openjev：0.8B Qwen 复刻 Jev 决策模型，支持多模态输入

alex nikolic @justALEXWORTEGA原推发布于 2026.09.21

在 X 上打开原推

https://huggingface.co/spaces/AlexWortega/openjev HUGE thx for @_akhaliq for domain, now openjev is: - 0.8b qwen - image-video-text train for input, audio coming дфеук - calibrated with RLCD (its somehow better then vanila grpo) - sglang support (5x faster then jev by @typesafe lol) 4b and 35b and blog coming later this week

这条是引用着另一条写的：「在 Qwen 4B 上训练 MLP 复刻 Jev」