Recommendation on MovieLens 20M (Accuracy/Recall@K)
67.6AccuracyMixRAGRec
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| MixRAGRecBackbone=LLaMA3-8B2026.05 | 67.6 | 90 | 94 | |
| MixRAGRecBackbone=Mistral-7B2026.05 | 65.2 | 89.2 | 93.5 | |
| Rec-r1Backbone=LLaMA3-8B2026.05 | 59.4 | 76.1 | 81.5 | |
| K-RagRecBackbone=LLaMA3-8B2026.05 | 57.8 | 72 | 75.5 | |
| K-RagRecBackbone=Mistral-7B2026.05 | 57.3 | 80.1 | 88.8 | |
| TallRecBackbone=Mistral-7B2026.05 | 54.5 | 80.9 | 88.3 | |
| G-retrieverBackbone=Mistral-7B2026.05 | 54.2 | 75.8 | 84.6 | |
| GPT-4oEvaluation Protocol=Zero-shot Inference2026.05 | 52.4 | — | — | |
| TallRecBackbone=LLaMA3-8B2026.05 | 52.1 | 78.1 | 86.8 | |
| Rec-r1Backbone=Mistral-7B2026.05 | 50.5 | 59.9 | 64.8 | |
| G-retrieverBackbone=LLaMA3-8B2026.05 | 49.1 | 72.2 | 80.1 | |
| Gemini-2.0-FlashEvaluation Protocol=Zero-shot Inference2026.05 | 46.7 | — | — | |
| Deepseek-r1Evaluation Protocol=Zero-shot Inference2026.05 | 44.5 | — | — | |
| KG-TextBackbone=LLaMA3-8B2026.05 | 30.1 | — | — | |
| KG-TextBackbone=Mistral-7B2026.05 | 27.7 | — | — | |
| KAPINGBackbone=LLaMA3-8B2026.05 | 27.4 | — | — | |
| KAPINGBackbone=Mistral-7B2026.05 | 27.2 | — | — | |
| LLaMA3-8BBackbone=LLaMA3-8B, Evaluation Protocol=Zero-shot Inference2026.05 | 12.9 | — | — | |
| Mistral-7BBackbone=Mistral-7B, Evaluation Protocol=Zero-shot Inference2026.05 | 9.7 | — | — |