Wallface Intelligence collaborates with Tsinghua University's open-source edge-side large model BitCPM-CANN

Wallfacer AI, in collaboration with Tsinghua University and the OpenBMB open-source community, has open-sourced BitCPM-CANN, China's first 1.58-bit ternary large-scale model trained end-to-end on Huawei's Ascend domestic computing platform. The model includes four sizes ranging from 0.5B to 8B. Employing a quantization-aware training approach, the model releases approximately 6 times the GPU memory advantage during the inference phase, achieving a capability retention rate of 90%–97.2%. This provides crucial infrastructure for edge AI deployment and the domestic NPU ecosystem.