Commit
•
088965d
1
Parent(s):
50a7e6a
Create 2302.07842.atom
Browse files- arxiv/2302.07842.atom +76 -0
arxiv/2302.07842.atom
ADDED
@@ -0,0 +1,76 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
<?xml version="1.0" encoding="UTF-8"?>
|
2 |
+
<feed xmlns="http://www.w3.org/2005/Atom">
|
3 |
+
<link href="http://arxiv.org/api/query?search_query%3D%26id_list%3D2302.07842%26start%3D0%26max_results%3D1" rel="self" type="application/atom+xml"/>
|
4 |
+
<title type="html">ArXiv Query: search_query=&id_list=2302.07842&start=0&max_results=1</title>
|
5 |
+
<id>http://arxiv.org/api/wq9zF2OG08gdPxTTIUbwyNNXJ+Q</id>
|
6 |
+
<updated>2023-06-09T00:00:00-04:00</updated>
|
7 |
+
<opensearch:totalResults xmlns:opensearch="http://a9.com/-/spec/opensearch/1.1/">1</opensearch:totalResults>
|
8 |
+
<opensearch:startIndex xmlns:opensearch="http://a9.com/-/spec/opensearch/1.1/">0</opensearch:startIndex>
|
9 |
+
<opensearch:itemsPerPage xmlns:opensearch="http://a9.com/-/spec/opensearch/1.1/">1</opensearch:itemsPerPage>
|
10 |
+
<entry>
|
11 |
+
<id>http://arxiv.org/abs/2302.07842v1</id>
|
12 |
+
<updated>2023-02-15T18:25:52Z</updated>
|
13 |
+
<published>2023-02-15T18:25:52Z</published>
|
14 |
+
<title>Augmented Language Models: a Survey</title>
|
15 |
+
<summary> This survey reviews works in which language models (LMs) are augmented with
|
16 |
+
reasoning skills and the ability to use tools. The former is defined as
|
17 |
+
decomposing a potentially complex task into simpler subtasks while the latter
|
18 |
+
consists in calling external modules such as a code interpreter. LMs can
|
19 |
+
leverage these augmentations separately or in combination via heuristics, or
|
20 |
+
learn to do so from demonstrations. While adhering to a standard missing tokens
|
21 |
+
prediction objective, such augmented LMs can use various, possibly
|
22 |
+
non-parametric external modules to expand their context processing ability,
|
23 |
+
thus departing from the pure language modeling paradigm. We therefore refer to
|
24 |
+
them as Augmented Language Models (ALMs). The missing token objective allows
|
25 |
+
ALMs to learn to reason, use tools, and even act, while still performing
|
26 |
+
standard natural language tasks and even outperforming most regular LMs on
|
27 |
+
several benchmarks. In this work, after reviewing current advance in ALMs, we
|
28 |
+
conclude that this new research direction has the potential to address common
|
29 |
+
limitations of traditional LMs such as interpretability, consistency, and
|
30 |
+
scalability issues.
|
31 |
+
</summary>
|
32 |
+
<author>
|
33 |
+
<name>Grégoire Mialon</name>
|
34 |
+
</author>
|
35 |
+
<author>
|
36 |
+
<name>Roberto Dessì</name>
|
37 |
+
</author>
|
38 |
+
<author>
|
39 |
+
<name>Maria Lomeli</name>
|
40 |
+
</author>
|
41 |
+
<author>
|
42 |
+
<name>Christoforos Nalmpantis</name>
|
43 |
+
</author>
|
44 |
+
<author>
|
45 |
+
<name>Ram Pasunuru</name>
|
46 |
+
</author>
|
47 |
+
<author>
|
48 |
+
<name>Roberta Raileanu</name>
|
49 |
+
</author>
|
50 |
+
<author>
|
51 |
+
<name>Baptiste Rozière</name>
|
52 |
+
</author>
|
53 |
+
<author>
|
54 |
+
<name>Timo Schick</name>
|
55 |
+
</author>
|
56 |
+
<author>
|
57 |
+
<name>Jane Dwivedi-Yu</name>
|
58 |
+
</author>
|
59 |
+
<author>
|
60 |
+
<name>Asli Celikyilmaz</name>
|
61 |
+
</author>
|
62 |
+
<author>
|
63 |
+
<name>Edouard Grave</name>
|
64 |
+
</author>
|
65 |
+
<author>
|
66 |
+
<name>Yann LeCun</name>
|
67 |
+
</author>
|
68 |
+
<author>
|
69 |
+
<name>Thomas Scialom</name>
|
70 |
+
</author>
|
71 |
+
<link href="http://arxiv.org/abs/2302.07842v1" rel="alternate" type="text/html"/>
|
72 |
+
<link title="pdf" href="http://arxiv.org/pdf/2302.07842v1" rel="related" type="application/pdf"/>
|
73 |
+
<arxiv:primary_category xmlns:arxiv="http://arxiv.org/schemas/atom" term="cs.CL" scheme="http://arxiv.org/schemas/atom"/>
|
74 |
+
<category term="cs.CL" scheme="http://arxiv.org/schemas/atom"/>
|
75 |
+
</entry>
|
76 |
+
</feed>
|