search

deepseek-ai / DeepSeek-V2

DeepSeek-V2: A Strong, Economical, and Efficient Mixture-of-Experts Language Model
MIT License
3.47k stars 143 forks source link

代码开源相关 #22

Closed DXZDXZ closed 4 months ago

DXZDXZ commented 4 months ago

请问 Aux Loss 部分的 device-level balance loss 和 communication balance loss 代码会开源吗，还有后面的 Token Dropping 策略

luofuli commented 4 months ago

暂无开源计划 ---- 回复的原邮件 ---- @.>发送日期2024年05月11日 18:09 @.> @.>主题[deepseek-ai/DeepSeek-V2] 代码开源相关 (Issue #22) 请问 Aux Loss 部分的 device-level balance loss 和 communication balance loss 代码会开源吗，还有后面的 Token Dropping 策略 — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.Message ID: @.>