10 Commits (8af6289d0c695f6139c80acbbc9341395c401798)

Author SHA1 Message Date
BlinkDL fd098b1d2e small update 5 years ago
BlinkDL 3b60c5b266 add wandb, and rename variables 5 years ago
BlinkDL 440bebff1a fixed nan in large models 5 years ago
BlinkDL 62e2cb06d6 fixing nan in large models 5 years ago
BlinkDL d699a69169 misc improvements 5 years ago
BlinkDL 6266f481da minor changes 5 years ago
BlinkDL 89eab46e60 + info 5 years ago
BlinkDL e9fbd9bf70 remove layernorm -> better RWKV 5 years ago
BlinkDL 447eae5841 add MHA-plus model 5 years ago
BlinkDL aa4e2a68f4 first commit 5 years ago