File size: 11,438 Bytes
7cbc767
 
 
 
 
 
 
 
 
 
653fbd5
 
 
 
 
 
 
 
 
 
 
 
 
4e4582d
653fbd5
 
 
7bab2fa
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7cbc767
 
7bab2fa
7cbc767
7bab2fa
41fddea
7cbc767
7bab2fa
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7cbc767
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7bab2fa
7cbc767
7bab2fa
7cbc767
 
 
7bab2fa
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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

---
license: apache-2.0
library_name: span-marker
tags:
- span-marker
- token-classification
- ner
- named-entity-recognition
pipeline_tag: token-classification
widget:
- text: "the model is initially fit on a training dataset, the model (e.g. a neural net or a naive bayes classifier) is trained on the training dataset using a supervised learning method, for example using optimization methods such as gradient descent or stochastic gradient descent."
  example_title: "AI"
- text: "it restricted the barbarians' selectorial options but they still boast 13 internationals including england full-back tim stimpson and recalled wing tony underwood, plus all black forwards ian jones and norm hewitt."
  example_title: "CoNLL"
- text: "two decades after frank herbert's death, his son brian herbert, along with kevin j. anderson, published two sequels - hunters of dune (2006) and sandworms of dune (2007) - based on notes left behind by frank herbert for what he referred to as dune 7, his own planned seventh novel in the dune series."
  example_title: "Literature"
- text: "polka is still a popular genre of folk music in many european countries and is performed by folk artists in poland, latvia, lithuania, czech republic, netherlands, croatia, slovenia, germany, hungary, austria, switzerland, italy, ukraine, belarus, russia and slovakia."
  example_title: "Music 1"
- text: "as a strong advocate of animal rights, linda lent her support to many organizations such as people for the ethical treatment of animals (peta), the campaign to protect rural england, and friends of the earth."
  example_title: "Music 2"
- text: "some of the most pronounced effects of hellenization can be seen in afghanistan and india, in the region of the relatively late-rising greco-bactrian kingdom (250-125 bc) (in modern afghanistan, pakistan, and tajikistan) and the indo-greek kingdom (180 bc - 10 ad) in modern afghanistan and india and created a culture of greco-buddhist art."
  example_title: "Politics"
- text: "that first evening session was organized by jack yardley from johns hopkins university, and included henry appelman (university of michigan), harvey goldman (beth israel deaconess medical center and harvard medical school), bill hawk (the cleveland clinic), tom kent (university of iowa), si-chun ming (temple university), tom norris (university of washington), and robert riddell (university of chicago)."
  example_title: "Science 1"
- text: "viral tk phosphorylates aciclovir into its monophosphate form, which is subsequently phosphorylated to active aciclovir triphoshate by cellular kinases, thus selectively inhibiting viral dna polymerase."
  example_title: "Science 2"
model-index:
  - name: SpanMarker w. bert-base-uncased on CrossNER by Tom Aarsen
    results:
      - task:
          type: token-classification
          name: Named Entity Recognition
        dataset:
          type: P3ps/Cross_ner
          name: CrossNER
          split: test
          revision: 7cecbbb3d2eb8c75c8571c53e5a5270cfd0c5a9e
        metrics:
          - type: f1
            value: 0.8708
            name: F1
          - type: precision
            value: 0.8763
            name: Precision
          - type: recall
            value: 0.8654
            name: Recall
datasets:
  - P3ps/Cross_ner
language:
  - en
metrics:
  - f1
  - recall
  - precision
---

# SpanMarker for uncased Named Entity Recognition

This is a [SpanMarker](https://github.com/tomaarsen/SpanMarkerNER) model that can be used for Named Entity Recognition. In particular, this SpanMarker model uses [bert-base-uncased](https://huggingface.co/bert-base-uncased) as the underlying encoder. See [train.py](train.py) for the training script.
It is trained on [P3ps/Cross_ner](https://huggingface.co/datasets/P3ps/Cross_ner), which I believe is a variant of [DFKI-SLT/cross_ner](https://huggingface.co/datasets/DFKI-SLT/cross_ner) that merged the validation set into the training set and applied deduplication.

Is your data always capitalized correctly? Then consider using the cased variant of this model instead for better performance: 
[tomaarsen/span-marker-bert-base-cross-ner](https://huggingface.co/tomaarsen/span-marker-bert-base-cross-ner).

## Labels & Metrics

| **Label**          | **Examples** |  **Precision** |   **Recall** |    **F1** |
|:-------------------|-|------------:|---------:|------:|
| **all**            | - |       87.63 |    86.54 | 87.08 |
| academicjournal    | "new journal of physics", "epl", "european physical journal b" |       82.22 |    90.24 | 86.05 |
| album              | "tellin' stories", "generation terrorists", "country airs" |       84.46 |    84.46 | 84.46 |
| algorithm          | "lda", "pca", "gradient descent" |       82.86 |    76.99 | 79.82 |
| astronomicalobject | "earth", "sun", "halley's comet" |       88.61 |    94.59 | 91.50 |
| award              | "nobel prize for literature", "acamedy award for best actress", "mandelbrot's awards" |       87.76 |    91.63 | 89.66 |
| band               | "clash", "parliament funkadelic", "sly and the family stone" |       82.72 |    85.35 | 84.01 |
| book               | "nietzsche contra wagner" , "dionysian-dithyrambs", "the rebel" |       68.51 |    79.49 | 73.59 |
| chemicalcompound   | "hydrogen sulfide", "starch", "lactic acid" |       73.33 |    66.67 | 69.84 |
| chemicalelement    | "potassium", "fluorine", "chlorine" |       95.65 |    73.33 | 83.02 |
| conference         | "siggraph", "ijcai", "ieee transactions on speech and audio processing" |       72.41 |    60.00 | 65.62 |
| country            | "united arab emirates", "u.s.", "canada" |       81.03 |    86.08 | 83.48 |
| discipline         | "physics", "meteorology", "geography" |       35.48 |    40.74 | 37.93 |
| election           | "2004 canadian federal election", "2006 canadian federal election", "1999 scottish parliament election" |       96.22 |    98.28 | 97.24 |
| enzyme             | "rna polymerase", "phosphoinositide 3-kinase", "protein kinase c" |       72.09 |    83.78 | 77.50 |
| event              | "cannes film festival", "2019 special olympics world summer games", "2017 western iraq campaign" |       68.12 |    60.22 | 63.93 |
| field              | "computational imaging", "electronics", "information theory" |       92.13 |    77.36 | 84.10 |
| literarygenre      | "novel", "satire", "short story" |       65.26 |    72.09 | 68.51 |
| location           | "china", "bombay", "serbia" |       94.78 |    93.68 | 94.23 |
| magazine           | "the atlantic", "the american spectator", "astounding science fiction" |       60.71 |    60.71 | 60.71 |
| metrics            | "bleu", "precision", "dcg" |       77.01 |    82.72 | 79.76 |
| misc               | "serbian", "belgian", "the birth of a nation" |       80.11 |    72.12 | 75.91 |
| musicalartist      | "chuck burgi", "john miceli", "john o'reilly" |       78.84 |    84.44 | 81.55 |
| musicalinstrument  | "koto", "bubens", "def" |       75.00 |    33.33 | 46.15 |
| musicgenre         | "christian rock", "punk rock", "romantic melodicism" |       88.21 |    88.21 | 88.21 |
| organisation       | "irish times", "comintern", "wimbledon" |       89.17 |    89.98 | 89.57 |
| person             | "gong zhichao", "liu lufung", "margret crowley" |       95.87 |    92.65 | 94.23 |
| poem               | "historia destructionis troiae", "i am joaquin", "the snow man" |       94.29 |    64.71 | 76.74 |
| politicalparty     | "new democratic party", "bloc québécois", "liberal party of canada" |       87.16 |    84.50 | 85.81 |
| politician         | "susan kadis", "simon strelchik", "lloyd helferty" |       85.23 |    90.71 | 87.89 |
| product            | "alphago", "wordnet", "facial recognition system" |       63.95 |    65.48 | 64.71 |
| programlang        | "r", "c++", "java" |       75.00 |    84.38 | 79.41 |
| protein            | "dna methyltransferase", "tau protein", "amyloid beta" |       57.50 |    66.67 | 61.74 |
| researcher         | "sirovich", "kirby", "matthew turk" |       93.06 |    75.28 | 83.23 |
| scientist          | "matjaž perc", "cotton", "singer" |       80.27 |    93.72 | 86.47 |
| song               | "right where i'm supposed to be", "easy", "three times a lady" |       89.87 |    82.56 | 86.06 |
| task               | "robot control", "elevator scheduling", "telecommunications" |       73.86 |    75.58 | 74.71 |
| theory             | "big bang", "general theory of relativity", "ptolemaic planetary theories" |        0.00 |     0.00 |  0.00 |
| university         | "university of göttingen", "duke", "imperial academy of sciences" |       79.78 |    79.78 | 79.78 |
| writer             | "thomas mann", "george bernard shaw", "thomas hardy" |       77.78 |    86.19 | 81.77 |

## Usage

To use this model for inference, first install the `span_marker` library:

```bash
pip install span_marker
```

You can then run inference with this model like so:

```python
from span_marker import SpanMarkerModel

# Download from the 🤗 Hub
model = SpanMarkerModel.from_pretrained("tomaarsen/span-marker-bert-base-uncased-cross-ner")
# Run inference
entities = model.predict("amelia earhart flew her single engine lockheed vega 5b across the atlantic to paris.")
```

See the [SpanMarker](https://github.com/tomaarsen/SpanMarkerNER) repository for documentation and additional information on this library.

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 3

### Training results

| Training Loss | Epoch | Step | Validation Loss | Overall Precision | Overall Recall | Overall F1 | Overall Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:-----------------:|:--------------:|:----------:|:----------------:|
| 0.0641        | 0.25  | 200  | 0.0445          | 0.7141            | 0.5496         | 0.6212     | 0.8700           |
| 0.0268        | 0.5   | 400  | 0.0224          | 0.8171            | 0.7510         | 0.7827     | 0.9314           |
| 0.0213        | 0.76  | 600  | 0.0187          | 0.8387            | 0.8013         | 0.8196     | 0.9444           |
| 0.017         | 1.01  | 800  | 0.0162          | 0.8623            | 0.8231         | 0.8422     | 0.9497           |
| 0.0141        | 1.26  | 1000 | 0.0163          | 0.8571            | 0.8384         | 0.8477     | 0.9535           |
| 0.0132        | 1.51  | 1200 | 0.0149          | 0.8711            | 0.8470         | 0.8589     | 0.9563           |
| 0.0113        | 1.76  | 1400 | 0.0150          | 0.8603            | 0.8523         | 0.8563     | 0.9556           |
| 0.0097        | 2.02  | 1600 | 0.0150          | 0.8710            | 0.8553         | 0.8631     | 0.9573           |
| 0.0083        | 2.27  | 1800 | 0.0148          | 0.8809            | 0.8568         | 0.8687     | 0.9586           |
| 0.0075        | 2.52  | 2000 | 0.0150          | 0.8733            | 0.8573         | 0.8652     | 0.9583           |
| 0.0068        | 2.77  | 2200 | 0.0148          | 0.8745            | 0.8642         | 0.8693     | 0.9600           |


### Framework versions

- SpanMarker 1.2.4
- Transformers 4.31.0
- Pytorch 2.0.1+cu118
- Datasets 2.14.3
- Tokenizers 0.13.2