LLM Reference

LLaVA 1.6 Hermes Yi 34B

About

LLaVA-1.6, specifically the Hermes Yi 34B variant, represents a leap in multimodal AI capabilities, enhanced from its predecessor, LLaVA 1.5. This open-source chatbot excels in processing and responding to both text and image inputs. The model boasts a fourfold increase in image resolution support, enhanced visual reasoning and OCR capabilities, and improved visual conversation and world knowledge. It leverages the Nous-Hermes-2-Yi-34B language model as its backbone, offering superior commercial licenses and bilingual support. LLaVA-1.6-34B outshines other open-source models and even competes with Google's Gemini Pro on some tasks. Its training efficiency is impressive, requiring just one day on 32 A100 GPUs, and a demo for chat, image captioning, and visual question answering is accessible online.

Capabilities

MultimodalFunction CallingTool UseJSON Mode

Providers(2)

ProviderInput (per 1M)Output (per 1M)Type
NVIDIA NIM
Provisioned
Fireworks AI Platform
Provisioned

Specifications

FamilyLLaVA 1.6
Released2024-01-31
Parameters34B
ArchitectureDecoder Only
Specializationgeneral