Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Question Answering on SQuAD (eval)
Loading...
59.84
EM (Clean)
SDBN-p
52.352
54.296
56.24
58.184
Jun 9, 2026
EM (Clean)
EM (Swap-Word)
EM (Homophone)
Updated 1mo ago
Evaluation Results
Method
Method
Links
EM (Clean)
EM (Swap-Word)
EM (Homophone)
SDBN-p
Backbone=LLaMA-3.2-1B,...
2026.06
59.84
35.08
52.2
NEFTune
Backbone=LLaMA-3.2-1B,...
2026.06
59.72
32.72
48.08
Vanilla
Backbone=LLaMA-3.2-1B,...
2026.06
58.92
32.44
47.28
EDA
Backbone=LLaMA-3.2-1B,...
2026.06
58.88
32.44
47.96
FreeLB
Backbone=LLaMA-3.2-1B,...
2026.06
57
31.64
44.04
SMART
Backbone=LLaMA-3.2-1B,...
2026.06
52.64
28.44
40.52
Feedback
Search any
task
Search any
task
