Downstream Task Accuracy via Paradigm Routing on Downstream Tasks (test)
73.4AccuracyOracle
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| OracleRouter Category=Oracle, Model=Gemini2026.04 | 73.4 | — | |
| OracleRouter Category=Oracle, Model=GPT-52026.04 | 72.9 | — | |
| OracleRouter Category=Oracle, Model=Qwen3-Max2026.04 | 72.5 | — | |
| Embedding (LR)Router Category=Emb, Model=GPT-52026.04 | 64.2 | — | |
| Combined (LR)Router Category=Emb, Model=GPT-52026.04 | 64.2 | — | |
| Best-singleRouter Category=Base, Model=GPT-52026.04 | 62.4 | — | |
| Handcrafted (MLP)Router Category=HC, Model=GPT-52026.04 | 62.4 | — | |
| Combined (MLP)Router Category=Emb, Model=GPT-52026.04 | 62.4 | — | |
| Combined (MLP)Router Category=Emb, Model=Gemini2026.04 | 62.2 | — | |
| Embedding (LR)Router Category=Emb, Model=Gemini2026.04 | 61.9 | — | |
| Combined (LR)Router Category=Emb, Model=Gemini2026.04 | 61 | — | |
| DirectRouter Category=Base, Model=GPT-52026.04 | 60.3 | — | |
| Self-routeRouter Category=ZS, Model=GPT-52026.04 | 60.3 | — | |
| Handcrafted (LR)Router Category=HC, Model=GPT-52026.04 | 59.4 | — | |
| Handcrafted (LR)Router Category=HC, Model=Gemini2026.04 | 58.5 | — | |
| Self-routeRouter Category=ZS, Model=Gemini2026.04 | 56.8 | — | |
| OracleRouter Category=Oracle, Model=Qwen3-30B2026.04 | 56.8 | — | |
| DirectRouter Category=Base, Model=Gemini2026.04 | 55.5 | — | |
| Best-singleRouter Category=Base, Model=Gemini2026.04 | 55.5 | — | |
| Handcrafted (MLP)Router Category=HC, Model=Gemini2026.04 | 55.5 | — | |
| Combined (LR)Router Category=Emb, Model=Qwen3-Max2026.04 | 54.6 | — | |
| Handcrafted (MLP)Router Category=HC, Model=Qwen3-Max2026.04 | 53.3 | — | |
| Embedding (LR)Router Category=Emb, Model=Qwen3-Max2026.04 | 53.3 | — | |
| Combined (MLP)Router Category=Emb, Model=Qwen3-Max2026.04 | 53.3 | — | |
| Handcrafted (LR)Router Category=HC, Model=Qwen3-Max2026.04 | 51.5 | — | |
| Best-singleRouter Category=Base, Model=Qwen3-Max2026.04 | 50.7 | — | |
| DirectRouter Category=Base, Model=Qwen3-Max2026.04 | 49.8 | — | |
| Self-routeRouter Category=ZS, Model=Qwen3-Max2026.04 | 42.4 | — | |
| Handcrafted (MLP)Router Category=HC, Model=Qwen3-30B2026.04 | 37.1 | — | |
| Handcrafted (LR)Router Category=HC, Model=Qwen3-30B2026.04 | 34.9 | — | |
| Combined (MLP)Router Category=Emb, Model=Qwen3-30B2026.04 | 33.6 | — | |
| Best-singleRouter Category=Base, Model=Qwen3-30B2026.04 | 32.8 | — | |
| Combined (LR)Router Category=Emb, Model=Qwen3-30B2026.04 | 32.8 | — | |
| Embedding (LR)Router Category=Emb, Model=Qwen3-30B2026.04 | 32.3 | — | |
| Self-routeRouter Category=ZS, Model=Qwen3-30B2026.04 | 27.5 | — | |
| DirectRouter Category=Base, Model=Qwen3-30B2026.04 | 24.9 | — | |
| Best-singleRouter Category=Base2026.04 | — | 50.3 | |
| Combined (LR)Router Category=Emb2026.04 | — | 53.1 | |
| Combined (MLP)Router Category=Emb2026.04 | — | 52.9 | |
| DirectRouter Category=Base2026.04 | — | 47.6 | |
| Embedding (LR)Router Category=Emb2026.04 | — | 52.9 | |
| Handcrafted (LR)Router Category=HC2026.04 | — | 51.1 | |
| Handcrafted (MLP)Router Category=HC2026.04 | — | 52.1 | |
| OracleRouter Category=Oracle2026.04 | — | 68.9 | |
| Self-routeRouter Category=ZS2026.04 | — | 46.7 |