[update] readme
Browse files
README.md
CHANGED
|
@@ -1,3 +1,19 @@
|
|
| 1 |
-
---
|
| 2 |
-
license: apache-2.0
|
| 3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
datasets:
|
| 4 |
+
- nics-efc/R2R_Router_Training
|
| 5 |
+
- nics-efc/R2R_query
|
| 6 |
+
language:
|
| 7 |
+
- en
|
| 8 |
+
metrics:
|
| 9 |
+
- accuracy
|
| 10 |
+
base_model:
|
| 11 |
+
- deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B
|
| 12 |
+
- deepseek-ai/DeepSeek-R1-Distill-Qwen-32B
|
| 13 |
+
---
|
| 14 |
+
|
| 15 |
+
This is the default router from the paper [R2R: Efficiently Navigating Divergent Reasoning Paths with Small-Large Model Token Routing](https://arxiv.org/abs/2505.21600).
|
| 16 |
+
|
| 17 |
+
Roads to Rome (R2R) is a neural token router that efficiently combines Large Language Models (LLMs) and Small Language Models (SLMs) by selectively routing only critical, reasoning-divergent tokens to the large model.
|
| 18 |
+
|
| 19 |
+
Please visit our [GitHub repo](https://github.com/thu-nics/R2R) for more information.
|