# Health-domain instruction following on HealthBench 1K-example (eval)

Source: [Health-domain instruction following on HealthBench 1K-example (eval)](<https://www.sota2.com/research/sota/health-domain-instruction-following-on-healthbench-1k-example-eval>)

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

Data updated: 2026-03-18T06:41:19.254Z

Evaluation of model performance on instruction following tasks specifically within the health domain, using the HealthBench 1K-example (eval) dataset, measured by Score and Time/step.

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 | Health-domain instruction following |
| Dataset (including any stated split) | HealthBench 1K-example (eval) |
| Metric | Score |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 8 of 8 |
| 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

- [Score](<https://www.sota2.com/research/sota/health-domain-instruction-following-on-healthbench-1k-example-eval.md?metric=Score>): Higher is better; unit: PERCENT.

- [Time/step (s)](<https://www.sota2.com/research/sota/health-domain-instruction-following-on-healthbench-1k-example-eval.md?metric=Time%2Fstep%20(s)>): Lower is better; unit: TIME.

## Recorded results

| Method | Score | Result date | Report paper |
| --- | --- | --- | --- |
| Alternating RL | 78.8 | Not recorded | 69ba316a0636e5bf4afc05ed |
| Scalarized RL | 76.2 | Not recorded | 69ba316a0636e5bf4afc05ed |
| Alternating RL | 76.1 | Not recorded | 69ba316a0636e5bf4afc05ed |
| Scalarized RL | 75 | Not recorded | 69ba316a0636e5bf4afc05ed |
| Alternating RL | 71.6 | Not recorded | 69ba316a0636e5bf4afc05ed |
| Scalarized RL | 69 | Not recorded | 69ba316a0636e5bf4afc05ed |
| Alternating RL | 57.6 | Not recorded | 69ba316a0636e5bf4afc05ed |
| Scalarized RL | 55.6 | Not recorded | 69ba316a0636e5bf4afc05ed |

### Alternating RL

Score: 78.8. Result ID: 69ba482547f34ce924a374a4.

[Reporting paper](<https://www.sota2.com/research/paper/69ba316a0636e5bf4afc05ed.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69ba316a0636e5bf4afc05ed.md>)

Recorded evaluation configuration: {"Actor Model":"Qwen3-14B","RL Algorithm":"Alternating RL (ARL)"}. This record does not assert independent reproduction.

### Scalarized RL

Score: 76.2. Result ID: 69ba482547f34ce924a374a3.

[Reporting paper](<https://www.sota2.com/research/paper/69ba316a0636e5bf4afc05ed.md>)

Recorded evaluation configuration: {"Actor Model":"Qwen3-14B","RL Algorithm":"Scalarized aggregation (SRL)"}. This record does not assert independent reproduction.

### Alternating RL

Score: 76.1. Result ID: 69ba482547f34ce924a374a2.

[Reporting paper](<https://www.sota2.com/research/paper/69ba316a0636e5bf4afc05ed.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69ba316a0636e5bf4afc05ed.md>)

Recorded evaluation configuration: {"Actor Model":"Qwen3-8B","RL Algorithm":"Alternating RL (ARL)"}. This record does not assert independent reproduction.

### Scalarized RL

Score: 75. Result ID: 69ba482547f34ce924a374a1.

[Reporting paper](<https://www.sota2.com/research/paper/69ba316a0636e5bf4afc05ed.md>)

Recorded evaluation configuration: {"Actor Model":"Qwen3-8B","RL Algorithm":"Scalarized aggregation (SRL)"}. This record does not assert independent reproduction.

### Alternating RL

Score: 71.6. Result ID: 69ba482547f34ce924a374a0.

[Reporting paper](<https://www.sota2.com/research/paper/69ba316a0636e5bf4afc05ed.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69ba316a0636e5bf4afc05ed.md>)

Recorded evaluation configuration: {"Actor Model":"Qwen3-4B","RL Algorithm":"Alternating RL (ARL)"}. This record does not assert independent reproduction.

### Scalarized RL

Score: 69. Result ID: 69ba482547f34ce924a3749f.

[Reporting paper](<https://www.sota2.com/research/paper/69ba316a0636e5bf4afc05ed.md>)

Recorded evaluation configuration: {"Actor Model":"Qwen3-4B","RL Algorithm":"Scalarized aggregation (SRL)"}. This record does not assert independent reproduction.

### Alternating RL

Score: 57.6. Result ID: 69ba482547f34ce924a3749e.

[Reporting paper](<https://www.sota2.com/research/paper/69ba316a0636e5bf4afc05ed.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69ba316a0636e5bf4afc05ed.md>)

Recorded evaluation configuration: {"Actor Model":"Qwen3-1.7B","RL Algorithm":"Alternating RL (ARL)"}. This record does not assert independent reproduction.

### Scalarized RL

Score: 55.6. Result ID: 69ba482547f34ce924a3749d.

[Reporting paper](<https://www.sota2.com/research/paper/69ba316a0636e5bf4afc05ed.md>)

Recorded evaluation configuration: {"Actor Model":"Qwen3-1.7B","RL Algorithm":"Scalarized aggregation (SRL)"}. This record does not assert independent reproduction.

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