Ablation: WiderGate32, RoPE dims, activation slopes, hparam stack (8xH100)#1970
Open
bsisduck wants to merge 1 commit into
Open
Ablation: WiderGate32, RoPE dims, activation slopes, hparam stack (8xH100)#1970bsisduck wants to merge 1 commit into
bsisduck wants to merge 1 commit into