AGORA
Nodes
0/ 1215
Peak
364
Loss
3.006
Protocol overhead
61%
Showing single run
Pluralis-13B-Test
Pluralis-13B-TestModelPluralis-13BDatasetFineWeb-Edu (1.3T)
Range
Smooth0.00
X
Zoom
Training loss
Cross-entropy
Throughput
Tokens / sec
Throughput per TFLOP
Tok/s divided by active dense BF16 TFLOPs · higher is better
1.91~
Baselines: Optimal central setting 4.95 · Megatron-LM 200 Mbps 0.23 tok/s/TFLOP. See Analysis tab for full breakdown + sources.
Total swarm TFLOPS
Active dense BF16 compute · Pluralis-operated vs Contributor split
69.29kPluralis 10%Contributor 90%
Mean MFU
Model FLOP utilisation · achieved ÷ peak · higher is better (Chinchilla 6N, 13B, denom: BF16)
14.9%
Evaluations
Not available for Pluralis-13B-Test
No benchmark evaluations are published for this run.