Hugging Face's logo

infosave
/
Qwen3.8-Flash-Next-Fast-Code-cmf

Text Generation
cortiq
English
Russian
Chinese
cmf
quantized
q4tp
Mixture of Experts
code
expert-pruning
speculative-decoding
vulkan
cuda
Model card Files
xet
 Community
Qwen3.8-Flash-Next-Fast-Code-cmf / README.md

Commit History

Yesterday

  1. Card: Chinese section links

    infosave
    Verified signature
    e685d2f

  2. Card by backend: CUDA (12 GB and up) and Vulkan (48 GB and up) as separate sections, one figure per stand

    infosave
    Verified signature
    d181f41

  3. Card: plain Chinese troubleshooting line

    infosave
    Verified signature
    e7a084f

  4. Card: requirements, install, run and settings for every card; no aquarium

    infosave
    Verified signature
    ab077e6

  5. cortiq 0.8.78: cards from 12 GB at 91-97 % of Strata Coder with 4-bit experts

    infosave
    Verified signature
    feff033

  6. CUDA on cards from 12 GB: 105 / 118 / 133 tok/s within 12 / 16 / 24 GB

    infosave
    Verified signature
    04344f2

4 days ago

  1. Card: the task-aware cut (fcaq256), aquarium page, new figures

    infosave
    Verified signature
    817ef23

  2. Upload README.md with huggingface_hub

    infosave
    Verified signature
    a10d7a7