Polinko is a research model exploring human-AI alignment through evals, evidence chains, and runtime tooling.
-
Updated
May 25, 2026 - Python
Polinko is a research model exploring human-AI alignment through evals, evidence chains, and runtime tooling.
读懂大模型的“脑回路”,追踪其思维过程。模型可视化。生成归因图(attribution graphs),其作用类似于大脑的神经网络示意图,通过可视化模型内部超节点及其连接关系,呈现LLM处理信息的路径。
A minimal BART-style encoder-decoder model using Rougeformer components (GQA, RoPE, RMSNorm). For research and experimentation in PyTorch.
polinko toy factory: pick a colour. huey’s is better.
polinko toy factory: scorey keeps the score and you’ve already lost. sorry.
polinko toy factory: probably a mini oracle. definitely a mini chatbot.
Add a description, image, and links to the research-models topic page so that developers can more easily learn about it.
To associate your repository with the research-models topic, visit your repo's landing page and select "manage topics."