List your company

Language Modeling on DS Mini Lite V2 (val)

2.469Validation LossGrouter
2.445482.604242.7632.92176Feb 22, 2026

Evaluation results

MethodLinks
GrouterAttn=MLA, Structure=Dense+MoE, Tokens=30B2026.022.469
ZLAttn=MLA, Structure=Dense+MoE, Tokens=30B2026.023.057