Datasets:
Tasks:
Question Answering
Modalities:
Text
Formats:
parquet
Languages:
English
Size:
10M - 100M
DOI:
spacemanidol
commited on
Commit
•
c9db5fd
1
Parent(s):
3c5b4f9
Update README.md
Browse files
README.md
CHANGED
@@ -26,6 +26,42 @@ configs:
|
|
26 |
This dataset contains the embeddings for the MSMARCO-V2.1 dataset which is used as the corpora for [TREC RAG](https://trec-rag.github.io/)
|
27 |
All embeddings are created using [Snowflake's Arctic Embed L](https://huggingface.co/Snowflake/snowflake-arctic-embed-l) and are intended to serve as a simple baseline for dense retrieval-based methods.
|
28 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
29 |
## Loading the dataset
|
30 |
|
31 |
### Loading the document embeddings
|
|
|
26 |
This dataset contains the embeddings for the MSMARCO-V2.1 dataset which is used as the corpora for [TREC RAG](https://trec-rag.github.io/)
|
27 |
All embeddings are created using [Snowflake's Arctic Embed L](https://huggingface.co/Snowflake/snowflake-arctic-embed-l) and are intended to serve as a simple baseline for dense retrieval-based methods.
|
28 |
|
29 |
+
|
30 |
+
## Retrieval Performance
|
31 |
+
Retrieval performance for the TREC DL21-23, MSMARCOV2-Dev and Raggy Queries can be found below with BM25 as a baseline. For both systems retrieval is at the segment level and Doc Score = Max (passage score).
|
32 |
+
Retrieval is done via dot product and happens in BF16.
|
33 |
+
### NDCG@10
|
34 |
+
| Dataset | BM25 | Snowflake Arctic Embed L |
|
35 |
+
|---|---|---|
|
36 |
+
| Deep Learning 2021 | 0.5778 | 0.70682 |
|
37 |
+
| Deep Learning 2022 | 0.3576 | 0.5444 |
|
38 |
+
| Deep Learning 2023 | 0.3356 | 0.47372 |
|
39 |
+
| msmarcov2-dev | N/A | 0.35844 |
|
40 |
+
| msmarcov2-dev2 | N/A | 0.35821 |
|
41 |
+
| Raggy Queries | 0.4227 | 0.57759 |
|
42 |
+
|
43 |
+
### Recall@100
|
44 |
+
| Dataset | BM25 | Snowflake Arctic Embed L |
|
45 |
+
|---|---|---|
|
46 |
+
| Deep Learning 2021 | 0.3811 | 0.41361 |
|
47 |
+
| Deep Learning 2022 | 0.233 | 0.31351 |
|
48 |
+
| Deep Learning 2023 | 0.3049 | 0.34793 |
|
49 |
+
| msmarcov2-dev | 0.6683 | 0.85131 |
|
50 |
+
| msmarcov2-dev2 | 0.6771 | 0.84767 |
|
51 |
+
| Raggy Queries | 0.2807 | 0.36228 |
|
52 |
+
|
53 |
+
|
54 |
+
### Recall@1000
|
55 |
+
| Dataset | BM25 | Snowflake Arctic Embed L |
|
56 |
+
|---|---|---|
|
57 |
+
| Deep Learning 2021 | 0.7115 | 0.7193 |
|
58 |
+
| Deep Learning 2022 | 0.479 | 0.54566 |
|
59 |
+
| Deep Learning 2023 | 0.5852 | 0.59577 |
|
60 |
+
| msmarcov2-dev | 0.8528 | 0.93966 |
|
61 |
+
| msmarcov2-dev2 | 0.8577 | 0.93947 |
|
62 |
+
| Raggy Queries | 0.5745 | 0.63092 |
|
63 |
+
|
64 |
+
|
65 |
## Loading the dataset
|
66 |
|
67 |
### Loading the document embeddings
|