nreimers commited on
Commit
f387ca9
1 Parent(s): 33dbf06
Files changed (1) hide show
  1. README.md +45 -18
README.md CHANGED
@@ -1,34 +1,61 @@
1
  # Cross-Encoder for MS Marco
2
 
3
- This model uses [Electra-base](https://huggingface.co/google/electra-base-discriminator).
4
 
5
- It was trained on [MS Marco Passage Ranking](https://github.com/microsoft/MSMARCO-Passage-Ranking) task.
6
 
7
- The model can be used for Information Retrieval: Given a query, encode the query will all possible passages (e.g. retrieved with ElasticSearch). Then sort the passages in a decreasing order. See [SBERT.net Information Retrieval](https://github.com/UKPLab/sentence-transformers/tree/master/examples/applications/information-retrieval) for more details. The training code is available here: [SBERT.net Training MS Marco](https://github.com/UKPLab/sentence-transformers/tree/master/examples/training/ms_marco)
8
 
9
- ## Usage and Performance
10
 
11
- Pre-trained models can be used like this:
 
 
 
 
 
 
 
 
 
 
 
 
12
  ```
 
 
 
 
 
 
13
  from sentence_transformers import CrossEncoder
14
  model = CrossEncoder('model_name', max_length=512)
15
  scores = model.predict([('Query', 'Paragraph1'), ('Query', 'Paragraph2') , ('Query', 'Paragraph3')])
16
  ```
17
 
 
 
18
  In the following table, we provide various pre-trained Cross-Encoders together with their performance on the [TREC Deep Learning 2019](https://microsoft.github.io/TREC-2019-Deep-Learning/) and the [MS Marco Passage Reranking](https://github.com/microsoft/MSMARCO-Passage-Ranking/) dataset.
19
 
20
 
21
- | Model-Name | NDCG@10 (TREC DL 19) | MRR@10 (MS Marco Dev) | Docs / Sec (BertTokenizerFast) | Docs / Sec |
22
- | ------------- |:-------------| -----| --- | --- |
23
- | cross-encoder/ms-marco-TinyBERT-L-2 | 67.43 | 30.15 | 9000 | 780
24
- | cross-encoder/ms-marco-TinyBERT-L-4 | 68.09 | 34.50 | 2900 | 760
25
- | cross-encoder/ms-marco-TinyBERT-L-6 | 69.57 | 36.13 | 680 | 660
26
- | cross-encoder/ms-marco-electra-base | 71.99 | 36.41 | 340 | 340
27
- | *Other models* | | | |
28
- | nboost/pt-tinybert-msmarco | 63.63 | 28.80 | 2900 | 760
29
- | nboost/pt-bert-base-uncased-msmarco | 70.94 | 34.75 | 340 | 340|
30
- | nboost/pt-bert-large-msmarco | 73.36 | 36.48 | 100 | 100 |
31
- | Capreolus/electra-base-msmarco | 71.23 | | 340 | 340 |
32
- | amberoad/bert-multilingual-passage-reranking-msmarco | 68.40 | | 330 | 330
 
 
 
 
 
 
 
 
33
 
34
- Note: Runtime was computed on a V100 GPU. A bottleneck for smaller models is the standard Python tokenizer from Huggingface v3. Replacing it with the fast tokenizer based on Rust, the throughput is significantly improved:
 
1
  # Cross-Encoder for MS Marco
2
 
3
+ This model was trained on the [MS Marco Passage Ranking](https://github.com/microsoft/MSMARCO-Passage-Ranking) task.
4
 
5
+ The model can be used for Information Retrieval: Given a query, encode the query will all possible passages (e.g. retrieved with ElasticSearch). Then sort the passages in a decreasing order. See [SBERT.net Retrieve & Re-rank](https://www.sbert.net/examples/applications/retrieve_rerank/README.html) for more details. The training code is available here: [SBERT.net Training MS Marco](https://github.com/UKPLab/sentence-transformers/tree/master/examples/training/ms_marco)
6
 
 
7
 
8
+ ## Usage with Transformers
9
 
10
+ ```python
11
+ from transformers import AutoTokenizer, AutoModelForSequenceClassification
12
+ import torch
13
+
14
+ model = AutoModelForSequenceClassification.from_pretrained('model_name')
15
+ tokenizer = AutoTokenizer.from_pretrained('model_name')
16
+
17
+ features = tokenizer(['How many people live in Berlin?', 'How many people live in Berlin?'], ['Berlin has a population of 3,520,031 registered inhabitants in an area of 891.82 square kilometers.', 'New York City is famous for the Metropolitan Museum of Art.'], padding=True, truncation=True, return_tensors="pt")
18
+
19
+ model.eval()
20
+ with torch.no_grad():
21
+ scores = model(**features).logits
22
+ print(scores)
23
  ```
24
+
25
+
26
+ ## Usage with SentenceTransformers
27
+
28
+ The usage becomes easier when you have [SentenceTransformers](https://www.sbert.net/) installed. Then, you can use the pre-trained models like this:
29
+ ```python
30
  from sentence_transformers import CrossEncoder
31
  model = CrossEncoder('model_name', max_length=512)
32
  scores = model.predict([('Query', 'Paragraph1'), ('Query', 'Paragraph2') , ('Query', 'Paragraph3')])
33
  ```
34
 
35
+
36
+ ## Performance
37
  In the following table, we provide various pre-trained Cross-Encoders together with their performance on the [TREC Deep Learning 2019](https://microsoft.github.io/TREC-2019-Deep-Learning/) and the [MS Marco Passage Reranking](https://github.com/microsoft/MSMARCO-Passage-Ranking/) dataset.
38
 
39
 
40
+ | Model-Name | NDCG@10 (TREC DL 19) | MRR@10 (MS Marco Dev) | Docs / Sec |
41
+ | ------------- |:-------------| -----| --- |
42
+ | **Version 2 models** | | |
43
+ | cross-encoder/ms-marco-TinyBERT-L-2-v2 | 69.84 | 32.56 | 9000
44
+ | cross-encoder/ms-marco-MiniLM-L-2-v2 | 71.01 | 34.85 | 4100
45
+ | cross-encoder/ms-marco-MiniLM-L-4-v2 | 73.04 | 37.70 | 2500
46
+ | cross-encoder/ms-marco-MiniLM-L-6-v2 | 74.30 | 39.01 | 1800
47
+ | cross-encoder/ms-marco-MiniLM-L-12-v2 | 74.31 | 39.02 | 960
48
+ | **Version 1 models** | | |
49
+ | cross-encoder/ms-marco-TinyBERT-L-2 | 67.43 | 30.15 | 9000
50
+ | cross-encoder/ms-marco-TinyBERT-L-4 | 68.09 | 34.50 | 2900
51
+ | cross-encoder/ms-marco-TinyBERT-L-6 | 69.57 | 36.13 | 680
52
+ | cross-encoder/ms-marco-electra-base | 71.99 | 36.41 | 340
53
+ | **Other models** | | |
54
+ | nboost/pt-tinybert-msmarco | 63.63 | 28.80 | 2900
55
+ | nboost/pt-bert-base-uncased-msmarco | 70.94 | 34.75 | 340
56
+ | nboost/pt-bert-large-msmarco | 73.36 | 36.48 | 100
57
+ | Capreolus/electra-base-msmarco | 71.23 | 36.89 | 340
58
+ | amberoad/bert-multilingual-passage-reranking-msmarco | 68.40 | 35.54 | 330
59
+ | sebastian-hofstaetter/distilbert-cat-margin_mse-T2-msmarco | 72.82 | 37.88 | 720
60
 
61
+ Note: Runtime was computed on a V100 GPU.