update readme
Browse files
README.md
CHANGED
|
@@ -1 +1,14 @@
|
|
| 1 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
## Dataset Summary
|
| 2 |
+
|
| 3 |
+
This repo provides pre-built indexes for MSMARCO-v2.1 Doc Segmented corpus. All indexes are created through Anserini with different encoders (i.e., the lucene-inverted.msmarco-v2.1-doc-segmented.splade-v3.20250707.4039c3.tar.gz is encoded with the SPLADE-v3 model). This dataset is less than complete and will be updated over time to migrate other pre-built MSMARCO-v2.1 Doc Segmented indexes here.
|
| 4 |
+
|
| 5 |
+
## Load Dataset
|
| 6 |
+
An example to load the dataset:
|
| 7 |
+
```python
|
| 8 |
+
dataset = load_dataset("castorini/prebuilt-indexes-msmarco-v2.1-doc-segmented", data_files="lucene-inverted.msmarco-v2.1-doc-segmented.splade-v3.20250707.4039c3.tar.gz")
|
| 9 |
+
```
|
| 10 |
+
|
| 11 |
+
An example URL for downloading a pre-built index:
|
| 12 |
+
```
|
| 13 |
+
https://huggingface.co/datasets/castorini/prebuilt-indexes-msmarco-v2.1-doc-segmented/resolve/main/lucene-inverted.msmarco-v2.1-doc-segmented.splade-v3.20250707.4039c3.tar.gz
|
| 14 |
+
```
|