Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Exact Retrieval on Needle-in-a-Haystack (NIAH) 64K
Loading...
97.6
Average Accuracy
Baseline SFT
95.104
95.752
96.4
97.048
Apr 15, 2026
Average Accuracy
Updated 2d ago
Evaluation Results
Method
Method
Links
Average Accuracy
Baseline SFT
2026.04
97.6
MEAN0.1
Anchor=MEAN, p=0.1
2026.04
97.53
MEAN0.2
Anchor=MEAN, p=0.2
2026.04
96.47
MIN0.1
Anchor=MIN, p=0.1
2026.04
96.33
MIN0.2
Anchor=MIN, p=0.2
2026.04
95.2
Feedback
Search any
task
Search any
task