readme
Browse files
README.md
CHANGED
@@ -72,13 +72,12 @@ The format is the following:
|
|
72 |
- "top1k_offsets": Passage ID (int) when the numpy matrices are loaded sequentially and vertically stacked
|
73 |
- "top1k_passage_ids": Passage ID (string) as they appear in the dataset
|
74 |
- "top1k_cossim": Cosine similarities
|
75 |
-
- "qrels": Relevance annotations for the 215 annotated queries by NIST
|
76 |
|
77 |
### Queries - JSONL
|
78 |
The folder `queries_jsonl/` contains the queries in a `.jsonl.gz` format.
|
79 |
|
80 |
-
|
81 |
-
|
82 |
|
83 |
### Queries - Parquet
|
84 |
|
|
|
72 |
- "top1k_offsets": Passage ID (int) when the numpy matrices are loaded sequentially and vertically stacked
|
73 |
- "top1k_passage_ids": Passage ID (string) as they appear in the dataset
|
74 |
- "top1k_cossim": Cosine similarities
|
75 |
+
- "qrels": Relevance annotations for the 215 annotated queries by NIST. The **document relevance** scores are provided. You can get the doc_id for a passage via `row['_id'].split("#")[0]`
|
76 |
|
77 |
### Queries - JSONL
|
78 |
The folder `queries_jsonl/` contains the queries in a `.jsonl.gz` format.
|
79 |
|
80 |
+
Note: qrels are provided here as a dictionary lookup, while in the parquet format as a list in the format `[doc_id, score]` due to the limited support for dictionaries in parquet.
|
|
|
81 |
|
82 |
### Queries - Parquet
|
83 |
|