forked from OpenPipe/ART
-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathTHIRD-PARTY-NOTICES
More file actions
74 lines (53 loc) · 3.13 KB
/
Copy pathTHIRD-PARTY-NOTICES
File metadata and controls
74 lines (53 loc) · 3.13 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
This project includes code from PyTorch TorchTune:
- Repository: https://github.com/pytorch/torchtune
- License: BSD 3-Clause License
- License URL: https://github.com/pytorch/torchtune/blob/main/LICENSE
The following files contain code copied and modified from TorchTune:
- src/art/torchtune/recipe.py
Note: This project also uses TorchTune as a dependency. The above notice applies
specifically to files containing copied/modified code from the TorchTune repository.
Full license text:
BSD 3-Clause License
Copyright 2024 Meta
Redistribution and use in source and binary forms, with or without modification,
are permitted provided that the following conditions are met:
1. Redistributions of source code must retain the above copyright notice, this list
of conditions and the following disclaimer.
2. Redistributions in binary form must reproduce the above copyright notice, this
list of conditions and the following disclaimer in the documentation
and/or other materials provided with the distribution.
3. Neither the name of the copyright holder nor the names of its contributors may
be used to endorse or promote products derived from this software without specific
prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY
EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT
SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED
TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH
DAMAGE.
This project vendors a modified HybridEP runtime derived from DeepEP:
- Repository: https://github.com/deepseek-ai/DeepEP
- License: MIT License
- Vendored path: src/art/megatron/_hybrid_ep
The applicable license text is retained at:
- src/art/megatron/_hybrid_ep/LICENSE
Megatron release wheels also embed pinned source archives used only as headers
when building ART's HybridEP extension:
- NVIDIA NIXL 1.3.2 (commit de8115ca97d3f8fb63a4988e9b4d4a038b2e0f72),
Apache License 2.0, https://github.com/ai-dynamo/nixl
- OpenUCX 1.21.0, BSD 3-Clause License,
https://github.com/openucx/ucx
The complete license and notice files remain in their respective bundled source
archives. ART links HybridEP against the matching NIXL and UCX libraries shipped
by the official `nixl-cu12` or `nixl-cu13` wheel; it does not redistribute those
libraries itself.
On first cross-host HybridEP use, ART may download the checksum-pinned etcd
3.5.33 executable from its official GitHub release. etcd is licensed under the
Apache License 2.0: https://github.com/etcd-io/etcd
The CUDA 12 managed Megatron environment installs NVIDIA Apex from its pinned
25.09 source tag under the BSD 3-Clause License:
https://github.com/NVIDIA/apex