Apr 30

1. 22:38

   natural language prompt example per Z-Anime prompting guide Running

   Nekochu
   2852dd0
2. 22:33

   fix model filename mismatch, lighter neg prompt

   Nekochu
   4f0eee1
3. 22:27

   fix Examples (needs fn+outputs), full README with API/CLI/MCP docs

   Nekochu
   eaaf713
4. 22:25

   CLI mode, CFG slider 1.0-1.5, better neg prompt, lazy example, MCP, local model paths

   Nekochu
   b107f30
5. 22:01

   add MCP server, REST API, docstring, gradio[mcp], ~30 min timing

   Nekochu
   b7cce06
6. 21:24

   remove offload-to-cpu (disk swapping), keep preload + other opts, ~42 min timing

   Nekochu
   f75cb69
7. 20:10

   preload models into page cache, offload-to-cpu, IQ4_XS text enc, conv-direct, mmap, vae f32, miku theme

   Nekochu
   00c018e
8. 19:51

   add NoCrypt/miku theme

   Nekochu
   0137123
9. 19:24

   use 4-step distill Q5_0 GGUF from WeReCooking repo

   Nekochu
   cd0310f
10. 19:02

    revert to 8-step Q5_0 GGUF (4-step OOMs during build conversion)

    Nekochu
    c9b2c0d
11. 18:50

    fix: use q5_0 quantization type (q5_k_m not supported by sd.cpp)

    Nekochu
    4393113
12. 18:44

    4-step distill: download BF16, convert to Q5_K_M GGUF at build time

    Nekochu
    6fa797e
13. 18:38

    remove --schedule beta (not supported by sd.cpp)

    Nekochu
    5a55ab1
14. 18:19

    Z-Anime 6B CPU: distill 8-step Q5_0, Qwen3-4B Q8_0, euler_a, beta schedule

    Nekochu
    736cf48
15. 18:05

    initial commit

    Nekochu
    Verified signature

    18070fb