 Tuan Tran
|
400a00707a
add debug messages
|
il y a 1 an |
|
Tuan Tran
|
747002a537
merge with wa_models
|
il y a 1 an |
 Ning
|
b20ffea609
Unity.cpp dec sync (#257)
|
il y a 1 an |
|
Ning
|
ff54f81e8c
Release to main (#239)
|
il y a 1 an |
 Guillaume Wenzek
|
a768cdf55f
Unity inc (#159)
|
il y a 1 an |
|
Guillaume Wenzek
|
eb7810b81f
force little-endian
|
il y a 1 an |
|
Guillaume Wenzek
|
c31926c1a8
working out new way of saving hparams
|
il y a 1 an |
|
Guillaume Wenzek
|
184c1e1da3
};
|
il y a 1 an |
|
Guillaume Wenzek
|
f1f33dbec1
has_layer + transformer decoder
|
il y a 1 an |
|
Guillaume Wenzek
|
b07a08102a
MultiheadAttention_forward
|
il y a 1 an |
|
Guillaume Wenzek
|
fa85f05545
test flash_attn
|
il y a 1 an |
|
Guillaume Wenzek
|
c2e6384e29
nb
|
il y a 1 an |
|
Guillaume Wenzek
|
3f5912b973
forward
|
il y a 1 an |
|
Guillaume Wenzek
|
2b6130855c
handle 4 dims tensors and eof
|
il y a 1 an |
|
Guillaume Wenzek
|
772f90dfdc
load_fairseq2_ggml_file
|
il y a 1 an |
|
Guillaume Wenzek
|
506dee42d8
move layers to fairseq2.cpp
|
il y a 1 an |
|
Guillaume Wenzek
|
e7dc2b86fc
add ggml python bindings
|
il y a 1 an |
 Naji El Hachem
|
853b53bad0
start working out the loader api
|
il y a 1 an |
