 Aleksandr Borzunov
|
74c086ea35
DO NOT MERGE UNDER ANY CIRCUMSTANCES
|
2 years ago |
|
Alexander Borzunov
|
c735dd7ba3
Update transformers to 4.31.0 and peft to 0.4.0 (#371)
|
2 years ago |
 justheuristic
|
1ab35c2826
Typo in inference_session.py
|
2 years ago |
|
Alexander Borzunov
|
a6fdfc0556
Fix AssertionError on rebalancing (#370)
|
2 years ago |
|
Alexander Borzunov
|
f97582fb5f
Require transformers < 4.31.0 until we're compatible (#369)
|
2 years ago |
|
Alexander Borzunov
|
3b300c32e4
Update readme to show new models (#365)
|
2 years ago |
|
Alexander Borzunov
|
62d9ed5ce7
Implement shortest-path routing for inference (#362)
|
2 years ago |
 Ikko Eltociear Ashimine
|
fd30f7ce10
Fix typo in generation_algorithms.py (#364)
|
2 years ago |
|
Alexander Borzunov
|
11f0d992d7
Report inference, forward, and network RPS separately (#358)
|
2 years ago |
|
Alexander Borzunov
|
9517dd1e3d
Update readme and "Getting started" link (#360)
|
2 years ago |
|
Alexander Borzunov
|
3f733a96e3
Use bitsandbytes 0.40.1.post1 (#357)
|
2 years ago |
|
Alexander Borzunov
|
81c4a45ca2
Make a server ping next servers (#356)
|
2 years ago |
|
Alexander Borzunov
|
2c8959e713
Share more info about a server in DHT (#355)
|
2 years ago |
|
justheuristic
|
37fdcb3fe0
Switch adapters slightly faster (#353)
|
2 years ago |
|
Alexander Borzunov
|
9703358df0
Fix bugs in _choose_num_blocks() added in #346 (#354)
|
2 years ago |
|
Alexander Borzunov
|
1a78638c02
Test that bitsandbytes is not imported when it's not used (#351)
|
2 years ago |
|
justheuristic
|
c511990236
Remove unused import os (#352)
|
2 years ago |
|
Alexander Borzunov
|
e12d4c666b
Spam less in server logs (#350)
|
2 years ago |
|
justheuristic
|
010857a834
Estimate adapter memory overhead in choose_num_blocks() (#346)
|
2 years ago |
|
Alexander Borzunov
|
f605f093f7
Support LLaMA repos without "-hf" suffix (#349)
|
2 years ago |
|
Alexander Borzunov
|
90fbaab61e
Fix Docker build by avoiding Python 3.11 (#348)
|
2 years ago |
|
Alexander Borzunov
|
43acfe52a7
Import petals.utils.peft only when needed to avoid unnecessary import of bitsandbytes (#345)
|
2 years ago |
|
Alexander Borzunov
|
294970fe18
Update Colab link
|
2 years ago |
|
Alexander Borzunov
|
515a5120cb
Mention LLaMA in readme (#344)
|
2 years ago |
 Max Ryabinin
|
13f4e3a88a
Fix convergence issues and switch to LLaMA in the SST-2 example (#343)
|
2 years ago |
 Artem Chumachenko
|
b9f0a5467f
Support peft LoRA adapters (#335)
|
2 years ago |
|
Alexander Borzunov
|
dfc6578c8e
Use bitsandbytes 0.40.0.post4 with bias hotfix (#342)
|
2 years ago |
|
Alexander Borzunov
|
b28f5016ea
Delete deprecated petals.cli scripts (#336)
|
2 years ago |
|
Alexander Borzunov
|
fa095f6461
Use 4-bit for llama by default, use bitsandbytes 0.40.0.post3 (#340)
|
2 years ago |
|
Alexander Borzunov
|
158013a671
Implement direct server-to-server communication (#331)
|
2 years ago |
