Vector Storage Based Long-term Memory Research on LLM

Kun Li; Xin Jing; Chengang Jing

doi:10.2478/ijanmc-2024-0029

.blurhash-client-img { display: none !important; }

Vector Storage Based Long-term Memory Research on LLM

International Journal of Advanced Network, Monitoring and Controls

Volume 9 (2024): Issue 3 (September 2024)

By: Kun Li, Xin Jing and Chengang Jing

Open Access

|Sep 2024

Figures & Tables

TABLE I.

Experimental Environment

Experimental Environment	Version
CPU	Intel Core i9-10900K
GPU	NVIDIA Tesla V100 PCIe
	32G
Language	Python 3.9
Framework	LangChain

TABLE II.

Experimental results on various datasets

DataSet	Model	NoAgent	ReAct	InterAct	VIMBank
ALFWorld	Qwen2-7b	48.8	54.7	60.1	72.3
ALFWorld	ChatGLM3	46.2	49.2	55.8	64.9
HotpotQA	Qwen2-7b	51.6	57.3	63.4	76.3
HotpotQA	ChatGLM3	45.9	51.8	59.7	71.5
KAgentBench	Qwen2-7b	34.2	48.5	52.6	58.7
KAgentBench	ChatGLM3	32.6	44.7	46.3	54.2

TABLE III.

Reasoning cost of ALFWorld environment

	200	600	1000
NoAgent	63.2K	164.7K	334.7K
VIMBank	56.8K	142.6K	258.3K

References

Authors

Metrics

Articles in this issue

DOI: https://doi.org/10.2478/ijanmc-2024-0029 | Journal eISSN: 2470-8038

Journal RSS Feed

Language: English

Page range: 69 - 79

Published on: Sep 30, 2024

Published by: Xi’an Technological University

In partnership with: Paradigm Publishing Services

Publication frequency: 4 issues per year

Keywords:

Large Language Model,

Long Term Memory,

Vector Storage

Related subjects:

Computer sciences,

Computer sciences, other

© 2024 Kun Li, Xin Jing, Chengang Jing, published by Xi’an Technological University
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License.

Volume 9 (2024): Issue 3 (September 2024)

Vector Storage Based Long-term Memory Research on LLM

Figures & Tables

Figure 1.

Figure 2.

Figure 3.

TABLE I.

Figure 4.

TABLE II.

TABLE III.

Paradigm

My account