MRCR
text
+
+
+
+
About
MRCR (Multi-hop Reasoning for Conversational Recommendation) is a benchmark designed to evaluate models' ability to perform multi-step reasoning in conversational recommendation scenarios. It tests systems' capability to understand user preferences, conduct multi-hop reasoning across knowledge graphs, and provide accurate recommendations through natural dialogue interactions.
+
+
+
+
Evaluation Stats
Total Models6
Organizations1
Verified Results0
Self-Reported6
+
+
+
+
Benchmark Details
Max Score1
Language
en
+
+
+
+
Performance Overview
Score distribution and top performers
Score Distribution
6 models
Top Score
93.0%
Average Score
67.2%
High Performers (80%+)
2Top Organizations
#1Google
6 models
67.2%
+
+
+
+
Leaderboard
6 models ranked by performance on MRCR
| License | Links | ||||
|---|---|---|---|---|---|
| May 20, 2025 | Proprietary | 93.0% | |||
| May 1, 2024 | Proprietary | 82.6% | |||
| May 1, 2024 | Proprietary | 71.9% | |||
| Dec 1, 2024 | Proprietary | 69.2% | |||
| Mar 15, 2024 | Proprietary | 54.7% | |||
| May 20, 2025 | Proprietary | 32.0% |
