| .. |
|
attention
|
5e82533d02
upstream: add option to specify tokenizer
|
1 year ago |
|
activation.cpp
|
28866137ea
feat: add swiglu activation
|
1 year ago |
|
activation_kernels.cu
|
28866137ea
feat: add swiglu activation
|
1 year ago |
|
attention.cpp
|
d40a8d6bb0
chore: bind single_query_cached_kv_attention to python
|
1 year ago |
|
cache.cpp
|
a409431c40
feat: draft for cuda kernels
|
1 year ago |
|
cache_kernels.cu
|
a409431c40
feat: draft for cuda kernels
|
1 year ago |
|
layernorm.cpp
|
0ec53128b6
feat: add layernorm kernels
|
1 year ago |
|
layernorm_kernels.cu
|
0ec53128b6
feat: add layernorm kernels
|
1 year ago |
|
pos_encoding.cpp
|
67a17a1e93
feat: add rotary embeddings
|
1 year ago |
|
pos_encoding_kernels.cu
|
67a17a1e93
feat: add rotary embeddings
|
1 year ago |
|
reduction.cuh
|
0ec53128b6
feat: add layernorm kernels
|
1 year ago |
AlpinDale
5e82533d02
upstream: add option to specify tokenizer