Update README.md

README.md

@@ -113,8 +113,8 @@ for i, definition in enumerate(definitions):
 - **Max Sequence Length:** 256 tokens (both mentions and descriptions)
 
 ### Benchmark Results
-- **Dataset:** Zero-Shot Entity Linking [(Logeswaran et al., 2019)](https://arxiv.org/abs/1906.07348) test set. More details [here](https://github.com/lajanugen/zeshel/blob/master/README.md)
-- **Metric:** Recall@64
+- **Dataset:** Zero-Shot Entity Linking [(Logeswaran et al., 2019)](https://arxiv.org/abs/1906.07348) test set.
+- **Metric:** Recall@64 (Average performance across the set of test worlds was computed by macroaveraging - we followed the same pattern as in [Meta AI's BLINK paper](https://arxiv.org/pdf/1911.03814))
 - **Score:** 80.29%
 - **Comparison:** [Meta AI's BLINK](https://arxiv.org/pdf/1911.03814) achieves 82.06% on the same test set - slightly higher than ours, however, their model was trained on the training set but ours was not.
 - **Conclusion:** Our model has strong zero-shot performance