 justheuristic
|
f482a98e3a
fix imports
|
3 rokov pred |
|
justheuristic
|
3513d9ad23
renames
|
3 rokov pred |
 Aleksandr Borzunov
|
47ffdb3c9f
renames
|
3 rokov pred |
|
justheuristic
|
0dbc2ac838
Update tests/test_block_exact_match.py
|
3 rokov pred |
|
justheuristic
|
228eee6fcd
Update src/client/remote_sequential.py
|
3 rokov pred |
|
justheuristic
|
ae971adf14
Update src/client/remote_sequential.py
|
3 rokov pred |
|
justheuristic
|
052116adb6
Update src/client/remote_sequential.py
|
3 rokov pred |
 Pavel Samygin
|
81df31ef27
Merge branch 'main' into remove-remote-block
|
3 rokov pred |
 Pavel Samygin
|
a0e3c8943b
fix chained calls
|
3 rokov pred |
|
Pavel Samygin
|
02161c848f
fix missprint
|
3 rokov pred |
|
Pavel Samygin
|
0c0d9119f4
delete info lazy init in tests
|
3 rokov pred |
|
Pavel Samygin
|
c7e3566803
improve type traits
|
3 rokov pred |
|
Pavel Samygin
|
378dc79d9f
fix readme about tests
|
3 rokov pred |
|
Pavel Samygin
|
3cf325004d
reimplement get_remote_module
|
3 rokov pred |
 Artem Chumachenko
|
77220c718c
Add shallow prefix-tuned inference (#55)
|
3 rokov pred |
|
Pavel Samygin
|
5149f47ca7
remove transformer block, implement as sequence size 1
|
3 rokov pred |
|
justheuristic
|
d271b75dd4
Let users specify sequence length instead of assuming 2048 (#52)
|
3 rokov pred |
 Dmitry Baranchuk
|
948877149c
Fix recovering for sequential_backward (#50)
|
3 rokov pred |
|
Dmitry Baranchuk
|
24ba3433e4
[Fix] make distributed seq cls to not create the full bloom model (#49)
|
3 rokov pred |
|
justheuristic
|
f12d0deee9
[quickfix 1/n] remove expensive assertions in inference code (#48)
|
3 rokov pred |
|
Dmitry Baranchuk
|
0fd2caa4be
Convert actual model weights (#46)
|
3 rokov pred |
|
justheuristic
|
a2634001e9
Reduce vocabulary size in test model, fix bug in routing when overlapped (#45)
|
3 rokov pred |
|
Dmitry Baranchuk
|
5745882c67
fix rpc_forward_stream
|
3 rokov pred |
|
Dmitry Baranchuk
|
6095f58681
Deep distributed prompt tuning (#42)
|
3 rokov pred |
|
justheuristic
|
9460220a10
make pytest outputs more verbose (#44)
|
3 rokov pred |
|
Dmitry Baranchuk
|
c4aa1f49df
Rename 350m -> 560m (#43)
|
3 rokov pred |
|
Dmitry Baranchuk
|
11a424837f
integrate mixed-8bit model (#39)
|
3 rokov pred |
 Alexander Borzunov
|
7d39d46966
Use "PETALS" as the readme title (#40)
|
3 rokov pred |
|
Dmitry Baranchuk
|
04a2b6f5e3
Support various backend dtypes & async serialization (#38)
|
3 rokov pred |
|
Artem Chumachenko
|
d989b94614
Pack of Inference Changes (#37)
|
3 rokov pred |
