Zhiqiang Shen

Uncover a model's "digital DNA" – its pretraining data mixture – from its outputs alone, even without access to the training data.

Claude Code's architecture reveals a surprising amount of complexity outside the core LLM loop, with most code dedicated to safety, context management, and extensibility.

LLMs can be surprisingly effective vision models, even without full fine-tuning, thanks to inherent foundational properties in specific layers that transfer via a simple "random label bridge training" technique.