INDEX

Explanations

how

np_max-act · gemini-2.0-flash

The neuron strongly activates on the adverb “how” (i.e. it detects occurrences of “how”).

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

cac

-0.07

âl

-0.06

ーパ

-0.06

елю

-0.06

lou

-0.06

heid

-0.06

صبح

-0.06

IDENT

-0.06

�a

-0.06

_shared

-0.06

POSITIVE LOGITS

 speculation

0.07

PSI

0.07

さらに

0.06

山

0.06

 Gamer

0.06

Ext

0.06

portfolio

0.06

 reviewing

0.06

-light

0.06

のも

0.06

Activations Density 0.032%

how

The neuron strongly activates on the adverb “how” (i.e. it detects occurrences of “how”).

No Comments

No Known Activations