INDEX

Explanations

acronyms, technical terms

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

the acronym "MIPS" (whether referring to the processor architecture or maximal inspiratory pressure in medical contexts).

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

The neuron activates on multi-letter uppercase sequences—that is, acronyms or initialisms written in all caps.

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

geſ

-2.42

⬮

-2.38



-2.25

Ꮸ

-2.20

ſy

-2.19

 prominence

-2.17

迓

-2.16

釭

-2.16

本章

-2.11

LETIN

-2.09

POSITIVE LOGITS

him

2.67

 Теперь

2.47

 BOTH

2.41



2.36

Now

2.34

2.22

 ยัง

2.20

 Both

2.13

ględ

2.13

 açıklama

2.11

Activations Density 0.002%

acronyms, technical terms

the acronym "MIPS" (whether referring to the processor architecture or maximal inspiratory pressure in medical contexts).

The neuron activates on multi-letter uppercase sequences—that is, acronyms or initialisms written in all caps.

No Comments

No Known Activations

acronyms, technical terms

the acronym "MIPS" (whether referring to the processor architecture or maximal inspiratory pressure in medical contexts).

The neuron activates on multi-letter uppercase sequences—that is, acronyms or initialisms written in all caps.

No Comments

No Known Activations