| .. |
|
kernels
|
fix format (#565)
|
2022-04-06 11:40:59 +08:00 |
|
colossal_C_frontend.cpp
|
refactor kernel (#142)
|
2022-01-13 16:47:17 +08:00 |
|
compat.h
|
refactor kernel (#142)
|
2022-01-13 16:47:17 +08:00 |
|
cpu_adam.cpp
|
fix bugs in CPU adam (#633)
|
2022-04-02 17:04:05 +08:00 |
|
cpu_adam.h
|
[zero] cpu adam kernel (#288)
|
2022-03-11 15:50:28 +08:00 |
|
layer_norm_cuda.cpp
|
add colossalai kernel module (#55)
|
2021-12-21 12:19:52 +08:00 |
|
layer_norm_cuda_kernel.cu
|
Optimized MoE layer and fixed some bugs;
|
2022-03-11 15:50:28 +08:00 |
|
moe_cuda.cpp
|
Optimized MoE layer and fixed some bugs;
|
2022-03-11 15:50:28 +08:00 |
|
moe_cuda_kernel.cu
|
Optimized MoE layer and fixed some bugs;
|
2022-03-11 15:50:28 +08:00 |
|
multi_tensor_adam.cu
|
[cuda] modify the fused adam, support hybrid of fp16 and fp32 (#497)
|
2022-03-25 14:15:53 +08:00 |
|
multi_tensor_apply.cuh
|
refactor kernel (#142)
|
2022-01-13 16:47:17 +08:00 |
|
multi_tensor_l2norm_kernel.cu
|
refactor kernel (#142)
|
2022-01-13 16:47:17 +08:00 |
|
multi_tensor_lamb.cu
|
refactor kernel (#142)
|
2022-01-13 16:47:17 +08:00 |
|
multi_tensor_scale_kernel.cu
|
fix format (#563)
|
2022-04-06 11:40:59 +08:00 |
|
multi_tensor_sgd_kernel.cu
|
refactor kernel (#142)
|
2022-01-13 16:47:17 +08:00 |
|
multihead_attention_1d.cpp
|
add colossalai kernel module (#55)
|
2021-12-21 12:19:52 +08:00 |
|
multihead_attention_1d.h
|
add colossalai kernel module (#55)
|
2021-12-21 12:19:52 +08:00 |
|
scaled_masked_softmax.cpp
|
add colossalai kernel module (#55)
|
2021-12-21 12:19:52 +08:00 |
|
scaled_masked_softmax.h
|
add colossalai kernel module (#55)
|
2021-12-21 12:19:52 +08:00 |
|
scaled_masked_softmax_cuda.cu
|
add colossalai kernel module (#55)
|
2021-12-21 12:19:52 +08:00 |
|
scaled_upper_triang_masked_softmax.cpp
|
add colossalai kernel module (#55)
|
2021-12-21 12:19:52 +08:00 |
|
scaled_upper_triang_masked_softmax.h
|
add colossalai kernel module (#55)
|
2021-12-21 12:19:52 +08:00 |
|
scaled_upper_triang_masked_softmax_cuda.cu
|
add colossalai kernel module (#55)
|
2021-12-21 12:19:52 +08:00 |
|
type_shim.h
|
[cuda] modify the fused adam, support hybrid of fp16 and fp32 (#497)
|
2022-03-25 14:15:53 +08:00 |
e014144c44