INDEX

Explanations

Transmissible, Poly, Nutrition

np_acts-logits-general · gemini-2.5-flash-lite

technical or scientific text with specialized terminology.

oai_token-act-pair · claude-3-7-sonnet-20250219 Triggered by @neilrathi

The neuron fires strongly on unusually rare or content‐heavy tokens (e.g. technical terms, punctuation like commas, numbers, hyphens)—i.e. it flags low‐frequency or high‐surprisal tokens.

oai_token-act-pair · o4-mini Triggered by @jyhe0408

New Auto-Interp

Configuration

google/gemma-scope-27b-pt-res/layer_22/width_131k

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

cuál

-1.02

 olvido

-1.01

 relatively

-1.01

 during

-1.00

by

-0.99

 cofre

-0.96

đèn

-0.95

laub

-0.94

aspetto

-0.94

POSITIVE LOGITS

"}";

0.98

,\\

0.95

tizen

0.94

isins

0.93

 actuator

0.92

!!!!!!

0.91

DOD

0.90

 Offers

0.90

 растения

0.90

Oh

0.89

Activations Density 0.002%

Transmissible, Poly, Nutrition

technical or scientific text with specialized terminology.

The neuron fires strongly on unusually rare or content‐heavy tokens (e.g. technical terms, punctuation like commas, numbers, hyphens)—i.e. it flags low‐frequency or high‐surprisal tokens.

No Comments

No Known Activations

Transmissible, Poly, Nutrition

technical or scientific text with specialized terminology.

The neuron fires strongly on unusually rare or content‐heavy tokens (e.g. technical terms, punctuation like commas, numbers, hyphens)—i.e. it flags low‐frequency or high‐surprisal tokens.

No Comments

No Known Activations