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

Language Modeling on WikiText-103 (Throughput and Speedup)

159,000Throughput (tokens/s)

AdamW

6,74446,27285,800125,328Mar 18, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.03
159,000-2.2
2026.03
142,900-4.25
2026.03
134,000-1.5
2026.03
116,2001.3-
2026.03
114,6001.58-
2026.03
89,400--
2026.03
72,400--
2026.03
68,5002.04-
2026.03
55,200-2.09
2026.03
49,300-3.91
2026.03
43,4501.65-
2026.03
33,600--
2026.03
30,000-1.32
2026.03
29,1002.31-
2026.03
28,100-1.89
2026.03
27,8001.22-
2026.03
26,400--
2026.03
23,0001.54-
2026.03
22,700--
2026.03
14,900--
2026.03
12,600--