New AI research from KAIST presents FLASK: a fine-grained assessment framework for language modeling based on skill sets

New AI research from KAIST presents FLASK: a fine-grained assessment framework for language modeling based on skill sets

Incredibly, LLMs have proven to correspond to human values, providing helpful, honest and harmless answers. In particular, this capability has been greatly enhanced by methods that refine a pre-trained LLM on various tasks or user preferences, such as instruction optimization and human feedback reinforcement learning (RLHF). Recent research suggests that by evaluating models based solely…

New AI research proposes a simple but effective structure-based encoder for learning the representation of proteins based on their 3D structures

New AI research proposes a simple but effective structure-based encoder for learning the representation of proteins based on their 3D structures

Proteins, the cell’s energy, are involved in various applications, including materials and treatments. They consist of a chain of amino acids that folds into a certain shape. A significant number of new protein sequences have recently been discovered thanks to the development of low-cost sequencing technology. Accurate and effective in silico protein function annotation methods…