INDEX

Explanations

Questions and answers

np_max-act · gemini-2.0-flash

This neuron detects question‐labeling tokens—specifically the words “Question,” the attached numbers (e.g. 1, 2), and the following punctuation (colons, quotes) that mark out Question 1, Question 2, etc.

oai_token-act-pair · o4-mini Triggered by @xinyanhu8

New Auto-Interp

Configuration

andyrdt/saes-llama-3.1-8b-instruct/resid_post_layer_11/trainer_1

Dataset (Dashboard)

Various

Features

131,072

Data Type

float32

Hook Name

blocks.11.hook_resid_post

Architecture

standard

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

_CONTENT

-0.07

 sust

-0.06

urar

-0.06

段

-0.06

justify

-0.06

_dma

-0.06

 Sword

-0.06

UTF

-0.06

 ус

-0.06

 SMALL

-0.06

POSITIVE LOGITS

ряду

0.07

 serie

0.07

 voleb

0.06

&r

0.06

 المج

0.06

thora

0.06

/facebook

0.06

вать

0.06

gün

0.06

věř

0.06

Activations Density 0.012%

Questions and answers

This neuron detects question‐labeling tokens—specifically the words “Question,” the attached numbers (e.g. 1, 2), and the following punctuation (colons, quotes) that mark out Question 1, Question 2, etc.

No Comments

No Known Activations