Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
BERTweet: A pre-trained language model for English Tweets (EMNLP-2020)
| Date | Stars |
|---|---|
| 2026-07-24 | 609 |
| 2026-07-25 | 609 |
| 2026-07-28 | 609 |
| 2026-07-30 | 609 |
| 2026-07-31 | 609 |
| 2026-08-06 | 609 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
#### Table of contents
1. [Introduction](#introduction)
2. [Main results](#results)
3. [Using BERTweet with `transformers`](#transformers)
- [Pre-trained models](#models2)
- [Example usage](#usage2)
- [Normalize raw input Tweets](#preprocess)
4. [Using BERTweet with `fairseq`](#fairseq)
# <a name="introduction"></a> BERTweet: A pre-trained language model for English Tweets
BERTweet is the first public large-scale language model pre-trained for English Tweets. BERTweet is trained based on the [RoBERTa](https://github.com/pytorch/fairseq/blob/master/examples/roberta/README.md) pre-training procedure. The corpus used to pre-train BERTweet consists of 850M English Tweets (16B word tokens ~ 80GB), containing 845M Tweets streamed from 01/2012 to 08/2019 and 5M Tweets related to the **COVID-19** pandemic. The general architecture and experimental results of BERTweet can be found in our [paper](https://aclanthology.org/2020.emnlp-demos.2/):
@inproceedings{bertweet,
title = {{BERTweet: A pre-trained language model for English Tweets}},
author = {Dat Quoc Nguyen and Thanh Vu and Anh Tuan Nguyen},
booktitle = {Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing: System Demonstrations},
pages = {9--14},
year = {2020}
}
**Please CITE** our paper when BERTweet is used to help produce published results or is incorporated into other software.
## <a name="results"></a> Main results
<img width="275" alt="postagging" src="https://user-images.githubusercontent.com/2412555/135724590-01d8d435-262d-44fe-a383-cd39324fe190.png"> <img width="275" alt="ner" src="https://user-images.githubusercontent.com/2412555/135724598-1e3605e7-d8ce-4c5e-be4a-62ae8501fae7.png">
<img width="275" alt="sentiment" src="https://user-images.githubusercontent.com/2412555/135724597-f1981f1e-fe73-4c03-b1ff-0cae0cc5f948.png"> <img width="275" alt="irony" src="https://user-images.githubusercontent.com/2412555/135724595-15f4f2c8-bbb6-4ee6-82a0-034769dec183.png">
## <a name="transformers"></a> Using BERTweet with [`transformers`](https://github.com/huggingface/transformers)
### Installation <a name="install2"></a>
- Install `transformers` with pip: `pip install transformers`, or [install `transformers` from source](https://huggingface.co/docs/transformers/installation#installing-from-source). <br />
Note that we merged a slow tokenizer for BERTweet into the main `transformers` branch. The process of merging a fast tokenizer for BERTweet is in the discussion, as mentioned in [this pull request](https://github.com/huggingface/transformers/pull/17254#issuecomment-1133932067). If users would like to utilize the fast tokenizer, the users might install `transformers` as follows:
```
git clone --single-branch --branch fast_tokenizers_BARTpho_PhoBERT_BERTweet https://github.com/datquocnguyen/transformers.git
cd transformers
pip3 install -e .
```
- Install `tokenizers` with pip: `pip3 install tokenizers`
### <a name="models2"></a> Pre-trained models
Model | #params | Arch. | Max length | Pre-training data
---|---|---|---|---
[`vinai/bertweet-base`](https://huggingface.co/vinai/bertweet-base) | 135M | base | 128 | 850M English Tweets (cased)
[`vinai/bertweet-covid19-base-cased`](https://huggingface.co/vinai/bertweet-covid19-base-cased) | 135M | base | 128 | [23M COVID-19 English Tweets (cased)](https://forms.gle/sdppxWdmG7bD9rXH7)
[`vinai/bertweet-covid19-base-uncased`](https://huggingface.co/vinai/bertweet-covid19-base-uncased) | 135M | base | 128 | 23M COVID-19 English Tweets (uncased)
[`vinai/bertweet-large`](https://huggingface.co/vinai/bertweet-large) | 355M | large | 512 | 873M English Tweets (cased)
- 09/2020: Two pre-trained models `vinai/bertweet-covid19-base-cased` and `vinai/bertweet-covid19-base-uncased` are resulted by further pre-training the pre-trained model `vinai/bertweet-base` on Excerpt of 7,633 characters
Read on GitHub
38
1
Would you bet a product on this? Bounded 0–100 and slow moving.
matched fp:d26e750a7517d08b, topic:named-entity-recognition, topic:text-classification, topic:sentiment-analysis