jamesdumay's picture
Add files using upload-large-folder tool
3235f35 verified
metadata
license: apache-2.0
base_model: Qwen/Qwen3-0.6B
tags:
  - mesh-llm
  - mlx
  - qwen3
  - parity

Qwen3 0.6B parity MLX (8bit)

Same-origin MLX parity artifact for Qwen/Qwen3-0.6B, produced for backend comparison work in mesh-llm.

  • Source checkpoint: Qwen/Qwen3-0.6B
  • Conversion flow: original checkpoint -> MLX 8bit
  • Intended pair: meshllm/qwen3-0.6b-parity-q8_0-gguf

This repo is for backend-parity testing rather than for claiming best overall model quality.