Early access
The full LLM
pipeline on
your Mac.
Fine-tune, quantize, test and ship small language models — from a single app. No terminal required.
Built for Apple Silicon. Dataset curation, MLX fine-tuning, GGUF export, and on-device inference — all in one place.
Join the waitlistFine-tune
Llama-3.2-1B · LoRA · r=8
Training
Loss
0.4231
Epoch
2 / 5
Steps
124/640
LR
2e-4
Training progress
19% · ~14 min left
Training loss
12:41:04
step 124 · loss=0.4231 · grad_norm=0.81
12:41:01
step 123 · loss=0.4289 · grad_norm=0.79
12:40:58
step 122 · loss=0.4341 · grad_norm=0.83