# Skill-selection on TOOLBENCH-60 soft-0.3 preference regime

Source: [Skill-selection on TOOLBENCH-60 soft-0.3 preference regime](<https://www.sota2.com/research/sota/skill-selection-on-toolbench-60-soft-0-3-preference-regime>)

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

Data updated: 2026-06-05T09:11:26.726Z

Evaluation of the model's ability to select the correct skill using the ToolBench-60 dataset under the soft-0.3 preference regime, measured by Regret, Accuracy, and SRC.

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 | Skill-selection |
| Dataset (including any stated split) | TOOLBENCH-60 soft-0.3 preference regime |
| Metric | Regret |
| Unit | SCALAR |
| Direction | Lower is better |
| Results loaded | 9 of 9 |
| 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

- [Regret](<https://www.sota2.com/research/sota/skill-selection-on-toolbench-60-soft-0-3-preference-regime.md?metric=Regret>): Lower is better; unit: SCALAR.

- [Accuracy](<https://www.sota2.com/research/sota/skill-selection-on-toolbench-60-soft-0-3-preference-regime.md?metric=Accuracy>): Higher is better; unit: PERCENT.

- [SRC](<https://www.sota2.com/research/sota/skill-selection-on-toolbench-60-soft-0-3-preference-regime.md?metric=SRC>): Higher is better; unit: SCALAR.

## Recorded results

| Method | Regret | Result date | Report paper |
| --- | --- | --- | --- |
| Bandit-as-Override | 264.8 | Not recorded | 6a2258a186173ad9c1468be8 |
| Pure-Bandit | 282 | Not recorded | 6a2258a186173ad9c1468be8 |
| Freq-as-Override | 295.3 | Not recorded | 6a2258a186173ad9c1468be8 |
| Freq-Greedy | 328.5 | Not recorded | 6a2258a186173ad9c1468be8 |
| Profile-Memory | 344.2 | Not recorded | 6a2258a186173ad9c1468be8 |
| Bandit-as-Context | 369.6 | Not recorded | 6a2258a186173ad9c1468be8 |
| InContext-Memory | 372.7 | Not recorded | 6a2258a186173ad9c1468be8 |
| ZeroShot-LLM | 377.5 | Not recorded | 6a2258a186173ad9c1468be8 |
| Random | 491.2 | Not recorded | 6a2258a186173ad9c1468be8 |

### Bandit-as-Override

Regret: 264.8. Result ID: 6a2291ddfbc373592c20dfc2.

[Reporting paper](<https://www.sota2.com/research/paper/6a2258a186173ad9c1468be8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a2258a186173ad9c1468be8.md>)

Recorded evaluation configuration: {"Method Category":"LLM with Statistical Prior","Backbone":"Qwen3-30B-Instruct"}. This record does not assert independent reproduction.

### Pure-Bandit

Regret: 282. Result ID: 6a2291ddfbc373592c20dfbd.

[Reporting paper](<https://www.sota2.com/research/paper/6a2258a186173ad9c1468be8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa60891e209131ce9efe66.md>)

Recorded evaluation configuration: {"Method Category":"Statistical","Backbone":"Qwen3-30B-Instruct"}. This record does not assert independent reproduction.

### Freq-as-Override

Regret: 295.3. Result ID: 6a2291ddfbc373592c20dfc1.

[Reporting paper](<https://www.sota2.com/research/paper/6a2258a186173ad9c1468be8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/6a2258a186173ad9c1468be8.md>)

Recorded evaluation configuration: {"Method Category":"LLM with Statistical Prior","Backbone":"Qwen3-30B-Instruct"}. This record does not assert independent reproduction.

### Freq-Greedy

Regret: 328.5. Result ID: 6a2291ddfbc373592c20dfbc.

[Reporting paper](<https://www.sota2.com/research/paper/6a2258a186173ad9c1468be8.md>)

Recorded evaluation configuration: {"Method Category":"Statistical","Backbone":"Qwen3-30B-Instruct"}. This record does not assert independent reproduction.

### Profile-Memory

Regret: 344.2. Result ID: 6a2291ddfbc373592c20dfbf.

[Reporting paper](<https://www.sota2.com/research/paper/6a2258a186173ad9c1468be8.md>)

Recorded evaluation configuration: {"Method Category":"LLM with Memory","Backbone":"Qwen3-30B-Instruct"}. This record does not assert independent reproduction.

### Bandit-as-Context

Regret: 369.6. Result ID: 6a2291ddfbc373592c20dfc0.

[Reporting paper](<https://www.sota2.com/research/paper/6a2258a186173ad9c1468be8.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa60891e209131ce9efe66.md>)

Recorded evaluation configuration: {"Method Category":"LLM with Memory","Backbone":"Qwen3-30B-Instruct"}. This record does not assert independent reproduction.

### InContext-Memory

Regret: 372.7. Result ID: 6a2291ddfbc373592c20dfbe.

[Reporting paper](<https://www.sota2.com/research/paper/6a2258a186173ad9c1468be8.md>)

Recorded evaluation configuration: {"Method Category":"LLM with Memory","Backbone":"Qwen3-30B-Instruct"}. This record does not assert independent reproduction.

### ZeroShot-LLM

Regret: 377.5. Result ID: 6a2291ddfbc373592c20dfbb.

[Reporting paper](<https://www.sota2.com/research/paper/6a2258a186173ad9c1468be8.md>)

Recorded evaluation configuration: {"Method Category":"No Learning","Backbone":"Qwen3-30B-Instruct"}. This record does not assert independent reproduction.

### Random

Regret: 491.2. Result ID: 6a2291ddfbc373592c20dfba.

[Reporting paper](<https://www.sota2.com/research/paper/6a2258a186173ad9c1468be8.md>)

Recorded evaluation configuration: {"Method Category":"No Learning","Backbone":"Qwen3-30B-Instruct"}. This record does not assert independent reproduction.

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