Code and Data for paper: "CMMaTH: A Chinese Multi-modal Math Skill Evaluation Benchmark for Foundation Models". To systematically evaluate the capability of multimodal large models in solving Chinese ...
$\boxed{\text{Chain-of-Embedding (CoE)}}$ is a brand-new interpretability tool, which captures a progressive embedding chain from input to output space by tracking the hidden states of language models ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results