LLM Reference

Llama 3 TenyxChat 70B

About

Llama 3 TenyxChat 70B, developed by Tenyx Research, is a finely-tuned 70-billion-parameter instruction model based on the open-source Llama 3 70B. It utilizes proprietary techniques such as Direct Preference Optimization to boost performance as an assistive tool. Trained with the UltraFeedback dataset, it aligns with human preferences for helpfulness and safety. The model excels in dialogue applications, text and code generation, reasoning, math, and handling multi-turn conversations while mitigating "catastrophic forgetting" issues. Despite its strengths, it requires further refinement for safety and still faces some challenges in reasoning and verbosity.

Capabilities

MultimodalFunction CallingTool UseJSON Mode

Providers(1)

ProviderInput (per 1M)Output (per 1M)Type
Azure OpenAI
Provisioned

Specifications

FamilyTenyxChat
ArchitectureDecoder Only
Specializationgeneral