): Higher is better; unit: PERCENT.

- [Intervention Pass@1](<https://www.sota2.com/research/sota/reasoning-failure-prediction-and-recovery-on-safim-l3.md?metric=Intervention%20Pass%401>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| thought-tree-based classifier | 81 | Not recorded | 69e704bc7d0172140ec7ee7c |
| thought-tree-based classifier | 71 | Not recorded | 69e704bc7d0172140ec7ee7c |

### thought-tree-based classifier

Accuracy: 81. Result ID: 69e7416947f34ce924cdc379.

[Reporting paper](<https://www.sota2.com/research/paper/69e704bc7d0172140ec7ee7c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69e704bc7d0172140ec7ee7c.md>)

Recorded evaluation configuration: {"Target Reasoning Model":"R1","Evaluation Protocol":"Generalization from L1 training"}. This record does not assert independent reproduction.

### thought-tree-based classifier

Accuracy: 71. Result ID: 69e7416947f34ce924cdc37a.

[Reporting paper](<https://www.sota2.com/research/paper/69e704bc7d0172140ec7ee7c.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69e704bc7d0172140ec7ee7c.md>)

Recorded evaluation configuration: {"Target Reasoning Model":"QwQ","Evaluation Protocol":"Generalization from L1 training"}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.