Hugging Face's logo

KaedeTai
/
dflash2-mlx-quantized-draft

MLX
qwen3
speculative-decoding
dflash2
apple-silicon
quantization
benchmark
4-bit precision
Model card Files
xet
 Community
dflash2-mlx-quantized-draft

Commit History

  1. fix: M5 Max is 128 GB, not 137 GB (byte-count/1e9 conversion error)

    KaedeTai
    Verified signature
    e3c1203

  2. fix: 16.1 GB resident (measured), not 15.5 GB of disk weights

    KaedeTai
    Verified signature
    97877d3

  3. DFlash2 quantized-draft loader + M5 Max measurements

    KaedeTai
    Verified signature
    78c877d

  4. initial commit

    KaedeTai
    Verified signature
    2589942