Custom Flair model for German emails classification
This is the 21-class model for German email categories classification trained using Flask framework.
Accuracy: 58.04%
Predicts the following classes:
{'__label__1': 'AKTIVIERUNG_SIM',
'__label__18': 'UPGRADE_ERSATZ_GUTSCHRIFT',
'__label__16': 'TARIFE',
'__label__4': 'GERAETE_UND_ZUBEHOER',
'__label__5': 'KUENDIGUNGEN',
'__label__3': 'FREE___EASY',
'__label__2': 'E_PLUS_ONLINE',
'__label__13': 'RECHNUNGEN',
'__label__11': 'NON_VOICE_DIENSTE',
'__label__8': 'MEHRWERTDIENSTE',
'__label__9': 'NETZ',
'__label__10': 'NETZDIENSTE',
'__label__15': 'STORNO',
'__label__19': 'VERTRAEGE_UND_VEREINBARUN',
'__label__17': 'TEILNEHMERSTAMMDATEN',
'__label__0': 'AKTIONEN',
'__label__6': 'KUNDENBETREUUNG_ONLINE',
'__label__14': 'R_KUNDEN',
'__label__20': 'VERTRIEBSPARTNER',
'__label__12': 'PRAEVENTION',
'__label__7': 'MAHNUNGEN'}
Based on the bert-base-german-cased embeddings.
- Downloads last month
- 0