mirror of https://github.com/hpcaitech/ColossalAI
e7d3afc9cc
* [optimizer] add div_scale for optimizers * [zero] use div_scale in zero optimizer * fix testing error |
||
---|---|---|
.. | ||
kernels | ||
colossal_C_frontend.cpp | ||
compat.h | ||
cpu_adam.cpp | ||
cpu_adam.h | ||
layer_norm_cuda.cpp | ||
layer_norm_cuda_kernel.cu | ||
moe_cuda.cpp | ||
moe_cuda_kernel.cu | ||
multi_tensor_adam.cu | ||
multi_tensor_apply.cuh | ||
multi_tensor_l2norm_kernel.cu | ||
multi_tensor_lamb.cu | ||
multi_tensor_scale_kernel.cu | ||
multi_tensor_sgd_kernel.cu | ||
multihead_attention_1d.cpp | ||
multihead_attention_1d.h | ||
scaled_masked_softmax.cpp | ||
scaled_masked_softmax.h | ||
scaled_masked_softmax_cuda.cu | ||
scaled_upper_triang_masked_softmax.cpp | ||
scaled_upper_triang_masked_softmax.h | ||
scaled_upper_triang_masked_softmax_cuda.cu | ||
type_shim.h |