INDEX

Explanations

scores and rankings

The neuron activates on numeric tokens, i.e. standalone numbers.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

urants

-0.92

EXPECT

-0.87

union

-0.83

SELL

-0.80

srand

-0.78

 Schedule

-0.76

zünd

-0.75

当選

-0.74

Schedule

-0.73

budget

-0.73

POSITIVE LOGITS

 ranking

2.11

 leaderboard

2.05

 rankings

1.99

 ranked

1.63

 rank

1.52

Ranking

1.50

 Rankings

1.48

 Ranking

1.48

 leader

1.48

 Leaderboard

1.47

Activations Density 0.014%