▌system-one

~ / open-implementations / trained-replicas / valen

Valen

Train-your-own multimodal System One model: evaluates text, images, and video against task instructions and returns probabilities over supplied candidates. Ships the model code, data pipeline, SFT plus experimental RLCD training, and evals.

section
Open Implementations
group
Trained replicas
source
github.com
repo
liuziyu77/valen
stars
782 ▕████████░░░░▏
7-day gain
+190
forks
71
language
Jupyter Notebook
license
Apache-2.0
last pushed
oct 13
added to list
oct 8

## history

2026-09-200 → 782 stars2026-10-13

also in Trained replicas