Chat Memory Reasoning on Chat Memory
67AccuracyQwen3-235B (w/ Pensieve)
Evaluation results
| Method | Links | |
|---|---|---|
| Qwen3-235B (w/ Pensieve)Context=256K2026.02 | 67 | |
| StateLM-14B-RLContext=32K2026.02 | 64.47 | |
| StateLM-14BContext=32K2026.02 | 64.4 | |
| StateLM-8B-RLContext=32K2026.02 | 59.73 | |
| StateLM-4BContext=32K2026.02 | 59.33 | |
| RL-MemoryAgent-14BContext=32K2026.02 | 59 | |
| StateLM-8BContext=32K2026.02 | 58.93 | |
| Qwen3-14BContext=128K2026.02 | 54.07 | |
| Qwen3-8BContext=128K2026.02 | 45.4 | |
| RL-MemoryAgent-7BContext=32K2026.02 | 40.6 | |
| Qwen3-4BContext=128K2026.02 | 39.53 | |
| ReadAgent-14BContext=32K2026.02 | 14.6 | |
| ReadAgent-8BContext=32K2026.02 | 0 |