Commit Graph

457 Commits

Author SHA1 Message Date
Dmitry Baranchuk
f5463812ad
Shallow prompt tuning (#22) 2022-07-15 17:37:26 +03:00
Alexander Borzunov
7e9f337a63
Remove excess line from readme 2022-07-14 22:27:17 +04:00
Dmitry Baranchuk
db966a76dd
rm heuristic for num_handlers 2022-07-14 02:35:23 +03:00
Dmitry Baranchuk
f114a6d417
set default num_handlers=16 2022-07-14 02:32:58 +03:00
Aleksandr Borzunov
f3cf5f4d8d Fix choose_best_blocks() 2022-07-13 22:59:01 +00:00
Alexander Borzunov
75856e4769
Measure and cache network & compute throughput (#21) 2022-07-13 05:46:26 +04:00
Dmitry Baranchuk
ac7df18dfa
Merge pull request #19 from learning-at-home/lm_head
add a modified LM head
2022-07-12 14:25:24 +03:00
Dmitry Baranchuk
fd0bf064f3
minor refactoring 2022-07-12 14:22:11 +03:00
Alexander Borzunov
aba43f1308
Implement block selection on servers (#20) 2022-07-12 14:42:30 +04:00
dbaranchuk
21e1f42f04 mv set_requires_grad to remote_model 2022-07-10 23:41:05 +03:00
dbaranchuk
5168a3405a fix comments 2022-07-10 20:35:13 +03:00
dbaranchuk
79280c4371 refactoring 2022-07-10 20:27:38 +03:00
dbaranchuk
6bffeff0a1 fix 2022-07-10 20:17:11 +03:00
dbaranchuk
b3cc9e0d99 add LM head for DistributedBloomCausalLM 2022-07-10 15:13:46 +03:00
dbaranchuk
df42822f26 LM head for CausalLM & chunked forward 2022-07-10 15:10:13 +03:00
Dmitry Baranchuk
f055135b08
rm prefix 2022-07-08 21:59:25 +03:00
justheuristic
4eadd00a2c rm prefix from tests 2022-07-08 19:11:55 +03:00
justheuristic
1c68670d06 it works 2022-07-08 19:05:57 +03:00
justheuristic
2e90ac30a0 use default prefix in readme 2022-07-07 13:22:03 +03:00
justheuristic
5695897620 fix imports 2022-07-07 04:13:20 +03:00
justheuristic
88c1bf9896 black-isort 2022-07-07 03:49:53 +03:00
justheuristic
3223e8b19b straighten import order 2022-07-07 03:49:19 +03:00
justheuristic
de556c99be straighten import order 2022-07-07 03:49:04 +03:00
justheuristic
c792f50b9c fix import order 2022-07-07 03:42:28 +03:00
justheuristic
90d65e58aa set default DHT prefix 2022-07-07 03:34:58 +03:00
justheuristic
41e5a95e8e set client branch to main by default;
remove the concept of base branch (redundant)
2022-07-07 03:18:10 +03:00
justheuristic
899cefe588 set client branch to main by default;
remove the concept of base branch (redundant)
2022-07-07 03:16:47 +03:00
justheuristic
4695071ad2 WIP: make DistributedBloom compliant with HF interface 2022-07-07 03:11:28 +03:00
justheuristic
6c437c9249 use main as client branch 2022-07-07 03:08:41 +03:00
justheuristic
e32208c954 black-isort 2022-07-07 01:06:03 +03:00
justheuristic
4ad845bce3 black-isort 2022-07-07 01:04:47 +03:00
justheuristic
0f9cd687d4
Merge pull request #18 from learning-at-home/rpc
Chained rpc_forward & rpc_backward
2022-07-07 00:58:10 +03:00
Dmitry Baranchuk
4cb986f680 add chained rpc_forward & rpc_backward 2022-07-06 19:44:28 +03:00
Dmitry Baranchuk
0b5a68983f
Merge pull request #17 from learning-at-home/update-model
Update client and model
2022-07-05 16:41:36 +03:00
Dmitry Baranchuk
29999a6691 rm debug print 2022-07-05 02:02:19 +03:00
Dmitry Baranchuk
e66ab6f1f2 design interface & refactoring 2022-07-05 01:54:47 +03:00
Dmitry Baranchuk
be83e6d0cb refactoring 2022-07-04 22:43:51 +03:00
Dmitry Baranchuk
6a603f9cd6 set requires_grad=False, lm_layer -> h @ word_embeddings, rm lm_layer from comverted_model 2022-07-04 21:26:06 +03:00
Dmitry Baranchuk
d969172208 set requires_grad=False, lm_layer -> h @ word_embeddings, rm lm_layer from comverted_model 2022-07-04 21:18:29 +03:00
justheuristic
30d8bf3301 Merge branch 'main' of github.com:learning-at-home/bloom-demo into main 2022-07-01 16:58:28 +03:00
Aleksandr Borzunov
792dc4bcf8 review 2022-07-01 11:55:48 +03:00
justheuristic
8de7c1687b
list latest additions 2022-07-01 03:57:21 +03:00
justheuristic
d42e8abd38 Merge branch 'client' into main 2022-07-01 03:53:54 +03:00
justheuristic
d688cb0d22 stupid, slow, fragile, but correct full model inference 2022-07-01 03:53:08 +03:00
justheuristic
a6c4a606e0 compare logits to logits 2022-07-01 03:48:36 +03:00
justheuristic
d0c7f2a886 test full model exact match 2022-07-01 03:38:38 +03:00
justheuristic
3f42b3fb8d run inference with no grad 2022-07-01 02:49:02 +03:00
justheuristic
db3ad89c2e temporary fix for info 2022-07-01 02:42:56 +03:00
justheuristic
6113a7e955 create alibi inside each block 2022-07-01 02:36:14 +03:00
justheuristic
8ff45d136d check for absent args 2022-07-01 02:34:10 +03:00