MiMo-V2-Flash Hybrid Attention Reduces KV caching costs

MiMo-V2-Flash released: 256K long context and multi-token prediction to improve inference throughput

Xiaomi MiMo and the Xiaomi large model Core team have released and opened MiMo-V2-Flash-related resources, positioning it as a basic language model fo...