Experiments

A benchmark explorer for reported NH-LoRA runs.

Browse migrated training logs as benchmark tabs, summary metrics, configuration cards, task-level tables, and raw artifact links.

Benchmark

CIFAR-100

Placeholder CIFAR-100 experiment group. Replace these values with final benchmark runs when they are ready.

Configs

2

Placeholder configs

Top placeholder average

65.68%

Temporary value

Best placeholder forgetting

33.32

Temporary value

NH-LoRA
ViT-B/16
Seed 1

Config-1

Placeholder CIFAR-100 baseline entry. Replace this summary, metric row, and artifact link with the final run.

Avg Acc | Forgetting

65.68 | 36.63

MetricT1T2T3T4T5
Accuracy (%)36.3554.462.5578.4596.65
NH-LoRA
ViT-B/16
Seed 1

Config-2

Placeholder CIFAR-100 configuration entry using the Deployment-style format for later replacement.

Avg Acc | Forgetting

61.41 | 33.32

MetricT1T2T3T4T5
Accuracy (%)45.351.555271.9586.25