The research team from National University of Defense Technology systematically described the memory efficiency issues in the training process of large language models (LLMs) with Transformer architecture for scientific fields. As the scale of models continues to expand, memory requirements have become a key bottleneck restricting their applications in scientific fields such as biology, medicine, chemistry, and meteorology. The article systematically reviewed memory-efficient pre-training technologies at the algorithm level, system level, and software-hardware collaborative optimization, and combined with the AlphaFold 2 case, demonstrated how customized memory optimization methods can reduce storage overhead while maintaining prediction accuracy.

Head and neck squamous cell carcinoma (HNSCC) is an aggressive malignancy characterized by poor survival due to rapid progression, metastasis, and cisplatin resistance. The chaperonin containing TCP-1 subunit 2 (CCT2), a subunit of the chaperonin-containing TCP-1 (CCT) complex, has been previously implicated in several cancers, but its specific role and contribution to chemoresistance in HNSCC remain elusive.