 Dmitry Baranchuk
|
11a424837f
integrate mixed-8bit model (#39)
|
3 年之前 |
|
Dmitry Baranchuk
|
04a2b6f5e3
Support various backend dtypes & async serialization (#38)
|
3 年之前 |
 justheuristic
|
e2711a033b
Add automated tests (#23)
|
3 年之前 |
|
Dmitry Baranchuk
|
f5463812ad
Shallow prompt tuning (#22)
|
3 年之前 |
|
Dmitry Baranchuk
|
fd0bf064f3
minor refactoring
|
3 年之前 |
|
dbaranchuk
|
21e1f42f04
mv set_requires_grad to remote_model
|
3 年之前 |
|
dbaranchuk
|
5168a3405a
fix comments
|
3 年之前 |
|
dbaranchuk
|
79280c4371
refactoring
|
3 年之前 |
|
dbaranchuk
|
6bffeff0a1
fix
|
3 年之前 |
|
dbaranchuk
|
df42822f26
LM head for CausalLM & chunked forward
|
3 年之前 |
|
justheuristic
|
88c1bf9896
black-isort
|
3 年之前 |
|
justheuristic
|
4695071ad2
WIP: make DistributedBloom compliant with HF interface
|
3 年之前 |
|
justheuristic
|
e32208c954
black-isort
|
3 年之前 |
|
justheuristic
|
4ad845bce3
black-isort
|
3 年之前 |
|
Dmitry Baranchuk
|
e66ab6f1f2
design interface & refactoring
|
3 年之前 |
 Dmitry Baranchuk
|
d969172208
set requires_grad=False, lm_layer -> h @ word_embeddings, rm lm_layer from comverted_model
|
3 年之前 |
|
justheuristic
|
6113a7e955
create alibi inside each block
|
3 年之前 |
|
justheuristic
|
f8b7aaece4
proper __len__
|
3 年之前 |
|
justheuristic
|
eea6fbb318
check for past key values properly
|
3 年之前 |
|
justheuristic
|
3f32912937
check for past key values properly
|
3 年之前 |
|
justheuristic
|
471e47c0f5
black-isort
|
3 年之前 |
|
justheuristic
|
d03b38b9eb
black-isort
|
3 年之前 |
|
justheuristic
|
83cd4412a1
black-isort
|
3 年之前 |
|
justheuristic
|
1ab5fb1630
fetch a specific bloom block without downloading the entire model
|
3 年之前 |
|
justheuristic
|
a6fca51212
fetch a specific bloom block without downloading the entire model
|
3 年之前 |
|
justheuristic
|
e8241d2915
black everything
|
3 年之前 |
|
justheuristic
|
3b9351de1c
isort
|
3 年之前 |
|
justheuristic
|
1c49bcb741
basic backend
|
3 年之前 |
|
justheuristic
|
e5e8c9ed12
expel all bloom-specific files to src.bloom
|
3 年之前 |
