Language Models are Unsupervised Multitask Learners
E437278
"Language Models are Unsupervised Multitask Learners" is a 2019 OpenAI research paper that demonstrated how large-scale unsupervised language models like GPT-2 can perform a wide range of tasks without task-specific training.
All labels observed (1)
| Label | Occurrences |
|---|---|
| Language Models are Unsupervised Multitask Learners canonical | 3 |
How this entity was disambiguated
This entity first appeared as the object of triple T4389049 — 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: Language Models are Unsupervised Multitask Learners Context triple: [Rewon Child, coAuthorOf, Language Models are Unsupervised Multitask Learners]
-
A.
Exploring the Limits of Language Modeling
"Exploring the Limits of Language Modeling" is a research paper that investigates how far large-scale neural language models can be pushed in terms of performance, scalability, and generalization on natural language tasks.
-
B.
LLaMA
LLaMA is a family of large language models developed by Meta AI, designed for efficient training and inference across a range of natural language processing tasks.
-
C.
XLNet
XLNet is a generalized autoregressive pretraining model for natural language processing that improves on BERT by leveraging permutation-based language modeling to better capture bidirectional context.
-
D.
Longformer
Longformer is a transformer-based neural network architecture designed for efficient processing of very long sequences using sparse attention mechanisms.
-
E.
Distributed Representations of Sentences and Documents
"Distributed Representations of Sentences and Documents" is a seminal machine learning paper that introduced the Paragraph Vector (Doc2Vec) method for learning continuous vector representations of variable-length text such as sentences, paragraphs, and documents.
- 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: Language Models are Unsupervised Multitask Learners Target entity description: "Language Models are Unsupervised Multitask Learners" is a 2019 OpenAI research paper that demonstrated how large-scale unsupervised language models like GPT-2 can perform a wide range of tasks without task-specific training.
-
A.
Exploring the Limits of Language Modeling
"Exploring the Limits of Language Modeling" is a research paper that investigates how far large-scale neural language models can be pushed in terms of performance, scalability, and generalization on natural language tasks.
-
B.
LLaMA
LLaMA is a family of large language models developed by Meta AI, designed for efficient training and inference across a range of natural language processing tasks.
-
C.
XLNet
XLNet is a generalized autoregressive pretraining model for natural language processing that improves on BERT by leveraging permutation-based language modeling to better capture bidirectional context.
-
D.
Longformer
Longformer is a transformer-based neural network architecture designed for efficient processing of very long sequences using sparse attention mechanisms.
-
E.
Distributed Representations of Sentences and Documents
"Distributed Representations of Sentences and Documents" is a seminal machine learning paper that introduced the Paragraph Vector (Doc2Vec) method for learning continuous vector representations of variable-length text such as sentences, paragraphs, and documents.
- F. None of above. chosen
Statements (46)
| Predicate | Object |
|---|---|
| instanceOf |
research paper
ⓘ
scientific publication ⓘ |
| abbreviation | LMUML ⓘ |
| associatedWith | release of GPT-2 models in staged manner ⓘ |
| author |
Alec Radford
NERFINISHED
ⓘ
Dario Amodei NERFINISHED ⓘ David Luan NERFINISHED ⓘ Ilya Sutskever NERFINISHED ⓘ Jeff Wu NERFINISHED ⓘ Rewon Child NERFINISHED ⓘ |
| concernsAddressed | potential misuse of powerful language models ⓘ |
| concludes | task-agnostic unsupervised training can yield strong performance on many NLP tasks ⓘ |
| demonstrates |
few-shot learning capabilities
ⓘ
multitask performance without task-specific training ⓘ scaling laws for language models qualitatively ⓘ zero-shot learning capabilities ⓘ |
| field |
artificial intelligence
ⓘ
machine learning ⓘ natural language processing ⓘ |
| focusesOn |
language modeling
ⓘ
multitask learning ⓘ unsupervised learning ⓘ |
| hasVersion | technical report ⓘ |
| hostedOn | OpenAI website NERFINISHED ⓘ |
| impact |
popularized the term large language model
ⓘ
sparked discussion on AI capabilities and safety ⓘ |
| influenced | subsequent large language model research ⓘ |
| introduces | GPT-2 1.5B parameter model NERFINISHED ⓘ |
| language | English ⓘ |
| modelType | large-scale transformer language model ⓘ |
| organization | OpenAI NERFINISHED ⓘ |
| proposesModel | GPT-2 NERFINISHED ⓘ |
| publicationYear | 2019 ⓘ |
| publisher | OpenAI NERFINISHED ⓘ |
| relatedTo |
GPT series
NERFINISHED
ⓘ
self-supervised learning ⓘ transformer architecture ⓘ |
| shows |
language models can perform question answering without supervised training
ⓘ
language models can perform reading comprehension without supervised training ⓘ language models can perform summarization without supervised training ⓘ language models can perform text completion ⓘ language models can perform translation without supervised training ⓘ performance improves with model size and data scale ⓘ |
| title | Language Models are Unsupervised Multitask Learners NERFINISHED ⓘ |
| trainingObjective | next-token prediction ⓘ |
| uses | web text corpus for training ⓘ |
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: Language Models are Unsupervised Multitask Learners Description of subject: "Language Models are Unsupervised Multitask Learners" is a 2019 OpenAI research paper that demonstrated how large-scale unsupervised language models like GPT-2 can perform a wide range of tasks without task-specific training.
Referenced by (3)
Full triples — surface form annotated when it differs from this entity's canonical label.
subject surface form:
WebText