Personalized product reranking on Electronics (test)
42.47Accuracy+ MEMRERANK + think tag
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| + MEMRERANK + think tagReranker=o4-mini, Memory extractor=RL-trained Qwen2.5-7B-Instruct, Reasoning=think tag2026.03 | 42.47 | 10.61 | |
| + MEMRERANKReranker=o4-mini, Memory extractor=RL-trained Qwen2.5-7B-Instruct2026.03 | 40.72 | 8.86 | |
| + MEMRERANK + think tagReranker=GPT-4.1-mini, Memory extractor=RL-trained Qwen2.5-7B-Instruct, Reasoning=think tag2026.03 | 39.07 | 6.6 | |
| + Base-extractor memory + think tagReranker=o4-mini, Base-extractor=Qwen2.5-7B-Instruct, Reasoning=think tag2026.03 | 38.66 | 6.8 | |
| + MEMRERANKReranker=GPT-4.1-mini, Memory extractor=RL-trained Qwen2.5-7B-Instruct2026.03 | 38.14 | 5.67 | |
| + Base-extractor memoryReranker=o4-mini, Base-extractor=Qwen2.5-7B-Instruct2026.03 | 36.29 | 4.43 | |
| MR.RecReranker=o4-mini2026.03 | 36.08 | 4.22 | |
| Mem0Reranker=o4-mini2026.03 | 35.88 | 4.02 | |
| + GPT-5.2 memoryReranker=o4-mini, Memory extractor=GPT-5.22026.03 | 35.77 | 3.91 | |
| + Within + cross-category productsReranker=o4-mini2026.03 | 34.85 | 2.99 | |
| + Within + cross-category productsReranker=GPT-4.1-mini2026.03 | 34.02 | 1.55 | |
| + Within-category productsReranker=o4-mini2026.03 | 34.02 | 2.16 | |
| + GPT-5.2 memoryReranker=GPT-4.1-mini, Memory extractor=GPT-5.22026.03 | 33.92 | 1.45 | |
| + Base-extractor memory + think tagReranker=GPT-4.1-mini, Base-extractor=Qwen2.5-7B-Instruct, Reasoning=think tag2026.03 | 33.3 | 0.83 | |
| + Within-category productsReranker=GPT-4.1-mini2026.03 | 32.58 | 0.11 | |
| No memoryReranker=GPT-4.1-mini2026.03 | 32.47 | — | |
| MR.RecReranker=GPT-4.1-mini2026.03 | 32.47 | 0 | |
| + Base-extractor memoryReranker=GPT-4.1-mini, Base-extractor=Qwen2.5-7B-Instruct2026.03 | 31.96 | -0.51 | |
| No memoryReranker=o4-mini2026.03 | 31.86 | — | |
| Mem0Reranker=GPT-4.1-mini2026.03 | 30.72 | -1.75 |