Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
distilbert-base-uncased trained on MSMARCO Document Reranking task,
|
2 |
+
|
3 |
+
#### usage
|
4 |
+
```
|
5 |
+
from transformers import AutoTokenizer, AutoModelForSequenceClassification
|
6 |
+
|
7 |
+
tokenizer = AutoTokenizer.from_pretrained('brutusxu/distilbert-base-cross-encoder-first-p')
|
8 |
+
model = AutoModelForSequenceClassification.from_pretrained('brutusxu/distilbert-base-cross-encoder-first-p')
|
9 |
+
|
10 |
+
query = 'I love New York'
|
11 |
+
document = 'I like New York'
|
12 |
+
input = '<P>' + query + tokenizer.sep_token + '<Q>' + document
|
13 |
+
tokenized_input = tokenizer(input, return_tensors='pt')
|
14 |
+
ranking_score = model(**tokenized_input)
|
15 |
+
```
|
16 |
+
|
17 |
+
#### performance
|
18 |
+
on MSMARCO Document Reranking w. top-100 documents from BM25
|
19 |
+
```
|
20 |
+
MRR@10: 0.373
|
21 |
+
MRR@100: 0.381
|
22 |
+
nDCG@10: 0.442
|
23 |
+
nDCG@10: 0.475
|
24 |
+
```
|