File size: 939 Bytes
b698673 9dd0c9b b698673 088a889 b698673 d50d629 b698673 |
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 |
---
language: en
license: mit
tags:
- ner
widget:
- text: "These shoes I recently bought from Tommy Hilfiger fit quite well. The shirt, however, has got a hole"
---
### Description
A Named Entity Recognition model trained on a customer feedback data using DistilBert.
Possible labels are in BIO-notation. Performance of the PERS tag could be better because of low data samples:
- PROD: for certain products
- BRND: for brands
- PERS: people names
The following tags are simply in place to help better categorize the previous tags
- MATR: relating to materials, e.g. cloth, leather, seam, etc.
- TIME: time related entities
- MISC: any other entity that might skew the results
### Usage
```
from transformers import AutoTokenizer, AutoModelForTokenClassification
tokenizer = AutoTokenizer.from_pretrained("CouchCat/ma_ner_v7_distil")
model = AutoModelForTokenClassification.from_pretrained("CouchCat/ma_ner_v7_distil")
```
|