LLM Reference

Llama 3 8B Instruct

About

The Llama 3 8B Instruct model, released on April 18, 2024, is Meta's latest instruction-following language model with 8 billion parameters. It utilizes an auto-regressive transformer architecture with Grouped-Query Attention for improved scalability. Trained on over 15 trillion tokens and fine-tuned with 10 million human-annotated examples, it excels in dialogue and conversational tasks. The model outperforms its predecessors on industry benchmarks, scoring 68.4 on MMLU (5-shot). Designed for commercial and research applications, it prioritizes safety and helpfulness, making it suitable for chatbots, virtual assistants, and other interactive AI applications. For more details, visit the Hugging Face page [1].

Capabilities

MultimodalFunction CallingTool UseJSON Mode

Providers(18)

ProviderInput (per 1M)Output (per 1M)Type
AWS Bedrock$0.3$0.6Serverless
GroqCloud$0.05$0.08Serverless
deepinfra APIServerless
OctoAI API$0.15$0.15Serverless
Fireworks AI Platform$0.2$0.2Serverless
Alibaba Cloud PAI-EASServerless
Baseten APIServerless
Lepton AI APIServerless
Replicate APIServerless
GCP Vertex AIServerless
Snowflake Cortex$0.38$0.38Serverless
Cloudflare Workers AIServerless
NVIDIA NIMProvisioned
Together AI API$0.18$0.18Serverless
Perplexity LabsServerless
Databricks Foundation Model ServingProvisioned
IBM watsonx$0.6$0.6Serverless
Azure OpenAI$0.37$1.1ServerlessProvisioned

Benchmark Scores(4)

BenchmarkScoreVersionSource
Google-Proof Q&A44.8diamondresearch
HellaSwag91.110-shotresearch
HumanEval68.2pass@1research
Massive Multitask Language Understanding76.95-shotresearch

Specifications

FamilyLlama 3
Released2024-04-18
Parameters8B
Context8K
ArchitectureDecoder Only
Specializationgeneral