File size: 9,191 Bytes
3bb4be0
 
 
 
 
 
177b1cc
 
 
 
 
 
 
 
 
 
 
 
1fb33ce
 
 
 
3bb4be0
 
267c6c2
 
3bb4be0
 
 
 
 
 
 
 
 
267c6c2
3bb4be0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
267c6c2
3bb4be0
267c6c2
3bb4be0
 
 
 
 
 
 
 
 
74f67cf
 
3bb4be0
 
 
74f67cf
3bb4be0
 
1fb33ce
 
3bb4be0
 
 
 
 
1fb33ce
3bb4be0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
267c6c2
3bb4be0
267c6c2
3bb4be0
 
 
 
 
 
 
 
 
 
 
 
 
084e962
3bb4be0
 
084e962
3bb4be0
084e962
3bb4be0
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
---
license: apache-2.0
datasets:
- allenai/scirepeval
---

## SPECTER2

<!-- Provide a quick summary of what the model is/does. -->

SPECTER2 is a family of models that succeeds [SPECTER](https://huggingface.co/allenai/specter) and is capable of generating task specific embeddings for scientific tasks when paired with [adapters](https://huggingface.co/models?search=allenai/specter-2_).
This is the base encoder to be used with relevant task specific adapters.
Given the combination of title and abstract of a scientific paper or a short texual query, the model can be used to generate effective embeddings to be used in downstream applications.

**Note:For general embedding purposes, please use [allenai/specter2](https://huggingface.co/allenai/specter2).**

**To get the best performance on a downstream task type please load the associated adapter () with the base model as in the example below.**

**Dec 2023 Update:**

Model usage updated to be compatible with latest versions of transformers and adapters (newly released update to adapter-transformers) libraries.

**\*\*\*\*\*\*Update\*\*\*\*\*\***

This update introduces a new set of SPECTER2 models with the base transformer encoder pre-trained on an extended citation dataset containing more recent papers.
For benchmarking purposes please use the existing SPECTER2 [models](https://huggingface.co/allenai/specter2) w/o the **aug2023refresh** suffix.

**Note:For general embedding purposes, please use [allenai/specter2](https://huggingface.co/allenai/specter2).**

**To get the best performance on a downstream task type please load the associated adapter with the base model as in the example below.**

# Model Details

## Model Description

SPECTER2 has been trained on over 6M triplets of scientific paper citations, which are available [here](https://huggingface.co/datasets/allenai/scirepeval/viewer/cite_prediction_new/evaluation).
Post that it is trained with additionally attached task format specific adapter modules on all the [SciRepEval](https://huggingface.co/datasets/allenai/scirepeval) training tasks.

Task Formats trained on:
- Classification
- Regression
- Proximity
- Adhoc Search

  
It builds on the work done in [SciRepEval: A Multi-Format Benchmark for Scientific Document Representations](https://api.semanticscholar.org/CorpusID:254018137) and we evaluate the trained model on this benchmark as well.



- **Developed by:** Amanpreet Singh, Mike D'Arcy, Arman Cohan, Doug Downey, Sergey Feldman
- **Shared by :** Allen AI
- **Model type:** bert-base-uncased + adapters
- **License:** Apache 2.0
- **Finetuned from model:** [allenai/scibert](https://huggingface.co/allenai/scibert_scivocab_uncased).

## Model Sources

<!-- Provide the basic links for the model. -->

- **Repository:** [https://github.com/allenai/SPECTER2](https://github.com/allenai/SPECTER2)
- **Paper:** [https://api.semanticscholar.org/CorpusID:254018137](https://api.semanticscholar.org/CorpusID:254018137)
- **Demo:** [Usage](https://github.com/allenai/SPECTER2/blob/main/README.md)

# Uses

<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->

## Direct Use

|Model|Name and HF link|Description|
|--|--|--|
|Proximity*|[allenai/specter2_aug2023refresh](https://huggingface.co/allenai/specter2_aug2023refresh)|Encode papers as queries and candidates eg. Link Prediction, Nearest Neighbor Search|
|Adhoc Query|[allenai/specter2_aug2023refresh_adhoc_query](https://huggingface.co/allenai/specter2_aug2023refresh_adhoc_query)|Encode short raw text queries for search tasks. (Candidate papers can be encoded with the proximity adapter)|
|Classification|[allenai/specter2_aug2023refresh_classification](https://huggingface.co/allenai/specter2_aug2023refresh_classification)|Encode papers to feed into linear classifiers as features|
|Regression|[allenai/specter2_aug2023refresh_regression](https://huggingface.co/allenai/specter2_aug2023refresh_regression)|Encode papers to feed into linear regressors as features|

*Proximity model should suffice for downstream task types not mentioned above

```python
from transformers import AutoTokenizer
from adapters import AutoAdapterModel

# load model and tokenizer
tokenizer = AutoTokenizer.from_pretrained('allenai/specter2_aug2023refresh_base')

#load base model
model = AutoAdapterModel.from_pretrained('allenai/specter2_aug2023refresh_base')

#load the adapter(s) as per the required task, provide an identifier for the adapter in load_as argument and activate it
model.load_adapter("allenai/specter2_aug2023refresh", source="hf", load_as="proximity", set_active=True)
#other possibilities: allenai/specter2_aug2023refresh_<classification|regression|adhoc_query>

papers = [{'title': 'BERT', 'abstract': 'We introduce a new language representation model called BERT'},
          {'title': 'Attention is all you need', 'abstract': ' The dominant sequence transduction models are based on complex recurrent or convolutional neural networks'}]

# concatenate title and abstract
text_batch = [d['title'] + tokenizer.sep_token + (d.get('abstract') or '') for d in papers]
# preprocess the input
inputs = self.tokenizer(text_batch, padding=True, truncation=True,
                                   return_tensors="pt", return_token_type_ids=False, max_length=512)
output = model(**inputs)
# take the first token in the batch as the embedding
embeddings = output.last_hidden_state[:, 0, :]
```

## Downstream Use

<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->

For evaluation and downstream usage, please refer to [https://github.com/allenai/scirepeval/blob/main/evaluation/INFERENCE.md](https://github.com/allenai/scirepeval/blob/main/evaluation/INFERENCE.md).

# Training Details

## Training Data

<!-- This should link to a Data Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->

The base model is trained on citation links between papers and the adapters are trained on 8 large scale tasks across the four formats.
All the data is a part of SciRepEval benchmark and is available [here](https://huggingface.co/datasets/allenai/scirepeval).

The citation link are triplets in the form 

```json
{"query": {"title": ..., "abstract": ...}, "pos": {"title": ..., "abstract": ...}, "neg": {"title": ..., "abstract": ...}}
```

consisting of a query paper, a positive citation and a negative which can be from the same/different field of study as the query or citation of a citation.

## Training Procedure 

Please refer to the [SPECTER paper](https://api.semanticscholar.org/CorpusID:215768677).


### Training Hyperparameters


The model is trained in two stages using [SciRepEval](https://github.com/allenai/scirepeval/blob/main/training/TRAINING.md):
- Base Model: First a base model is trained on the above citation triplets.
``` batch size = 1024, max input length = 512, learning rate = 2e-5, epochs = 2 warmup steps = 10% fp16```
- Adapters: Thereafter, task format specific adapters are trained on the SciRepEval training tasks, where 600K triplets are sampled from above and added to the training data as well.
``` batch size = 256, max input length = 512, learning rate = 1e-4, epochs = 6 warmup = 1000 steps fp16```


# Evaluation

We evaluate the model on [SciRepEval](https://github.com/allenai/scirepeval), a large scale eval benchmark for scientific embedding tasks which which has [SciDocs] as a subset.
We also evaluate and establish a new SoTA on [MDCR](https://github.com/zoranmedic/mdcr), a large scale citation recommendation benchmark.

|Model|SciRepEval In-Train|SciRepEval Out-of-Train|SciRepEval Avg|MDCR(MAP, Recall@5)|
|--|--|--|--|--|
|[BM-25](https://api.semanticscholar.org/CorpusID:252199740)|n/a|n/a|n/a|(33.7, 28.5)|
|[SPECTER](https://huggingface.co/allenai/specter)|54.7|57.4|68.0|(30.6, 25.5)|
|[SciNCL](https://huggingface.co/malteos/scincl)|55.6|57.8|69.0|(32.6, 27.3)|
|[SciRepEval-Adapters](https://huggingface.co/models?search=scirepeval)|61.9|59.0|70.9|(35.3, 29.6)|
|[SPECTER2-Adapters](https://huggingface.co/models?search=allenai/specter-2)|**62.3**|**59.2**|**71.2**|**(38.4, 33.0)**|

Please cite the following works if you end up using SPECTER2:

[SPECTER paper](https://api.semanticscholar.org/CorpusID:215768677):  

```bibtex
@inproceedings{specter2020cohan,
  title={{SPECTER: Document-level Representation Learning using Citation-informed Transformers}},
  author={Arman Cohan and Sergey Feldman and Iz Beltagy and Doug Downey and Daniel S. Weld},
  booktitle={ACL},
  year={2020}
}
```
[SciRepEval paper](https://api.semanticscholar.org/CorpusID:254018137)
```bibtex
@inproceedings{Singh2022SciRepEvalAM,
  title={SciRepEval: A Multi-Format Benchmark for Scientific Document Representations},
  author={Amanpreet Singh and Mike D'Arcy and Arman Cohan and Doug Downey and Sergey Feldman},
  booktitle={Conference on Empirical Methods in Natural Language Processing},
  year={2022},
  url={https://api.semanticscholar.org/CorpusID:254018137}
}
```