longbridge/NOTICE at main · ConductorAILabs/longbridge · GitHub

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
longbridge
Copyright 2026 Conductor AI Labs (https://www.conductorailabs.com)

ATTRIBUTION REQUIRED. See LICENSE for the full terms.

----------------------------------------------------------------------------
COMPONENTS
----------------------------------------------------------------------------

Upstream code (Apache 2.0):
  NVlabs/LongLive (https://github.com/NVlabs/LongLive) — Copyright NVIDIA
  Wan-Video/Wan2.2 (https://github.com/Wan-Video/Wan2.2) — Copyright Alibaba

Conductor AI Labs additions (CC BY 4.0):
  - All `# Mac/MPS bridge:` patches inside `ref/`
  - `ref/scripts/static_bg.py` (temporal-median background lock post-process)
  - `ref/configs/inference_mac.yaml`
  - `ref/test_prompts/long_structured.txt`
  - `ref/sweep_prompt.sh` + `ref/_sweep_lib.sh` + `ref/sweep_v2.sh`
  - `ref/test_prompts/subj_*.txt` (per-subject prompt files)
  - `README.md`, `FINDINGS.md`, `PATCHES.md`, `CLEANUP_NOTES.md`

Model weights (downloaded separately, NOT in this repo):
  Wan-AI/Wan2.2-TI2V-5B               — Apache 2.0
  Efficient-Large-Model/LongLive-2.0-5B — CC BY-NC-SA 4.0 (research only)

IMPORTANT: The Conductor AI Labs CC BY 4.0 grant covers ONLY the patches and
additions listed above. It does NOT cover NVIDIA's or Alibaba's model weights.
You are responsible for reviewing those licenses BEFORE any use, especially
commercial use. Output generated with the LongLive-2.0-5B fine-tune is
non-commercial (CC BY-NC-SA 4.0) regardless of how this code is licensed.

  - Wan 2.2: https://huggingface.co/Wan-AI/Wan2.2-TI2V-5B
  - LongLive 2.0: https://huggingface.co/Efficient-Large-Model/LongLive-2.0-5B

----------------------------------------------------------------------------
HOW TO ATTRIBUTE
----------------------------------------------------------------------------

If you redistribute, fork, or build on the Conductor AI Labs additions
(including the Apple Silicon bridge patches), include the following notice
in your project's README or NOTICE file:

    Apple Silicon (MPS) bridge for NVIDIA LongLive 2.0 by Conductor AI Labs
    https://www.conductorailabs.com
    Source: https://github.com/ConductorAILabs/longbridge
    Licensed under CC BY 4.0 — https://creativecommons.org/licenses/by/4.0/

For papers, blog posts, demos, and videos that describe a derived work,
include an equivalent credit line and a link to this repository.

----------------------------------------------------------------------------
LINEAGE
----------------------------------------------------------------------------

The patching playbook and several MPS workarounds (real-arithmetic RoPE,
fp64-to-fp32 schedulers, write_video imageio shim) build on prior MPS
bridge work the Conductor AI Labs team did against NVIDIA SANA-WM.