Longformer
E435878
Longformer is a transformer-based neural network architecture designed for efficient processing of very long sequences using sparse attention mechanisms.
All labels observed (1)
| Label | Occurrences |
|---|---|
| Longformer canonical | 1 |
How this entity was disambiguated
This entity first appeared as the object of triple T4389206 — resolving that mention is where its identity was fixed. The disambiguator weighed these candidate entities and picked the highlighted one (or “None”, minting a new entity). This is how homonymy is resolved: the same surface form can point to different entities.
NED1
Entity disambiguation (via context triple)
gpt-5-mini-2025-08-07
Target entity: Longformer Context triple: [Hugging Face Transformers, supportsModelType, Longformer]
-
A.
Hugging Face Transformers
Hugging Face Transformers is a widely used open-source library that provides state-of-the-art transformer-based models and tools for natural language processing and related machine learning tasks.
-
B.
GPT-2
GPT-2 is a large transformer-based language model known for generating coherent, human-like text and sparking widespread discussion about the implications of advanced AI text generation.
-
C.
LLM
LLM is the ICAO airline designator assigned to Yamal Airlines, a Russian regional carrier.
-
D.
GPT-3
GPT-3 is a large-scale autoregressive language model known for generating human-like text and performing a wide range of natural language tasks with minimal fine-tuning.
-
E.
AllenNLP
AllenNLP is an open-source natural language processing research library built on PyTorch, designed to facilitate the development and evaluation of state-of-the-art NLP models.
- F. None of above. chosen
- G. Unsure - the case is ambiguous/there is not enough information to decide.
NED2
Entity disambiguation (via description)
gpt-5-mini-2025-08-07
Target entity: Longformer Target entity description: Longformer is a transformer-based neural network architecture designed for efficient processing of very long sequences using sparse attention mechanisms.
-
A.
Hugging Face Transformers
Hugging Face Transformers is a widely used open-source library that provides state-of-the-art transformer-based models and tools for natural language processing and related machine learning tasks.
-
B.
GPT-2
GPT-2 is a large transformer-based language model known for generating coherent, human-like text and sparking widespread discussion about the implications of advanced AI text generation.
-
C.
LLM
LLM is the ICAO airline designator assigned to Yamal Airlines, a Russian regional carrier.
-
D.
GPT-3
GPT-3 is a large-scale autoregressive language model known for generating human-like text and performing a wide range of natural language tasks with minimal fine-tuning.
-
E.
AllenNLP
AllenNLP is an open-source natural language processing research library built on PyTorch, designed to facilitate the development and evaluation of state-of-the-art NLP models.
- F. None of above. chosen
Statements (48)
| Predicate | Object |
|---|---|
| instanceOf |
deep learning model
ⓘ
natural language processing model ⓘ transformer-based neural network architecture ⓘ |
| attentionPattern | sparse attention ⓘ |
| availableIn | Hugging Face Transformers library NERFINISHED ⓘ |
| basedOn | Transformer architecture ⓘ |
| category | long-sequence Transformer model ⓘ |
| citationVenue | Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics NERFINISHED ⓘ |
| comparedTo |
BERT
NERFINISHED
ⓘ
RoBERTa NERFINISHED ⓘ |
| describedIn | Longformer: The Long-Document Transformer NERFINISHED ⓘ |
| designedFor | efficient processing of very long sequences ⓘ |
| developedBy | Allen Institute for AI NERFINISHED ⓘ |
| domain | natural language processing ⓘ |
| extends | RoBERTa pretraining scheme ⓘ |
| handles | documents longer than typical BERT limits ⓘ |
| hasAbbreviation | Longformer NERFINISHED ⓘ |
| hasArchitectureProperty |
combination of local and global attention
ⓘ
linear scaling with sequence length for attention ⓘ |
| hasComplexity | O(n) attention complexity with respect to sequence length ⓘ |
| hasComponent |
global attention pattern
ⓘ
local attention pattern ⓘ |
| implementedIn | PyTorch NERFINISHED ⓘ |
| improvesOver | quadratic attention complexity of standard Transformers ⓘ |
| inputType | token sequences ⓘ |
| introducedIn | 2020 ⓘ |
| language | primarily English in original experiments ⓘ |
| license | Apache 2.0 (via Hugging Face implementation) ⓘ |
| optimizedFor |
coreference resolution
ⓘ
document classification ⓘ long document NLP tasks ⓘ question answering ⓘ |
| proposedBy |
Arman Cohan
NERFINISHED
ⓘ
Iz Beltagy NERFINISHED ⓘ Matthew E. Peters NERFINISHED ⓘ |
| publishedAt | ACL 2020 NERFINISHED ⓘ |
| relatedTo |
BigBird
NERFINISHED
ⓘ
Reformer NERFINISHED ⓘ Sparse Transformer NERFINISHED ⓘ |
| supports | sequence lengths up to 4096 tokens or more ⓘ |
| supportsTask |
sequence classification
ⓘ
token classification ⓘ |
| trainingObjective | masked language modeling ⓘ |
| usedFor |
long document summarization
ⓘ
long-range dependency modeling ⓘ |
| uses |
global attention
ⓘ
sliding window attention ⓘ sparse attention mechanism ⓘ |
How these facts were elicited
The pipeline generated the facts above by prompting gpt-5.1 with this entity's name + description and the instruction below.
Instruction
You are a knowledge base construction expert. Given a subject entity and a description of it, return factual statements that you know for the subject as a JSON list of dictionaries(triples), where keys must be "subject", "predicate" and "object". The number of facts may be very high, between 25 to 50 or more, for very popular subjects. For less popular subjects, the number of facts can be very low, like 5 or 10. # Requirements - If you don't know the subject at all, return an empty list. - If the subject is not a named entity, return an empty list. - Include at least one triple where predicate is "instanceOf". - Do not get too wordy. - Separate several objects into multiple triples with one object.
Input
Subject: Longformer Description of subject: Longformer is a transformer-based neural network architecture designed for efficient processing of very long sequences using sparse attention mechanisms.
Referenced by (1)
Full triples — surface form annotated when it differs from this entity's canonical label.