main-database/MusicLDM.yaml at preview · Language-Technology-Assessment/main-database · 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
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
---
# Thank you for contributing!
# In filling out this yaml file, please follow the criteria as described here:
# https://osai-index.eu/contribute

# You're free to build on this work and reuse the data. It is licensed under CC-BY 4.0, with the
# stipulation that attribution should come in the form of a link to https://osai-index.eu/
# and a citation to the peer-reviewed paper in which the dataset & criteria were published:

# Liesenfeld, A. and Dingemanse, M., 2024. Rethinking open source generative AI: open-washing and the EU AI Act. In Proceedings of the 2024 ACM Conference on Fairness, Accountability, and Transparency (pp. 1774-1787).

# Organization tags:
# - National origin: United States
# - Contributor type: Non-academic (American Big Tech)

# This system assessment is based on work by the MusGO team, database published here: https://github.com/roserbatlleroca/MusGO_framework/tree/main/projects

system:
    name: MusicLDM
    link: https://github.com/RetroCirce/MusicLDM
    type: audio
    performanceclass: full
    basemodelname: musicldm
    endmodelname: mudicldm
    endmodel license: Attribution-NonCommercial-ShareAlike 4.0 International
    releasedate: 2023-08
    notes: Music generation model

org:
    name: University of California San Diego, Mila-Quebec AI Institute, University of Surre and LAION
    link:
    notes:

# availability:
datasources_basemodel:
    class: closed
    link:
    notes:  MusicLDM is trained on the Audiostock dataset, which contains 9000 music tracks for training and 1000 tracks for testing. Dataset is not directly provided, no information about the original sources accessibility or requirements is provided.


datasources_endmodel:
    class: closed
    link:
    notes: MusicLDM is trained on the Audiostock dataset, which contains 9000 music tracks for training and 1000 tracks for testing. Dataset is not directly provided, no information about the original sources accessibility or requirements is provided.

weights_basemodel:
    class: open
    link: https://drive.google.com/drive/folders/15VDVcIgf99YRM5oGXhRxa_Rowl54uWho
    notes:

weights_endmodel:
    class: open
    link: https://drive.google.com/drive/folders/15VDVcIgf99YRM5oGXhRxa_Rowl54uWho
    notes:

trainingcode:
    class: partial
    link: https://github.com/RetroCirce/MusicLDM
    notes:

# documentation:
code:
    class: open
    link: https://github.com/RetroCirce/MusicLDM/blob/main/README.md
    notes:

hardware_architecture:
    class: open
    link: https://arxiv.org/pdf/2308.01546
    notes: Training procedure and architecture is documented in the preprint and in the appendix additional page for the peer-reviewed version (https://musicldm.github.io/appendix/).


preprint:
    class: open
    link: https://arxiv.org/pdf/2308.01546
    notes:

paper:
    class: open
    link: https://ieeexplore.ieee.org/document/10447265
    notes: Accepted at ICASSP 2024

modelcard:
    class: open
    link: https://huggingface.co/ucsd-reach/musicldm
    notes:

datasheet:
    class: partial
    link: https://github.com/LAION-AI/audio-dataset/blob/main/data_card/Audiostock.md
    notes: Content is limited to data sources origin and details on how to reproduce the data collection. Details on curation and other considerations, such as consent, limitations and selection strategies are missing.

# access:
licenses:
    class: closed
    link: https://github.com/RetroCirce/MusicLDM/blob/main/LICENSE
    notes: Attribution-NonCommercial-ShareAlike 4.0 International