# Membership Inference on SteamMIA

Source: [Membership Inference on SteamMIA](<https://www.sota2.com/research/sota/membership-inference-on-steammia>)

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

Data updated: 2026-06-09T07:23:08.323Z

Evaluation of Membership Inference performance on the SteamMIA dataset, utilizing metrics focused on hit rates, statistical significance (P-value), and confidence intervals.

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 | Membership Inference |
| Dataset (including any stated split) | SteamMIA |
| Metric | Hit Rate (Non-members) |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 6 of 6 |
| 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

- [Hit Rate (Non-members)](<https://www.sota2.com/research/sota/membership-inference-on-steammia.md?metric=Hit%20Rate%20(Non-members)>): Higher is better; unit: PERCENT.

- [Hit Rate (Members)](<https://www.sota2.com/research/sota/membership-inference-on-steammia.md?metric=Hit%20Rate%20(Members)>): Higher is better; unit: PERCENT.

- [Hit Rate Difference](<https://www.sota2.com/research/sota/membership-inference-on-steammia.md?metric=Hit%20Rate%20Difference>): Higher is better; unit: PERCENT.

- [P-value](<https://www.sota2.com/research/sota/membership-inference-on-steammia.md?metric=P-value>): Lower is better; unit: SCALAR.

- [Confidence Interval Upper Bound](<https://www.sota2.com/research/sota/membership-inference-on-steammia.md?metric=Confidence%20Interval%20Upper%20Bound>): Lower is better; unit: PERCENT.

## Recorded results

| Method | Hit Rate (Non-members) | Result date | Report paper |
| --- | --- | --- | --- |
| Deepseek-V3-Chat | 49.77 | Not recorded | 6a279df890966ade919bf60e |
| Gemini-1.5-Flash\* | 44.37 | Not recorded | 6a279df890966ade919bf60e |
| LLaMA-3-70B | 43.92 | Not recorded | 6a279df890966ade919bf60e |
| Claude-3-Haiku | 40.54 | Not recorded | 6a279df890966ade919bf60e |
| GPT-4o-Mini | 37.16 | Not recorded | 6a279df890966ade919bf60e |
| LLaMA-3.1-8B-Instruct | 36.26 | Not recorded | 6a279df890966ade919bf60e |

### Deepseek-V3-Chat

Hit Rate (Non-members): 49.77. Result ID: 6a27beeffbc373592c241957.

[Reporting paper](<https://www.sota2.com/research/paper/6a279df890966ade919bf60e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa77001e209131cea7066a.md>)

Recorded evaluation configuration: {"LLM":"Deepseek-V3-Chat"}. This record does not assert independent reproduction.

### Gemini-1.5-Flash\*

Hit Rate (Non-members): 44.37. Result ID: 6a27beeffbc373592c24195a.

[Reporting paper](<https://www.sota2.com/research/paper/6a279df890966ade919bf60e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa76361e209131cea6be11.md>)

Recorded evaluation configuration: {"LLM":"Gemini-1.5-Flash\*","Knowledge Cutoff Date":"Later than December 2023"}. This record does not assert independent reproduction.

### LLaMA-3-70B

Hit Rate (Non-members): 43.92. Result ID: 6a27beeffbc373592c241959.

[Reporting paper](<https://www.sota2.com/research/paper/6a279df890966ade919bf60e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75d11e209131cea6a028.md>)

[Implementation](<https://github.com/meta-llama/llama-agentic-system>)

Recorded evaluation configuration: {"LLM":"LLaMA-3-70B"}. This record does not assert independent reproduction.

### Claude-3-Haiku

Hit Rate (Non-members): 40.54. Result ID: 6a27beeffbc373592c241956.

[Reporting paper](<https://www.sota2.com/research/paper/6a279df890966ade919bf60e.md>)

Recorded evaluation configuration: {"LLM":"Claude-3-Haiku"}. This record does not assert independent reproduction.

### GPT-4o-Mini

Hit Rate (Non-members): 37.16. Result ID: 6a27beeffbc373592c241958.

[Reporting paper](<https://www.sota2.com/research/paper/6a279df890966ade919bf60e.md>)

Recorded evaluation configuration: {"LLM":"GPT-4o-Mini"}. This record does not assert independent reproduction.

### LLaMA-3.1-8B-Instruct

Hit Rate (Non-members): 36.26. Result ID: 6a27beeffbc373592c241955.

[Reporting paper](<https://www.sota2.com/research/paper/6a279df890966ade919bf60e.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa75d11e209131cea6a028.md>)

[Implementation](<https://github.com/meta-llama/llama-agentic-system>)

Recorded evaluation configuration: {"LLM":"LLaMA-3.1-8B-Instruct"}. This record does not assert independent reproduction.

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