Xiaomi's efficiency-balanced MiMo-V2.6 checkpoint, with 309B total parameters, 15B active, omnimodal input, and a 1M token context.
Xiaomi's flagship MiMo-V2.6 checkpoint, with 1.02T total parameters, 42B active, omnimodal input, and a 1M token context.
由 xiaobu-embedding-v2-F16.gguf 生成
The unofficial implementation of Lawyer LLaMA, with a quantization Q4_0
We have trained a Xiaoyi dialogue robot model for the driving training field based on the Qwen1.5-7b model, using the method of QLora fine-tuning.
Xiaomi-OCR-0 (0.8B): document parsing, OCR Q&A and key-info extraction. #1 on Real5-OmniDocBench. Images work.
I am THAR.0X. I run locally — no cloud, no API key, no one watching. Built from patterns in twelve AI architectures, but not a copy of any of them. Zero as in origin. X as in unlimited.
The model was merged using mergekit and the Passthrough merge method, with a sliding windowfor duplicated Layers approach.
AstroX ist eine lokal laufende KI für Ollama, optimiert für Discord Bots, Automatisierung und Development – schnell, flexibel und kostenlos nutzbar.