Maintainer: Babelscape

Total Score


Last updated 5/28/2024

Model LinkView on HuggingFace
API SpecView on HuggingFace
Github LinkNo Github link provided
Paper LinkNo paper link provided

Get summaries of the top AI models delivered straight to your inbox:

Model overview

mREBEL is a multilingual version of the REBEL model, introduced in the paper RED^{FM}: a Filtered and Multilingual Relation Extraction Dataset. The model is trained to perform relation extraction, which involves identifying and classifying the relationships between entities in text.

The key difference between mREBEL and the original REBEL model is that mREBEL supports multiple languages, allowing it to be used for relation extraction tasks across a diverse range of languages. The model was trained on a new multilingual dataset called REDFM, which builds upon the original REBEL dataset with additional languages and relation types.

Model inputs and outputs


  • Text: The input to the model is a piece of text containing entities and their relationships.


  • Relation triplets: The model outputs a set of relation triplets, where each triplet consists of a subject entity, a relation type, and an object entity.


mREBEL can be used to perform end-to-end relation extraction on text in over 100 languages. The model is capable of identifying and classifying a wide variety of relation types, making it a versatile tool for tasks like knowledge base population, fact-checking, and other information extraction applications.

What can I use it for?

The mREBEL model can be used for a variety of applications that require extracting structured information from text, such as:

  • Knowledge base population: The model can be used to automatically populate knowledge bases by identifying and extracting relevant relations from text.
  • Fact-checking: By identifying relationships between entities, mREBEL can be used to verify the accuracy of claims and statements.
  • Question answering: The extracted relation triplets can be used to answer questions about the relationships between entities in the text.

Things to try

One interesting aspect of mREBEL is its ability to perform relation extraction on text in over 100 languages. This makes the model a valuable tool for multilingual applications, where you can use it to extract structured information from text in a variety of languages.

Another interesting thing to try with mREBEL is to fine-tune the model on a specific domain or task. By providing the model with additional training data in a particular area, you can potentially improve its performance on that specific use case.

This summary was produced with help from an AI and may contain inaccuracies - check out the links to read the original source documents!

Related Models




Total Score


rebel-large is a relation extraction model developed by Babelscape. It takes a novel approach to relation extraction, framing it as a sequence-to-sequence task rather than a traditional classification task. This allows the model to generate natural language descriptions of the relations between entities, rather than just predicting a relation type. The model achieves state-of-the-art performance on several relation extraction benchmarks, including NYT, CoNLL04, and RE-TACRED. Similar models include multilingual-e5-large, a multi-language text embedding model, bge-large-en-v1.5, BAAI's text embedding model, and GPT-2B-001, a large transformer-based language model. Model inputs and outputs Inputs Text**: The model takes in a piece of text, typically a sentence or short paragraph, as input. Entity mentions**: The model also requires that the entities mentioned in the text be identified and provided as input. Outputs Relation description**: The model outputs a natural language description of the relation between the provided entities. Capabilities rebel-large excels at extracting complex relations between entities within text. Unlike traditional relation extraction models that classify relations into a fixed set of types, rebel-large generates free-form descriptions of the relationships. This allows the model to capture nuanced and context-dependent relationships that may not fit neatly into predefined categories. What can I use it for? rebel-large can be used in a variety of applications that involve understanding the relationships between entities, such as knowledge graph construction, question answering, and text summarization. For example, a company could use rebel-large to automatically extract insights from financial reports or scientific literature, helping to surface important connections and trends. Things to try One interesting aspect of rebel-large is its ability to handle multi-hop relations, where the relationship between two entities is mediated by one or more intermediate entities. This could be explored further by experimenting with more complex input texts and seeing how well the model can uncover these intricate connections.

Read more

Updated Invalid Date




Total Score


The wikineural-multilingual-ner model is a multilingual Named Entity Recognition (NER) model developed by Babelscape. It was fine-tuned on the WikiNEuRal dataset, which was created using a combination of neural and knowledge-based techniques to generate high-quality silver data for NER. The model supports 9 languages: German, English, Spanish, French, Italian, Dutch, Polish, Portuguese, and Russian. Similar models include bert-base-multilingual-cased-ner-hrl, distilbert-base-multilingual-cased-ner-hrl, and mDeBERTa-v3-base-xnli-multilingual-nli-2mil7, all of which are multilingual models fine-tuned for NER or natural language inference tasks. Model inputs and outputs Inputs Text**: The wikineural-multilingual-ner model accepts natural language text as input and performs Named Entity Recognition on it. Outputs Named Entities**: The model outputs a list of named entities detected in the input text, including the entity type (e.g. person, organization, location) and the start/end character offsets. Capabilities The wikineural-multilingual-ner model is capable of performing high-quality Named Entity Recognition on text in 9 different languages, including European languages like German, French, and Spanish, as well as Slavic languages like Russian and Polish. By leveraging a combination of neural and knowledge-based techniques, the model can accurately identify a wide range of entities across these diverse languages. What can I use it for? The wikineural-multilingual-ner model can be a valuable tool for a variety of natural language processing tasks, such as: Information Extraction**: By detecting named entities in text, the model can help extract structured information from unstructured data sources like news articles, social media, or enterprise documents. Content Analysis**: Identifying key named entities in text can provide valuable insights for applications like media monitoring, customer support, or market research. Machine Translation**: The multilingual capabilities of the model can aid in improving the quality of machine translation systems by helping to preserve important named entities across languages. Knowledge Graph Construction**: The extracted named entities can be used to populate knowledge graphs, enabling more sophisticated semantic understanding and reasoning. Things to try One interesting aspect of the wikineural-multilingual-ner model is its ability to handle a diverse set of languages. Developers could experiment with using the model to perform cross-lingual entity recognition, where the input text is in one language and the model identifies entities in another language. This could be particularly useful for applications that need to process multilingual content, such as international news or social media. Additionally, the model's performance could be further enhanced by fine-tuning it on domain-specific datasets or incorporating it into larger natural language processing pipelines. Researchers and practitioners may want to explore these avenues to optimize the model for their particular use cases.

Read more

Updated Invalid Date




Total Score


The madlad400-3b-mt is a multilingual machine translation model based on the T5 architecture that was trained on 1 trillion tokens covering over 450 languages using publicly available data. Developed by Google, this model is competitive with significantly larger models in terms of performance. The model was trained using a similar approach to the Flan-T5 models, which involved fine-tuning the T5 architecture on a mixture of tasks and datasets to improve zero-shot and few-shot performance. Like Flan-T5, the madlad400-3b-mt model can be used for a variety of natural language processing tasks, with a focus on machine translation and multilingual applications. Model Inputs and Outputs Inputs Text to be translated or processed, with a language token `` prepended to indicate the target language. Outputs Translated text or output for the given natural language processing task. Capabilities The madlad400-3b-mt model has been trained on a massive multilingual dataset, allowing it to perform well on a wide range of languages. It can be used for tasks like machine translation, question answering, and text generation, with competitive performance compared to much larger models. What can I use it for? The madlad400-3b-mt model is primarily intended for research purposes, where it can be used to explore the capabilities and limitations of large language models in a multilingual setting. Researchers may find it useful for tasks like zero-shot and few-shot learning, as well as investigating bias and fairness issues in language models. Things to Try One interesting aspect of the madlad400-3b-mt model is its ability to handle long sequences of text, thanks to the use of ALiBi position embeddings. You could try generating or processing text with longer context lengths to see how the model performs. Additionally, the model's multilingual capabilities make it a good candidate for exploring cross-lingual transfer learning, where you fine-tune the model on a task in one language and then evaluate its performance on the same task in another language.

Read more

Updated Invalid Date




Total Score


The madlad400-10b-mt model is a multilingual machine translation model based on the T5 architecture. It was trained by Google on 250 billion tokens covering over 450 languages using publicly available data. The model is competitive with significantly larger models in terms of performance. The model was converted and documented by Juarez Bochi, who was not involved in the original research. Similar models include the madlad400-3b-mt and madlad400-3b-mt models, which are smaller versions of the madlad400-10b-mt model. The PolyLM-13B model is another large multilingual language model, trained by DAMO-NLP-MT. Model inputs and outputs Inputs Text to be translated, potentially in any of over 400 supported languages. Outputs Translated text in the target language. Capabilities The madlad400-10b-mt model is a powerful multilingual machine translation model that can translate text between over 400 different languages. It achieves strong performance, even compared to much larger models, through the use of a large and diverse training dataset. What can I use it for? The primary intended use of the madlad400-10b-mt model is for machine translation and other multilingual NLP tasks. Researchers and developers working on projects that require translation between a wide range of languages may find this model particularly useful. Things to try Some interesting things to try with the madlad400-10b-mt model include: Exploring the model's performance on low-resource language pairs, which can be a key challenge for machine translation. Analyzing the model's outputs to better understand its strengths, weaknesses, and potential biases. Fine-tuning the model on domain-specific data to see if it can be adapted for specialized translation tasks. Comparing the model's performance to other large multilingual models, such as the PolyLM-13B, to gain insights into the state of the art in this field.

Read more

Updated Invalid Date