OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

Bertweet

BERTweet is a pretrained language model for English-language tweets, trained using the RoBERTa pretraining procedure and distributed with model-use and tweet-normalization guidance.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Bertweet — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use BERTweet when a project needs a pretrained language model oriented toward English tweets and can prepare raw tweets using the repository's normalization guidance.

Who it is for

  • Researchers building language-processing workflows for English tweets.
  • Developers preparing tweet text for classification, named-entity recognition or irony detection.

Documented uses

  • Create model representations from normalized English tweet text.
  • Build downstream sentiment or other text-classification workflows.
  • Explore named-entity recognition and irony detection on tweet text.
  • Normalize mentions, URLs and emoji consistently before tokenization.
Before you choose

Requirements & limits

The repository is public and MIT-licensed.

  • The repository lists multiple checkpoints: base models with a maximum length of 128 and a large model with a maximum length of 512.
  • COVID-19-specific base checkpoints were further pretrained on a corpus of 23 million COVID-19 English tweets.
  • The project documents use with Transformers and fairseq; its Transformers example loads a model and tokenizer and returns model features.
  • TweetNormalizer guidance specifies emoji package versions 0.5.4 or 0.6.0.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Prepare raw tweets and tokenize them with Transformers

Normalize raw English tweets and convert one normalized tweet into model input IDs.

Before you start

  • Python environment with PyTorch and Transformers installed.
  • Install NLTK and emoji version 0.5.4 or 0.6.0 for TweetNormalizer.

Steps

  1. Install Transformers and tokenizers as described in the repository.

  2. Install NLTK and emoji 0.5.4 or 0.6.0.

  3. Load a BERTweet tokenizer with AutoTokenizer.from_pretrained.

  4. Pass raw tweet text through normalizeTweet, then encode the normalized text with the tokenizer.

Expected outcome

A tokenized input representation prepared for a BERTweet model.

Guide limitations

  • The repository warns that newer emoji package versions use updated emoji charts that do not match the preprocessing used for pretraining.
  • The documented example prepares model input; it does not provide a complete downstream classifier.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
16 October 2026
Original website title
VinAIResearch/BERTweet · Official homepage title · retained in its source language
How we review sourcesSuggest a correction