Skip to content

NLP Models

Free pre-trained models for natural language processing tasks.

Summarization

Model Size License Browser? Link
bart-large-cnn 400 MB Apache-2.0 No (too large) HF
t5-small 60 MB Apache-2.0 Yes HF
distilbart-cnn-12-6 300 MB Apache-2.0 No HF
pegasus-xsum 570 MB Apache-2.0 No HF

Translation

Model Languages Size License Link
Helsinki-NLP/opus-mt-** 1000+ pairs 150-300 MB CC-BY-4.0 HF
mbart-large-50-many-to-many 50 languages 2.4 GB MIT HF
NLLB-200-distilled-600M 200 languages 600 MB CC-BY-NC HF

Named Entity Recognition

Model Entities Size License Browser? Link
bert-base-NER PER, ORG, LOC, MISC 110 MB MIT Yes HF
distilbert-NER PER, ORG, LOC, MISC 67 MB Apache-2.0 Yes HF
spacy-en-core-web-sm 18 entities 12 MB MIT No (Python) spacy.io

Question Answering

Model Trained On Size License Browser? Link
distilbert-base-uncased-distilled-squad SQuAD 67 MB Apache-2.0 Yes HF
roberta-base-squad2 SQuAD 2.0 125 MB CC-BY-4.0 Yes HF

Token Classification & Parsing

Model Task Size License Link
bert-base-uncased-pos Part-of-speech tagging 110 MB MIT HF
distilbert-punctuation Punctuation restoration 67 MB MIT HF