@shixiangyu2

model-distill-master covers knowledge distillation from large to small models via a six-step workflow (diagnosis, environment setup, teacher analysis, data synthesis, training config, evaluation). defaults to gemma-3-4b-it as student model. procedurally ambitious but lacks actionable rigor in several key steps.

multi-writing-skills-main converts markdown to html and publishes drafts across wechat, zhihu, and toutiao. includes ai-assisted writing, image generation, and code formatting for chinese platforms.