INDEX

Explanations

special characters and symbols

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

line breaks or whitespace at the beginning of new lines in formatted text, particularly in code, HTML, or structured documents.

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

The neuron activates on isolated punctuation and special‐character tokens (like “#”, “:”, “<”, “}”, “&”, “*”, etc.) typically found in code or markup.

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

 because

-2.13

 first

-1.80

 only

-1.80

 with

-1.75

can

-1.75

↵

-1.69

 even

-1.66

we

-1.64

 when

-1.62

POSITIVE LOGITS

 คา

1.70



1.60

 vijf

1.53

 jezt

1.52

hehehe

1.49

퐿

1.49

 jesús

1.48

 Chriſt

1.48



1.48

蓠

1.46

Activations Density 0.222%

special characters and symbols

line breaks or whitespace at the beginning of new lines in formatted text, particularly in code, HTML, or structured documents.

The neuron activates on isolated punctuation and special‐character tokens (like “#”, “:”, “<”, “}”, “&”, “*”, etc.) typically found in code or markup.

No Comments

No Known Activations

special characters and symbols

line breaks or whitespace at the beginning of new lines in formatted text, particularly in code, HTML, or structured documents.

The neuron activates on isolated punctuation and special‐character tokens (like “#”, “:”, “<”, “}”, “&”, “*”, etc.) typically found in code or markup.

No Comments

No Known Activations