Update README.md
Browse filesAdded description for creation of testset
README.md
CHANGED
|
@@ -18,6 +18,7 @@ For this purpose, we translated the [MSMARCO-Passage-Ranking](https://github.com
|
|
| 18 |
|
| 19 |
We evaluated our model on the [GermanDPR testset](https://deepset.ai/germanquad) and followed the benchmark framework of [BEIR](https://github.com/UKPLab/beir).
|
| 20 |
In order to compare our results, we conducted an evaluation on the same test data with BM25 and presented the results in the table below.
|
|
|
|
| 21 |
|
| 22 |
|
| 23 |
| Model | NDCG@1 | NDCG@5 | NDCG@10 | Recall@1 | Recall@5 | Recall@10 |
|
|
|
|
| 18 |
|
| 19 |
We evaluated our model on the [GermanDPR testset](https://deepset.ai/germanquad) and followed the benchmark framework of [BEIR](https://github.com/UKPLab/beir).
|
| 20 |
In order to compare our results, we conducted an evaluation on the same test data with BM25 and presented the results in the table below.
|
| 21 |
+
We took every paragraph with negative and positive context out of the testset and deduplicated them. The resulting corpus size is 2871 against 1025 queries.
|
| 22 |
|
| 23 |
|
| 24 |
| Model | NDCG@1 | NDCG@5 | NDCG@10 | Recall@1 | Recall@5 | Recall@10 |
|