Hello World BERT

Live demo results generated by GitHub Actions

bert-base-uncased distilbert-sst-2 HuggingFace Transformers Generated: 2026-03-07 11:23 UTC

Fill-Mask — Masked Language Modelling

BERT predicts the most likely words for a [MASK] token using bidirectional context.

Input
Hello, I'm a [MASK] model.
role
0.1441
fashion
0.1418
new
0.0622
Input
BERT stands for Bidirectional Encoder Representations from [MASK].
bert
0.0612
c
0.0566
memory
0.0549
Input
The [MASK] of the Netherlands is Amsterdam.
capital
0.9997
capitals
0.0001
embassy
0.0000
Input
Natural language [MASK] is a subfield of artificial intelligence.
processing
0.5856
programming
0.1801
intelligence
0.0345

Sentiment Analysis — Text Classification

DistilBERT (fine-tuned on SST-2) classifies each sentence as POSITIVE or NEGATIVE.

Text
BERT is an amazing breakthrough in natural language processing!
POSITIVE
0.9999
Text
I struggled to understand the attention mechanism at first.
NEGATIVE
0.9996
Text
Transformers have completely revolutionised how we approach NLP tasks.
POSITIVE
0.9981
Text
The model was too slow and the accuracy was disappointing.
NEGATIVE
0.9997