Kanana-1.5-8B-Instruct-q4f32_1-MLC

This repository contains the Kanana-1.5-8B-Instruct model compiled into MLC format with q4f32_1 quantization. This format is specifically optimized for use with MLC-LLM and WebLLM.

Model Description

• Original Model: kakao-ai/Kanana-1.5-8B-Instruct
• Format: MLC (compiled for WebGPU/Wasm/Native)
• Quantization: q4f32_1 (4-bit quantization with float32 weights for certain parts)

Contact

For questions or feedback regarding this MLC conversion, please contact: Jin Kim (jinkim.kor.dev@gmail.com)