# Online model inference efficiency on Industrial Sponsored Search System (online serving)

Source: [Online model inference efficiency on Industrial Sponsored Search System (online serving)](<https://www.sota2.com/research/sota/online-model-inference-efficiency-on-industrial-sponsored-search-system-online-serving>)

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

Data updated: 2026-02-26T06:00:50.786Z

Evaluation of online model inference efficiency for industrial sponsored search systems, measuring latency (RT), prediction volume rate (PVR), and memory usage.

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 | Online model inference efficiency |
| Dataset (including any stated split) | Industrial Sponsored Search System (online serving) |
| Metric | RT (ms) |
| Unit | TIME |
| Direction | Lower is better |
| Results loaded | 3 of 3 |
| 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

- [RT (ms)](<https://www.sota2.com/research/sota/online-model-inference-efficiency-on-industrial-sponsored-search-system-online-serving.md?metric=RT%20(ms)>): Lower is better; unit: TIME.

- [PVR (%)](<https://www.sota2.com/research/sota/online-model-inference-efficiency-on-industrial-sponsored-search-system-online-serving.md?metric=PVR%20(%25)>): Higher is better; unit: PERCENT.

- [Memory (M)](<https://www.sota2.com/research/sota/online-model-inference-efficiency-on-industrial-sponsored-search-system-online-serving.md?metric=Memory%20(M)>): Lower is better; unit: SIZE.

## Recorded results

| Method | RT (ms) | Result date | Report paper |
| --- | --- | --- | --- |
| Base | 14.5 | Not recorded | 68fa717a1e209131cea517e4 |
| APG | 14.8 | Not recorded | 68fa717a1e209131cea517e4 |
| v1 | 57.1 | Not recorded | 68fa717a1e209131cea517e4 |

### Base

RT (ms): 14.5. Result ID: 6991d1733a6c3148490aca4f.

[Reporting paper](<https://www.sota2.com/research/paper/68fa717a1e209131cea517e4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa629e1e209131ce9fa416.md>)

[Implementation](<http://tensorflow.org/>)

Recorded evaluation configuration: {"Description":"Online model"}. This record does not assert independent reproduction.

### APG

RT (ms): 14.8. Result ID: 6991d1733a6c3148490aca50.

[Reporting paper](<https://www.sota2.com/research/paper/68fa717a1e209131cea517e4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa717a1e209131cea517e4.md>)

Recorded evaluation configuration: {"Model Version":"Full APG"}. This record does not assert independent reproduction.

### v1

RT (ms): 57.1. Result ID: 6991d1733a6c3148490aca51.

[Reporting paper](<https://www.sota2.com/research/paper/68fa717a1e209131cea517e4.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa717a1e209131cea517e4.md>)

Recorded evaluation configuration: {"Description":"Basic model of APG"}. This record does not assert independent reproduction.

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