Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Training Efficiency Measurement on Llama 3.1-8B Base (S=512, B200)

3.1Peak Memory Usage (GiB)

AdaLomo

-2.28834.08170.45106.819Jun 22, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.06
3.1584.9
2026.06
3.9229.7
2026.06
4.2232.7
2026.06
4.2249.4
2026.06
4.6221.5
2026.06
5.1181.5
2026.06
6.2248.6
2026.06
6.7186.9
2026.06
6.9614.3
2026.06
7.2188.4
2026.06
9.4267.3
2026.06
9.8246
2026.06
10.4220.6
2026.06
10.7180.6
2026.06
11242.4
2026.06
12.1905.6
2026.06
15255.9
2026.06
18192.1
2026.06
19.1274.2
2026.06
20150.5
2026.06
20.1347.2
2026.06
21.3222.1
2026.06
22.6311.3
2026.06
23.5708
2026.06
23.5708
2026.06
24.4299.7
2026.06
25.2946.7
2026.06
33.5293.3
2026.06
35.8172.6
2026.06
35.8172.6
2026.06
37.2242.1
2026.06
37.2242.1
2026.06
37.6222
2026.06
38.3279.5
2026.06
38.3279.5
2026.06
38.9269.2
2026.06
40332.8
2026.06
40.1237.4
2026.06
42.2218.7
2026.06
42.2218.7
2026.06
43.4268.7
2026.06
45.3364.6
2026.06
45.3364.6
2026.06
46.2251
2026.06
46.2397.2
2026.06
64.5240.6
2026.06
64.5-
2026.06
64.5240.6
2026.06
65.9300.1
2026.06
75205.2
2026.06
75205.2
2026.06
76.5252.6
2026.06
76.6344.4
2026.06
83.6555.3
2026.06
89.9240.1
2026.06
89.9240.1
2026.06
91.8283.9
2026.06
137.8226.3