 Alexander Borzunov
|
c68c1c3b92
Allow torch>=2.3.0 (#603)
|
vor 1 Jahr |
 Anton Sinitsin
|
02bbd85ed8
Added primitives for speculative decoding and tests (#598)
|
vor 1 Jahr |
 Aleksandr Borzunov
|
a2d4b65ae0
Update README.md
|
vor 1 Jahr |
|
Aleksandr Borzunov
|
10fab97e2b
Fix year in citation
|
vor 1 Jahr |
|
Alexander Borzunov
|
8ad5513bea
Fix server warnings, update license links and readme (#602)
|
vor 1 Jahr |
|
Alexander Borzunov
|
67ca11a282
Update hivemind to support torch >= 2.3.0, pydantic >= 2.0 (#601)
|
vor 1 Jahr |
|
Alexander Borzunov
|
103ef760da
Materialize buffers in get_block_size() (#600)
|
vor 1 Jahr |
 justheuristic
|
10f7525ce0
Fix typo in README
|
vor 1 Jahr |
|
justheuristic
|
19be29e89e
note about llama 3.1 RoPE support
|
vor 1 Jahr |
|
justheuristic
|
6477cb85e7
Bump transformers to 4.43.1 (#596)
|
vor 1 Jahr |
 Artem Chumachenko
|
f1e1b051d0
Update peft dependency, fix initialization and inference with new peft (#557)
|
vor 1 Jahr |
|
Anton Sinitsin
|
c0a4d2e3d5
Add option to rollback inference for a certain number of steps (#588)
|
vor 1 Jahr |
|
Anton Sinitsin
|
68585864ae
Update transformers to 4.41.2 (#583)
|
vor 1 Jahr |
 Priyanshupareek
|
e268c99a6b
Restrict PyTorch version to <2.3.0 to resolve import error (#577)
|
vor 1 Jahr |
|
Artem Chumachenko
|
30f522d1a0
Fix dummy cache allocation (#574)
|
vor 1 Jahr |
|
Artem Chumachenko
|
d6f4f80f3f
Fix Mixtral-related issues (#570)
|
vor 1 Jahr |
|
Artem Chumachenko
|
d2fcbbc72e
Add Mixtral models (#553)
|
vor 1 Jahr |
|
justheuristic
|
2ad0b2b936
Fix p2p pushing in rpc_inference (by @miaoqijun ) , support transformers 4.38.2 (#563)
|
vor 1 Jahr |
|
justheuristic
|
efee5d1fa8
Clean disk space in push-docker-image.yaml (#558)
|
vor 1 Jahr |
 Denis Mazur
|
0d91bbdac3
Bump transformers and accelerate versions (#554)
|
vor 1 Jahr |
|
justheuristic
|
d59c15c578
Bump version for inference diagnostics (#543)
|
vor 1 Jahr |
 Max Ryabinin
|
03cbe90234
Optimize LLaMA for inference (#513)
|
vor 1 Jahr |
|
justheuristic
|
25a0796b39
Hotfix: require peft version 0.5.0 (#539)
|
vor 1 Jahr |
|
justheuristic
|
dcce43670f
Hotfix: set transformers version <=4.34 temporarily (#538)
|
vor 1 Jahr |
|
Alexander Borzunov
|
82a97d6e9e
Fix beam search in GPU clients (#531)
|
vor 1 Jahr |
|
Alexander Borzunov
|
47d50e1e29
Improve default arguments for clients and servers (#530)
|
vor 1 Jahr |
|
Max Ryabinin
|
ae19b65095
Add position_ids argument to DistributedFalconModel (#525)
|
vor 1 Jahr |
|
Alexander Borzunov
|
1d9401ddce
Update README.md (#520)
|
vor 1 Jahr |
 FYY
|
a2484b3053
Fix file locks in NFS-mounted directories (#517)
|
vor 1 Jahr |
|
Alexander Borzunov
|
5ce4f1a159
Store (start_block, end_block) in each DHT record for reliability (#510)
|
vor 1 Jahr |
