< BACK TO NEWS
importantSYS.SOURCE: Baidu ERNIE• 2026-11-05T09:29:12Z

Baidu Releases ERNIE 5.5 Flash Base: A Sparse Mixture-of-Experts Language Model with 280B Parameters

#ai#moe#efficient_training#llm

Baidu released ERNIE 5.5 Flash Base, a sparse Mixture-of-Experts (MoE) language model with 280B parameters and 14B activated per token, designed for efficient further training and inference. The model features sparse attention mechanisms, latent MoE architecture, and optimized training infrastructure for long-context tasks.

Comments

Read original article

*** END OF TRANSMISSION ***

> YC STARTUP LAUNCHES $1M CHALLENGE FOR AI TO SOLVE KEY MATERIALS SCIENCE PROBLEMS> OPENSSH PASSWORD AUTHENTICATION FLAW IN MUSL-BASED DISTRIBUTIONS> OPENAI'S AI PIPELINE IDENTIFIES 948 LINUX PRIVILEGE ESCALATION VULNERABILITIES> AGILITY ROBOTICS GOES PUBLIC AS FIRST PURE HUMANOID ROBOT COMPANY> OPENSSH 10.6P2 CRITICAL WORKAROUND FOR ALPINE LINUX WITH DISABLED PAM> DECLINE IN QUALITY OF GITHUB AWESOME REPOSITORIES DUE TO AI INFLUENCE> AMAZON REPLACES GRUBHUB WITH UBER EATS FOR PRIME RESTAURANT DELIVERY> ACTIVE CYBER THREATS TARGETING US ENTERPRISES IN OCTOBER 2026> OPENSSH 10.6P2 RELEASE INCLUDES CRITICAL VULNERABILITY WORKAROUND FOR MUSL LIBC SYSTEMS> BAIDU RELEASES ERNIE 5.5 FLASH BASE: A SPARSE MIXTURE-OF-EXPERTS LANGUAGE MODEL WITH 280B PARAMETERS> YC STARTUP LAUNCHES $1M CHALLENGE FOR AI TO SOLVE KEY MATERIALS SCIENCE PROBLEMS> OPENSSH PASSWORD AUTHENTICATION FLAW IN MUSL-BASED DISTRIBUTIONS> OPENAI'S AI PIPELINE IDENTIFIES 948 LINUX PRIVILEGE ESCALATION VULNERABILITIES> AGILITY ROBOTICS GOES PUBLIC AS FIRST PURE HUMANOID ROBOT COMPANY> OPENSSH 10.6P2 CRITICAL WORKAROUND FOR ALPINE LINUX WITH DISABLED PAM> DECLINE IN QUALITY OF GITHUB AWESOME REPOSITORIES DUE TO AI INFLUENCE> AMAZON REPLACES GRUBHUB WITH UBER EATS FOR PRIME RESTAURANT DELIVERY> ACTIVE CYBER THREATS TARGETING US ENTERPRISES IN OCTOBER 2026> OPENSSH 10.6P2 RELEASE INCLUDES CRITICAL VULNERABILITY WORKAROUND FOR MUSL LIBC SYSTEMS> BAIDU RELEASES ERNIE 5.5 FLASH BASE: A SPARSE MIXTURE-OF-EXPERTS LANGUAGE MODEL WITH 280B PARAMETERS