| license: apache-2.0 | |
| base_model: yuxinlu1/gemma-4-12B-coder-fable5-composer2.5-v1 | |
| library_name: mlx | |
| tags: | |
| - mlx | |
| - quantized | |
| - oq | |
| - gemma4 | |
| - coding | |
| # Gemma 4 12B Coder Fable5 Composer2.5 oQ Quantization Sweep | |
| This repository is a public preservation archive for locally generated MLX quantizations of the 8-bit MLX conversion of [yuxinlu1/gemma-4-12B-coder-fable5-composer2.5-v1](https://huggingface.co/yuxinlu1/gemma-4-12B-coder-fable5-composer2.5-v1). | |
| ## Provenance | |
| - Input conversion: `mlx-community/gemma-4-12b-coder-fable5-composer2.5-8bit` | |
| - Base model: `google/gemma-4-12B-it` | |
| - Original fine-tune: `yuxinlu1/gemma-4-12B-coder-fable5-composer2.5-v1` | |
| - Quantization: local oMLX/oQ affine MLX safetensors, group size 64 | |
| Each subdirectory is an independently loadable MLX artifact. The directory name is the intended quantization label. This archive preserves the local experiment lineage; it does not claim an upstream benchmark or quality guarantee. | |