INDEX

Explanations

way

np_max-act · gemini-2.0-flash

This neuron activates on the phrase “way to,” i.e. it detects occurrences of the “way to” construction that introduces methods or suggestions.

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

Touch

-0.07

 Trek

-0.07

/original

-0.07

느

-0.07

 graded

-0.07

Traversal

-0.07

umblr

-0.07

리아

-0.06

 confused

-0.06

 illustrated

-0.06

POSITIVE LOGITS

 pParent

0.06

;left

0.06

mr

0.06

價

0.06

 хими

0.06

rades

0.06

 nedenle

0.05

 QWidget

0.05

#:

0.05

Activations Density 0.034%

way

This neuron activates on the phrase “way to,” i.e. it detects occurrences of the “way to” construction that introduces methods or suggestions.

No Comments

No Known Activations