BERT Tokenizer & Embeddings

Paste text below. The app will remove non-ASCII characters, lowercase the text, then use BERT (bert-base-uncased) to produce tokens and embeddings (last hidden state).

Notes

• Embeddings are 768-dim vectors from the last hidden state (one row per token).
• Special tokens like [CLS] and [SEP] are included.
• Truncation keeps the UI responsive; increase Max tokens if needed.