# Preference evolution over long multi-session histories on PersonaMem 32K context scale

Source: [Preference evolution over long multi-session histories on PersonaMem 32K context scale](<https://www.sota2.com/research/sota/preference-evolution-over-long-multi-session-histories-on-personamem-32k-context-scale>)

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

Data updated: 2026-05-04T08:02:40.147Z

Evaluation of model performance on the PersonaMem 32K context scale dataset, focusing on preference evolution across long, multi-session histories, measured by Accuracy and Overall scores.

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 | Preference evolution over long multi-session histories |
| Dataset (including any stated split) | PersonaMem 32K context scale |
| Metric | Accuracy |
| 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

- [Accuracy](<https://www.sota2.com/research/sota/preference-evolution-over-long-multi-session-histories-on-personamem-32k-context-scale.md?metric=Accuracy>): Higher is better; unit: PERCENT.

- [Overall Score](<https://www.sota2.com/research/sota/preference-evolution-over-long-multi-session-histories-on-personamem-32k-context-scale.md?metric=Overall%20Score>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Accuracy | Result date | Report paper |
| --- | --- | --- | --- |
| MemCoE | 57.06 | Not recorded | 69f827eac0f56ed843088914 |
| MemAgent | 53.58 | Not recorded | 69f827eac0f56ed843088914 |
| Mem-α | 53.37 | Not recorded | 69f827eac0f56ed843088914 |
| LightMem | 50.72 | Not recorded | 69f827eac0f56ed843088914 |
| RAG | 48.67 | Not recorded | 69f827eac0f56ed843088914 |
| Mem0 | 48.53 | Not recorded | 69f827eac0f56ed843088914 |
| A-Mem | 48.26 | Not recorded | 69f827eac0f56ed843088914 |
| Long Context | 34.36 | Not recorded | 69f827eac0f56ed843088914 |

### MemCoE

Accuracy: 57.06. Result ID: 69f8528b47f34ce924dbd796.

[Reporting paper](<https://www.sota2.com/research/paper/69f827eac0f56ed843088914.md>)

[Method origin paper](<https://www.sota2.com/research/paper/69f827eac0f56ed843088914.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-7B-Instruct","Retrieval model":"all-MiniLM-L6-v2","Top-K retrieval":"10"}. This record does not assert independent reproduction.

### MemAgent

Accuracy: 53.58. Result ID: 69f8528b47f34ce924dbd794.

[Reporting paper](<https://www.sota2.com/research/paper/69f827eac0f56ed843088914.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa79c71e209131cea7f569.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-7B-Instruct","Retrieval model":"all-MiniLM-L6-v2","Top-K retrieval":"10"}. This record does not assert independent reproduction.

### Mem-α

Accuracy: 53.37. Result ID: 69f8528b47f34ce924dbd795.

[Reporting paper](<https://www.sota2.com/research/paper/69f827eac0f56ed843088914.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7bc91e209131cea8a773.md>)

Recorded evaluation configuration: {"Backbone":"Qwen3-4B","Retrieval model":"all-MiniLM-L6-v2","Top-K retrieval":"10"}. This record does not assert independent reproduction.

### LightMem

Accuracy: 50.72. Result ID: 69f8528b47f34ce924dbd793.

[Reporting paper](<https://www.sota2.com/research/paper/69f827eac0f56ed843088914.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa7c6d1e209131cea8e04d.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-7B-Instruct","Retrieval model":"all-MiniLM-L6-v2","Top-K retrieval":"10"}. This record does not assert independent reproduction.

### RAG

Accuracy: 48.67. Result ID: 69f8528b47f34ce924dbd790.

[Reporting paper](<https://www.sota2.com/research/paper/69f827eac0f56ed843088914.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-7B-Instruct","Retrieval model":"all-MiniLM-L6-v2","Top-K retrieval":"10"}. This record does not assert independent reproduction.

### Mem0

Accuracy: 48.53. Result ID: 69f8528b47f34ce924dbd791.

[Reporting paper](<https://www.sota2.com/research/paper/69f827eac0f56ed843088914.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa784b1e209131cea77ad6.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-7B-Instruct","Retrieval model":"all-MiniLM-L6-v2","Top-K retrieval":"10"}. This record does not assert independent reproduction.

### A-Mem

Accuracy: 48.26. Result ID: 69f8528b47f34ce924dbd792.

[Reporting paper](<https://www.sota2.com/research/paper/69f827eac0f56ed843088914.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa78171e209131cea76b84.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-7B-Instruct","Retrieval model":"all-MiniLM-L6-v2","Top-K retrieval":"10"}. This record does not assert independent reproduction.

### Long Context

Accuracy: 34.36. Result ID: 69f8528b47f34ce924dbd78f.

[Reporting paper](<https://www.sota2.com/research/paper/69f827eac0f56ed843088914.md>)

Recorded evaluation configuration: {"Backbone":"Qwen2.5-7B-Instruct"}. This record does not assert independent reproduction.

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