 justheuristic
|
28971dcedd
black-isort
|
пре 3 година |
 Artem Chumachenko
|
5af1c9e3b4
fix test
|
пре 3 година |
|
justheuristic
|
aa769ce846
fix bug when cache assignment surpasses max length
|
пре 3 година |
|
justheuristic
|
93951779e1
fix bug when cache assignment surpasses max length
|
пре 3 година |
|
justheuristic
|
01d6ba8a9c
dedup
|
пре 3 година |
|
Artem Chumachenko
|
1b28432533
wip
|
пре 3 година |
|
justheuristic
|
33aa952d41
client-side prompts
|
пре 3 година |
|
Artem Chumachenko
|
85c2ec7b06
actually merge
|
пре 3 година |
|
justheuristic
|
e0bb3762b4
Merge branch 'generation-inference' into deep_prompt_inference
|
пре 3 година |
|
Artem Chumachenko
|
01c1e198b8
Fix merge conficts
|
пре 3 година |
|
Artem Chumachenko
|
88e6a75996
Add part of deepprompts
|
пре 3 година |
|
Artem Chumachenko
|
c003830cc6
fix
|
пре 3 година |
|
Artem Chumachenko
|
1afd59a071
introduce hypo_ids
|
пре 3 година |
|
Artem Chumachenko
|
f62c65ec23
fixes
|
пре 3 година |
|
Artem Chumachenko
|
53e19de6e0
Add tests
|
пре 3 година |
|
Artem Chumachenko
|
ade986ca58
Return multibatch mode
|
пре 3 година |
|
Artem Chumachenko
|
fb142375cb
WIP
|
пре 3 година |
 Alexander Borzunov
|
54ad745bed
Warn that current instructions involve 6B model but we will replace them soon (#63)
|
пре 3 година |
|
Alexander Borzunov
|
5f0c5329d4
Update readme with arxiv link and more discussions (#62)
|
пре 3 година |
|
Alexander Borzunov
|
9bea7b9ea8
Update bullet points with feedback from Tim and other people (#61)
|
пре 3 година |
|
Alexander Borzunov
|
7653562aa1
Use latest version of Petals scheme, shrink Petals logo (#59)
|
пре 3 година |
|
Alexander Borzunov
|
2eb5843852
Update readme for the 1st public release (#57)
|
пре 3 година |
 Pavel Samygin
|
0be21775af
remove transformer block, implement as sequential of size 1 (#54)
|
пре 3 година |
|
Artem Chumachenko
|
77220c718c
Add shallow prefix-tuned inference (#55)
|
пре 3 година |
|
justheuristic
|
d271b75dd4
Let users specify sequence length instead of assuming 2048 (#52)
|
пре 3 година |
 Dmitry Baranchuk
|
948877149c
Fix recovering for sequential_backward (#50)
|
пре 3 година |
|
Dmitry Baranchuk
|
24ba3433e4
[Fix] make distributed seq cls to not create the full bloom model (#49)
|
пре 3 година |
|
justheuristic
|
f12d0deee9
[quickfix 1/n] remove expensive assertions in inference code (#48)
|
пре 3 година |
|
Dmitry Baranchuk
|
0fd2caa4be
Convert actual model weights (#46)
|
пре 3 година |
|
justheuristic
|
a2634001e9
Reduce vocabulary size in test model, fix bug in routing when overlapped (#45)
|
пре 3 година |
