-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathCITATION.cff
More file actions
75 lines (75 loc) · 2.33 KB
/
Copy pathCITATION.cff
File metadata and controls
75 lines (75 loc) · 2.33 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
75
cff-version: 1.2.0
message: >-
If you use this software or the Fisher-MoE method, please cite the
EMNLP 2026 paper below.
title: >-
Less is MoE: Trimming Experts in Domain-Specialist Language Models
abstract: >-
Official implementation of Fisher-MoE: intra-expert structured pruning and
compression for Mixture-of-Experts (MoE) large language models. Prunes FFN
intermediate dimensions ranked by Fisher importance instead of dropping or
merging whole experts, preserving general-purpose capability at a 50%
compression ratio while reducing weight memory by ~45% and improving
inference throughput by 21%.
type: software
license: Apache-2.0
repository-code: https://github.com/HectorHHZ/Less-is-MoE
url: https://arxiv.org/abs/2606.05538
keywords:
- mixture-of-experts
- moe
- expert-pruning
- model-compression
- structured-pruning
- large-language-models
- efficient-inference
- fisher-information
authors:
- given-names: Haoze
family-names: He
- given-names: Xinkai
family-names: Zou
- given-names: Xuan
family-names: Jiang
- given-names: Xingyuan
family-names: Ding
- given-names: Ao
family-names: Qu
- given-names: Juncheng Billy
family-names: Li
- given-names: Heather
family-names: Miller
# Accepted at EMNLP 2026 (Main Conference, Oral). The proceedings are not
# published yet, so the preferred citation points at the arXiv version, which is
# the record Google Scholar and Semantic Scholar currently index. Switch this to
# type: conference-paper with the ACL Anthology collection-title, pages, and URL
# once the proceedings appear.
preferred-citation:
type: article
title: >-
Less is MoE: Trimming Experts in Domain-Specialist Language Models
journal: arXiv preprint arXiv:2606.05538
year: 2026
url: https://arxiv.org/abs/2606.05538
identifiers:
- type: other
value: arXiv:2606.05538
description: arXiv preprint
- type: doi
value: 10.48550/arXiv.2606.05538
description: arXiv DOI
authors:
- given-names: Haoze
family-names: He
- given-names: Xinkai
family-names: Zou
- given-names: Xuan
family-names: Jiang
- given-names: Xingyuan
family-names: Ding
- given-names: Ao
family-names: Qu
- given-names: Juncheng Billy
family-names: Li
- given-names: Heather
family-names: Miller