README.md
2.0 KB · 57 lines · markdown Raw
1 ---
2 language:
3 - en
4 - nl
5 - de
6 - fr
7 - it
8 - es
9
10 license: mit
11 ---
12
13 # bert-base-multilingual-uncased-sentiment
14
15 Visit the [NLP Town website](https://www.nlp.town) for an updated version of this model, with a 40% error reduction on product reviews.
16
17 This is a bert-base-multilingual-uncased model finetuned for sentiment analysis on product reviews in six languages: English, Dutch, German, French, Spanish, and Italian. It predicts the sentiment of the review as a number of stars (between 1 and 5).
18
19 This model is intended for direct use as a sentiment analysis model for product reviews in any of the six languages above or for further finetuning on related sentiment analysis tasks.
20
21 ## Training data
22
23 Here is the number of product reviews we used for finetuning the model:
24
25 | Language | Number of reviews |
26 | -------- | ----------------- |
27 | English | 150k |
28 | Dutch | 80k |
29 | German | 137k |
30 | French | 140k |
31 | Italian | 72k |
32 | Spanish | 50k |
33
34 ## Accuracy
35
36 The fine-tuned model obtained the following accuracy on 5,000 held-out product reviews in each of the languages:
37
38 - Accuracy (exact) is the exact match for the number of stars.
39 - Accuracy (off-by-1) is the percentage of reviews where the number of stars the model predicts differs by a maximum of 1 from the number given by the human reviewer.
40
41
42 | Language | Accuracy (exact) | Accuracy (off-by-1) |
43 | -------- | ---------------------- | ------------------- |
44 | English | 67% | 95%
45 | Dutch | 57% | 93%
46 | German | 61% | 94%
47 | French | 59% | 94%
48 | Italian | 59% | 95%
49 | Spanish | 58% | 95%
50
51 ## Contact
52
53 In addition to this model, [NLP Town](http://nlp.town) offers custom models for many languages and NLP tasks.
54
55 If you found this model useful, you can [buy us a coffee](https://www.buymeacoffee.com/yvespeirsman).
56
57 Feel free to contact us for questions, feedback and/or requests for similar models.