Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Mixture of Block Attention (MoBA) Implementation Performance on New-feature tasks
Loading...
38.7
Session Duration
PithTrain
36.044
53.972
71.9
89.828
May 29, 2026
Session Duration
Active GPU Time
Agent Turns
Per-Turn Context Size
Output Tokens
Updated 2d ago
Evaluation Results
Method
Method
Links
Session Duration
Active GPU Time
Agent Turns
Per-Turn Context Size
Output Tokens
PithTrain
Framework=PithTrain
2026.05
38.7
27.7
57
69,000
32,400
Megatron-LM
Framework=Megatron-LM
2026.05
61.6
49.5
134
120,900
53,800
TorchTitan
Framework=TorchTitan
2026.05
105.1
77.9
91
166,900
111,800
Feedback
Search any
task
Search any
task