LLaMA 3: Open Foundation and Instruction-Tuned Models
GPTKB entity
Statements (52)
| Predicate | Object |
|---|---|
| gptkbp:instanceOf |
language model family
|
| gptkbp:architecture |
decoder-only transformer
|
| gptkbp:availableOn |
open license
|
| gptkbp:developedBy |
gptkb:Meta_AI
|
| gptkbp:foundationModel |
yes
|
| gptkbp:hasModel |
gptkb:transformation
|
| https://www.w3.org/2000/01/rdf-schema#label |
LLaMA 3: Open Foundation and Instruction-Tuned Models
|
| gptkbp:instructionTuned |
yes
|
| gptkbp:intendedUse |
research
commercial |
| gptkbp:notableFeature |
supports academic research
open weights supports code generation supports function calling improved performance over LLaMA 2 supports fine-tuning instruction-tuned variants supports alignment research supports chat applications supports code completion supports document analysis supports enterprise use cases supports industry applications supports longer context windows supports multi-turn dialogue supports open innovation supports open-source community contributions supports plug-in ecosystem supports question answering supports reasoning tasks supports research reproducibility supports responsible AI research supports retrieval plugins supports retrieval-augmented generation supports safety research supports summarization supports text completion supports text generation supports tool integration supports tool use |
| gptkbp:parameter |
70B
8B 400B |
| gptkbp:predecessor |
gptkb:LLaMA_2
|
| gptkbp:releaseDate |
2024
|
| gptkbp:supports |
English
multilingual |
| gptkbp:trainer |
publicly available data
Meta-licensed data data from Meta partners |
| gptkbp:bfsParent |
gptkb:Hugo_Touvron
|
| gptkbp:bfsLayer |
7
|