أطلق مستخدم GitHub volotat mini-AGI، نموذج لغوي على مستوى البايت مصمم للتدريب من الصفر على hardware المستهلك بذاكرة فيديو 8 جيجابايت فقط. يعالج المشروع limitation أن training models at scale 1B+ عادة ما يتطلب infrastructure من مستوى المؤسسات. بتنفيذ architecture Mixture of Experts (MoE) that dynamically adds and prunes experts during training, and utilizing batch-1 training on a single continuous data stream, the model bounds parameter count by available disk space rather than VRAM capacity.

يتم تحميل الخبراء وإلغائهم من ذاكرة GPU فقط عند الحاجة، eliminating the need to store large randomized batches and gradients. This allows the model to read continuous interleaved passages—each 32K characters long—mimicking human sequential learning. التدريب الحالي proceses a corpus of 7.8B characters, expected to take several weeks to complete.

المؤلف acknowledges extensive collaboration with Claude during development, stating the project would be impossible without AI assistance. Although labeled a "toy-level model" not matching frontier capabilities, it demonstrates that continual learning without catastrophic forgetting is achievable on modest hardware. the repository is publicly available for cloning and observation, enabling anyone to train or continue training their own aligned models.