Beta
Classificação/Llama 2-34B
Meta

Llama 2-34B

Código aberto

por Meta · Lançado em 2024-01-01

50.4
pontuação média
N/A
Preço de entrada
N/A
Preço de saída
N/A
Janela de contexto
text
Tipo

Tested on 8 benchmarks with 50.4% average. Top scores: TriviaQA (84.6%), PIQA (63.8%), Winogrande (53.4%).

Pontuações de benchmark

BenchmarkCategoriaPontuaçãoBar
TriviaQAknowledge84.6
PIQAknowledge63.8
Winograndeknowledge53.4
MMLUknowledge50.1
OpenBookQAknowledge44.3
GSM8Kmath42.2
ARC AI2knowledge39.3
BBHreasoning25.5

Modelos similares

Google DeepMindGemini 3 Flash Preview
Google DeepMind
50.5
Google DeepMindGemini 2.0 Flash
Google DeepMind
50.0
DeepSeekDeepSeek V3.1
DeepSeek
51.1
OpenAIo3
OpenAI
51.3

Meta Llama 2 Cronologia

Llama 2-13B
53.7
N/AN/Actx11 benchmarks
Llama 2-34B
50.4-3.3
N/AN/Actx8 benchmarks
Llama 2-70B
56.6+6.2
N/AN/Actx12 benchmarks
Llama 2-7B
43.0-13.6
N/AN/Actx11 benchmarks