List your company

Language Modeling on QWEN3-like (train)

2.398LossMACRO-spec
2.37362.53832.7032.8677May 6, 2026

Evaluation results

MethodLinks
MACRO-specParameter Scale=1B2026.052.398
MACRO-froParameter Scale=1B2026.052.403
MuonH-froParameter Scale=1B2026.052.404
MuonH-specParameter Scale=1B2026.052.409
MuonParameter Scale=1B2026.052.419
MACRO-specParameter Scale=330M2026.052.665
MACRO-froParameter Scale=330M2026.052.67
SSOParameter Scale=330M2026.052.675
MuonH-specParameter Scale=330M2026.052.679
MuonH-froParameter Scale=330M2026.052.68
MuonParameter Scale=330M2026.052.684
FSOParameter Scale=330M2026.052.69
FSOParameter Scale=120M2026.052.99
MACRO-froParameter Scale=120M2026.052.995
MuonH-froParameter Scale=120M2026.052.997
SSOParameter Scale=120M2026.053.001
MACRO-specParameter Scale=120M2026.053.007
MuonParameter Scale=120M2026.053.008
MuonH-specParameter Scale=120M2026.053.008