|
gptkbp:instanceOf
|
large language model
|
|
gptkbp:architecture
|
gptkb:transformation
|
|
gptkbp:availableOn
|
gptkb:Hugging_Face
gptkb:Meta_AI_website
|
|
gptkbp:context
|
128K tokens (for some variants)
8K tokens
|
|
gptkbp:developedBy
|
gptkb:Meta_AI
|
|
https://www.w3.org/2000/01/rdf-schema#label
|
LLaMA 3
|
|
gptkbp:input
|
gptkb:text
|
|
gptkbp:intendedUse
|
research
commercial
|
|
gptkbp:language
|
English
multilingual
|
|
gptkbp:license
|
custom Meta license
|
|
gptkbp:notableBenchmarks
|
gptkb:SQuAD
gptkb:TriviaQA
gptkb:BBH
gptkb:BoolQ
gptkb:GSM8K
gptkb:HellaSwag
gptkb:HumanEval
gptkb:MMLU
gptkb:OpenBookQA
gptkb:PIQA
gptkb:WinoGrande
gptkb:ARC
gptkb:DROP
|
|
gptkbp:notableFor
|
competitive with GPT-4-class models
improved performance over LLaMA 2
open weights for research use
strong multilingual capabilities
|
|
gptkbp:notableUser
|
gptkb:researchers
academia
developers
companies
|
|
gptkbp:openSource
|
partially
|
|
gptkbp:output
|
gptkb:text
|
|
gptkbp:parameter
|
70B
8B
|
|
gptkbp:predecessor
|
gptkb:LLaMA_2
|
|
gptkbp:releaseYear
|
2024
|
|
gptkbp:supports
|
chat
reasoning
code generation
|
|
gptkbp:trainer
|
reinforcement learning from human feedback
supervised fine-tuning
|
|
gptkbp:trainingDataCutoff
|
2023
|
|
gptkbp:usedFor
|
translator
question answering
summarization
text generation
natural language understanding
chatbots
code completion
reasoning tasks
|
|
gptkbp:bfsParent
|
gptkb:LLaMA
|
|
gptkbp:bfsLayer
|
6
|