Update README.md
Browse files
README.md
CHANGED
@@ -3,97 +3,43 @@ language:
|
|
3 |
- 'da'
|
4 |
- 'no'
|
5 |
library_name: transformers
|
|
|
6 |
---
|
7 |
-
# Model Card for A
|
8 |
|
9 |
Text classification model that determines whether a not a short text contains an attack.
|
10 |
|
11 |
-
# Model Details
|
12 |
|
13 |
-
|
14 |
|
15 |
-
The model is based on the T5
|
16 |
|
17 |
|
18 |
-
- **Developed by:**
|
19 |
-
- **
|
20 |
-
- **
|
21 |
-
- **Language(s) (NLP):** [More Information Needed]
|
22 |
- **License:** [More Information Needed]
|
23 |
-
- **Finetuned from model
|
24 |
|
25 |
-
## Model Sources [optional]
|
26 |
|
27 |
-
|
|
|
28 |
|
29 |
-
- **Repository:** [More Information Needed]
|
30 |
-
- **Paper [optional]:** [More Information Needed]
|
31 |
-
- **Demo [optional]:** [More Information Needed]
|
32 |
-
|
33 |
-
# Uses
|
34 |
-
|
35 |
-
<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->
|
36 |
-
|
37 |
-
## Direct Use
|
38 |
-
|
39 |
-
<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->
|
40 |
-
|
41 |
-
[More Information Needed]
|
42 |
-
|
43 |
-
## Downstream Use [optional]
|
44 |
-
|
45 |
-
<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->
|
46 |
-
|
47 |
-
[More Information Needed]
|
48 |
-
|
49 |
-
## Out-of-Scope Use
|
50 |
-
|
51 |
-
<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->
|
52 |
-
|
53 |
-
[More Information Needed]
|
54 |
|
55 |
# Bias, Risks, and Limitations
|
56 |
-
|
57 |
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
58 |
-
|
59 |
-
[More Information Needed]
|
60 |
-
|
61 |
-
## Recommendations
|
62 |
-
|
63 |
-
<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->
|
64 |
-
|
65 |
-
Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.
|
66 |
-
|
67 |
-
# Training Details
|
68 |
-
|
69 |
-
## Training Data
|
70 |
-
|
71 |
-
<!-- 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. -->
|
72 |
-
|
73 |
-
[More Information Needed]
|
74 |
-
|
75 |
-
## Training Procedure [optional]
|
76 |
-
|
77 |
-
<!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->
|
78 |
-
|
79 |
-
### Preprocessing
|
80 |
-
|
81 |
[More Information Needed]
|
82 |
|
83 |
-
|
|
|
84 |
|
85 |
-
|
86 |
-
|
87 |
-
[More Information Needed]
|
88 |
|
89 |
# Evaluation
|
90 |
-
|
91 |
<!-- This section describes the evaluation protocols and provides the results. -->
|
92 |
-
|
93 |
## Testing Data, Factors & Metrics
|
94 |
|
95 |
### Testing Data
|
96 |
-
|
97 |
<!-- This should link to a Data Card if possible. -->
|
98 |
|
99 |
[More Information Needed]
|
@@ -106,7 +52,7 @@ Users (both direct and downstream) should be made aware of the risks, biases and
|
|
106 |
|
107 |
### Metrics
|
108 |
|
109 |
-
|
110 |
|
111 |
[More Information Needed]
|
112 |
|
@@ -118,11 +64,6 @@ Users (both direct and downstream) should be made aware of the risks, biases and
|
|
118 |
|
119 |
|
120 |
|
121 |
-
# Model Examination [optional]
|
122 |
-
|
123 |
-
<!-- Relevant interpretability work for the model goes here -->
|
124 |
-
|
125 |
-
[More Information Needed]
|
126 |
|
127 |
# Environmental Impact
|
128 |
|
@@ -132,65 +73,33 @@ Carbon emissions can be estimated using the [Machine Learning Impact calculator]
|
|
132 |
|
133 |
- **Hardware Type:** [More Information Needed]
|
134 |
- **Hours used:** [More Information Needed]
|
135 |
-
- **Cloud Provider:**
|
136 |
-
- **Compute Region:**
|
137 |
- **Carbon Emitted:** [More Information Needed]
|
138 |
|
139 |
-
# Technical Specifications [optional]
|
140 |
-
|
141 |
-
## Model Architecture and Objective
|
142 |
-
|
143 |
-
[More Information Needed]
|
144 |
-
|
145 |
-
## Compute Infrastructure
|
146 |
-
|
147 |
-
[More Information Needed]
|
148 |
-
|
149 |
-
### Hardware
|
150 |
-
|
151 |
-
[More Information Needed]
|
152 |
-
|
153 |
-
### Software
|
154 |
-
|
155 |
-
[More Information Needed]
|
156 |
-
|
157 |
-
# Citation [optional]
|
158 |
-
|
159 |
-
<!-- If there is a paper or blog post introducing the model, the APA and Bibtex information for that should go in this section. -->
|
160 |
-
|
161 |
-
**BibTeX:**
|
162 |
-
|
163 |
-
[More Information Needed]
|
164 |
|
165 |
-
|
|
|
166 |
|
167 |
-
|
168 |
-
|
169 |
-
# Glossary [optional]
|
170 |
-
|
171 |
-
<!-- If relevant, include terms and calculations in this section that can help readers understand the model or model card. -->
|
172 |
-
|
173 |
-
[More Information Needed]
|
174 |
-
|
175 |
-
# More Information [optional]
|
176 |
-
|
177 |
-
[More Information Needed]
|
178 |
-
|
179 |
-
# Model Card Authors [optional]
|
180 |
-
|
181 |
-
[More Information Needed]
|
182 |
|
183 |
-
|
184 |
|
185 |
-
|
|
|
186 |
|
187 |
-
#
|
|
|
|
|
188 |
|
189 |
-
|
|
|
|
|
190 |
|
191 |
-
|
192 |
-
|
|
|
|
|
193 |
|
194 |
-
|
195 |
|
196 |
-
</details>
|
|
|
3 |
- 'da'
|
4 |
- 'no'
|
5 |
library_name: transformers
|
6 |
+
f1-score: 0.76
|
7 |
---
|
8 |
+
# Model Card for A&ttack2
|
9 |
|
10 |
Text classification model that determines whether a not a short text contains an attack.
|
11 |
|
|
|
12 |
|
13 |
+
# Model Description
|
14 |
|
15 |
+
The model is based on the [North-T5-NCC Large](https://huggingface.co/north/t5_large_NCC) (developed by Per E. Kummervold) which is a Scandinavian language built upon [T5](https://github.com/google-research/text-to-text-transfer-transformer) and [T5X](https://github.com/google-research/t5x). The model is further trained on ~70k Norwegian and ~67k Danish social media posts which have been classified as either 'attack' or 'not attack', making it a text-to-text model manipulated to do classification. The model is described in Danish in [this report](https://strapi.ogtal.dk/uploads/966f1ebcfa9942d3aef338e9920611f4.pdf).
|
16 |
|
17 |
|
18 |
+
- **Developed by:** The development team at Analyse & Tal
|
19 |
+
- **Model type:** Language model restricted to classification
|
20 |
+
- **Language(s) (NLP):** Danish and Norwegian
|
|
|
21 |
- **License:** [More Information Needed]
|
22 |
+
- **Finetuned from model:** [More information needed]
|
23 |
|
|
|
24 |
|
25 |
+
# Direct Use
|
26 |
+
The model can be used directly to classify Danish and Norwegian social media posts (or similar pieces of text).
|
27 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
28 |
|
29 |
# Bias, Risks, and Limitations
|
|
|
30 |
<!-- This section is meant to convey both technical and sociotechnical limitations. -->
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
31 |
[More Information Needed]
|
32 |
|
33 |
+
# Training Data
|
34 |
+
A collection of ~70k Norwegian and ~67k Danish social media posts have been manually annotated as 'attack' or 'not attack' by six individual coders. 5% of the posts have been annotated by more then one annotator, with the annotators in agreement for 83% of annotations.
|
35 |
|
36 |
+
*Hvad er data-split metoden? Hvad er training-validation-test split?*
|
|
|
|
|
37 |
|
38 |
# Evaluation
|
|
|
39 |
<!-- This section describes the evaluation protocols and provides the results. -->
|
|
|
40 |
## Testing Data, Factors & Metrics
|
41 |
|
42 |
### Testing Data
|
|
|
43 |
<!-- This should link to a Data Card if possible. -->
|
44 |
|
45 |
[More Information Needed]
|
|
|
52 |
|
53 |
### Metrics
|
54 |
|
55 |
+
Macro-averaged f1-score: 0.76
|
56 |
|
57 |
[More Information Needed]
|
58 |
|
|
|
64 |
|
65 |
|
66 |
|
|
|
|
|
|
|
|
|
|
|
67 |
|
68 |
# Environmental Impact
|
69 |
|
|
|
73 |
|
74 |
- **Hardware Type:** [More Information Needed]
|
75 |
- **Hours used:** [More Information Needed]
|
76 |
+
- **Cloud Provider:** Azure
|
77 |
+
- **Compute Region:** North-Europe
|
78 |
- **Carbon Emitted:** [More Information Needed]
|
79 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
80 |
|
81 |
+
# Model Card Authors
|
82 |
+
This model card was written by the developer team at Analyse & Tal. Contact: oyvind@ogtal.dk.
|
83 |
|
84 |
+
# How to Get Started with the Model
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
85 |
|
86 |
+
Use the code below to get started with the model.
|
87 |
|
88 |
+
```
|
89 |
+
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
|
90 |
|
91 |
+
# Download/load tokenizer and language model
|
92 |
+
tokenizer = AutoTokenizer.from_pretrained("ogtal/A-og-ttack2")
|
93 |
+
model = AutoModelForSeq2SeqLM.from_pretrained("ogtal/A-og-ttack2")
|
94 |
|
95 |
+
# Give sample text. The example is from a social media comment.
|
96 |
+
sample_text = "Velbekomme dit klamme usle løgnersvin!"
|
97 |
+
input_ids = tokenizer("Velbekomme", return_tensors="pt").input_ids
|
98 |
|
99 |
+
# Forward pass and print the output
|
100 |
+
outputs = model.generate(input_ids)
|
101 |
+
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
|
102 |
+
```
|
103 |
|
104 |
+
Running the above code will print "angreb" (attack in Danish)
|
105 |
|
|