pierreprudh/ gemma4-12b-mtp:latest

481 2 weeks ago

Gemma 4 12B (QAT Q4_0) with Multi-Token Prediction speculative decoding — a fast, tool-capable local model for 16 GB - Apple Silicon Macs (~21 tok/s, ~1.4× via MTP).

tools thinking
... /
params
76ada76b335b · 31B
{
"stop": [
"<turn|>"
]
}