JiaqiXue commited on
Commit
771fd80
·
verified ·
1 Parent(s): 9e0e244

docs: note sub_10 only training following chayan

Browse files
Files changed (1) hide show
  1. README.md +3 -1
README.md CHANGED
@@ -185,7 +185,9 @@ checkpoints/
185
 
186
  ## Training Details
187
 
188
- - **Training Data**: RouterArena sub_10 split (809 queries, 10% of full 8,400)
 
 
189
  - **Method**: Nearest-neighbor regression with cosine distance, distance-weighted
190
  - **Evaluation**: Full 8,400 RouterArena queries (no data leakage)
191
  - **Training Time**: < 1 second
 
185
 
186
  ## Training Details
187
 
188
+ Following [chayan](https://huggingface.co/adaptive-classifier/chayan), we only use the official **sub_10 split** (809 queries, 10% of the full 8,400) for training. No full-set data is used during training or hyperparameter tuning.
189
+
190
+ - **Training Data**: RouterArena sub_10 split (809 queries)
191
  - **Method**: Nearest-neighbor regression with cosine distance, distance-weighted
192
  - **Evaluation**: Full 8,400 RouterArena queries (no data leakage)
193
  - **Training Time**: < 1 second