Mimids
Collection
A series of models designed specifically for Roleplaying • 4 items • Updated
spoomplesmaxx-mockingbird-36B — MLX 3-bit
3-bit MLX quant of spoomplesmaxx-mockingbird-36B for Apple silicon. ~18 GB on disk — runs on a 24 GB Mac with room to breathe, comfortable on 32 GB.
pip install mlx-lm
mlx_lm.chat --model aimeri/spoomplesmaxx-mockingbird-36B-mlx-3bit \
--temp 1.0 --top-p 0.9
The seed-native chat template ships in the tokenizer config; mlx_lm.chat
and LM Studio use it automatically.
Sampling — read this part
temperature 1.0 · top_p 0.9 · no penalties
âš Never use repetition, presence, or frequency penalties. The template ends every message with
<seed:eos>; context-wide penalties suppress that token, the model stops ending its turns, and generation degenerates into the base model's untrained Chinese vocabulary.
Usable temperature window is ~0.95–1.05: lower loops verbatim, higher frays. Full details on the main model card.
mimids 01 · Apache 2.0
- Downloads last month
- 106
Model size
5B params
Tensor type
BF16
·
U32 ·
Hardware compatibility
Log In to add your hardware
3-bit
Model tree for aimeri/spoomplesmaxx-mockingbird-36B-mlx-3bit
Base model
ByteDance-Seed/Seed-OSS-36B-Base-woSyn Finetuned
aimeri/spoomplesmaxx-mockingbird-36B