build-small-hackathon/mind-of-tashi-mini-sft-lora

Mind of Tashi — mini student LoRA adapter

The 21.8 MB LoRA adapter behind mind-of-tashi-mini-sft (see that card for the fine-tune config and eval results): r=32 / α=64 over the attention + MLP projections, trained on the mind-of-tashi-selfplay corpus. Apply with peft onto the base model, or use the merged repo directly.