INDEX

Explanations

initialisms and abbreviations

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

periods following single capital letters in abbreviations and initials.

oai_token-act-pair · claude-4-5-sonnet Triggered by @bfgcn7frf5

The neuron is highly sensitive to isolated single-letter uppercase tokens (e.g. “C,” “A,” “P,” “D,” etc.), which commonly appear in citations, abbreviations, and headings.

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

New Auto-Interp

Configuration

google/gemma-scope-27b-pt-res/layer_10/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

 gave

-3.45

 went

-3.34

 took

-3.25

 dazzling

-3.20

 chose

-3.20

 came

-3.05

 allows

-3.05

 promotes

-3.05

 unusually

-3.02

 However

-2.94

POSITIVE LOGITS

amd

2.92

鹟

2.81

蠓

2.77

ኧ

2.72

だけではなく

2.55

 REALLY

2.53

และ

2.53

 favoriser

2.52

 rése

2.47

 allerlei

2.44

Activations Density 0.069%

initialisms and abbreviations

periods following single capital letters in abbreviations and initials.

The neuron is highly sensitive to isolated single-letter uppercase tokens (e.g. “C,” “A,” “P,” “D,” etc.), which commonly appear in citations, abbreviations, and headings.

No Comments

No Known Activations

initialisms and abbreviations

periods following single capital letters in abbreviations and initials.

The neuron is highly sensitive to isolated single-letter uppercase tokens (e.g. “C,” “A,” “P,” “D,” etc.), which commonly appear in citations, abbreviations, and headings.

No Comments

No Known Activations