Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Sparse Representation Learning on Llama 3.1 8B activations
Loading...
127.8
Achieved L0
Top-K
28.168
54.034
79.9
105.766
May 6, 2026
Achieved L0
Dead Features (%)
Explained Variance
Mean Squared Error
Coherence (p50)
Coherence (p90)
Coherence (Max)
Updated 26d ago
Evaluation Results
Method
Method
Links
Achieved L0
Dead Features (%)
Explained Variance
Mean Squared Error
Coherence (p50)
Coherence (p90)
Coherence (Max)
Top-K
Target L0=≈ 128
2026.05
127.8
80.3
65.6
0.294
0.1
0.438
1
AEN-SAE
Target L0=≈ 128
2026.05
125.8
39.3
46.9
0.443
0.087
0.218
0.989
AEN-SAE
Target L0=≈ 64
2026.05
68.8
45
40
0.503
0.093
0.182
0.862
Top-K
Target L0=≈ 64
2026.05
64
80.2
58
0.342
0.097
0.36
1
AEN-SAE
Target L0=≈ 32
2026.05
34.5
55.4
19.9
0.671
0.097
0.143
0.585
Top-K
Target L0=≈ 32
2026.05
32
88
50.5
0.406
0.127
0.313
1
Feedback
Search any
task
Search any
task
