 jmorganca
|
ec17359a68
wip
|
11 months ago |
|
jmorganca
|
fbc8572859
add `llava` to `runner`
|
11 months ago |
|
jmorganca
|
28bedcd807
wip
|
11 months ago |
|
jmorganca
|
b22d78720e
cuda linux
|
11 months ago |
|
jmorganca
|
9547aa53ff
disable log file
|
11 months ago |
|
jmorganca
|
a8f91d3cc1
add llava
|
11 months ago |
|
jmorganca
|
e86db9381a
`avx2` should only add `avx2`
|
11 months ago |
|
jmorganca
|
9fe48978a8
move `runner` package down
|
11 months ago |
|
jmorganca
|
01ccbc07fe
replace static build in `llm`
|
11 months ago |
|
jmorganca
|
0110994d06
Initial `llama` Go module
|
1 year ago |
|
jmorganca
|
2ef3a217d1
add sync of llama.cpp
|
1 year ago |
 Michael Yang
|
fccf8d179f
partial decode ggml bin for more info
|
1 year ago |
 Bruce MacDonald
|
984c9c628c
fix embeddings invalid values
|
1 year ago |
|
Bruce MacDonald
|
09d8bf6730
fix build errors
|
1 year ago |
|
Bruce MacDonald
|
7a5f3616fd
embed text document in modelfile
|
1 year ago |
|
Michael Yang
|
f2074ed4c0
Merge pull request #306 from jmorganca/default-keep-system
|
1 year ago |
|
Bruce MacDonald
|
a6f6d18f83
embed text document in modelfile
|
1 year ago |
|
Jeffrey Morgan
|
5eb712f962
trim whitespace before checking stop conditions
|
1 year ago |
|
Michael Yang
|
4dc5b117dd
automatically set num_keep if num_keep < 0
|
1 year ago |
|
Michael Yang
|
b9f4d67554
configurable rope frequency parameters
|
1 year ago |
|
Michael Yang
|
c5bcf32823
update llama.cpp
|
1 year ago |
|
Michael Yang
|
74a5f7e698
no gpu for 70B model
|
1 year ago |
|
Michael Yang
|
319f078dd9
remove -Werror
|
1 year ago |
|
Jeffrey Morgan
|
7da249fcc1
only build metal for `darwin,arm` target
|
1 year ago |
|
Bruce MacDonald
|
184ad8f057
allow specifying stop conditions in modelfile
|
1 year ago |
|
Michael Yang
|
3549676678
embed ggml-metal.metal
|
1 year ago |
|
Michael Yang
|
fadf75f99d
add stop conditions
|
1 year ago |
|
Michael Yang
|
ad3a7d0e2c
add NumGQA
|
1 year ago |
|
Michael Yang
|
cca61181cb
sample metrics
|
1 year ago |
|
Michael Yang
|
c490416189
lock on llm.lock(); decrease batch size
|
1 year ago |
