[GLM4MoE] Set attention_softmax_in_fp32 and bf16 defaults in GLMMoEMo…#4314
Open
zhanghonggeng wants to merge 1 commit intoPaddlePaddle:developfrom
Open
[GLM4MoE] Set attention_softmax_in_fp32 and bf16 defaults in GLMMoEMo…#4314zhanghonggeng wants to merge 1 commit intoPaddlePaddle:developfrom
zhanghonggeng wants to merge 1 commit intoPaddlePaddle:developfrom