INDEX

Explanations

to

np_max-act · gemini-2.0-flash

The neuron activates on imperative or directive action words in step‐by‐step instructions (e.g., “edit,” “research,” “remember,” “understand,” “keep”), indicating it’s spotting verbs that introduce how-to steps.

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

 scales

-0.07

 ways

-0.06

 equations

-0.06

 Synopsis

-0.06

�

-0.06

licas

-0.06

 paras

-0.06

-large

-0.06

 hacker

-0.06

 experiment

-0.06

POSITIVE LOGITS

ुए

0.07

ानसभ

0.06

 famous

0.06

ذي

0.06

 brought

0.06

_barang

0.06

VK

0.06

 наруш

0.06

О

0.06

-к

0.06

Activations Density 0.011%

to

The neuron activates on imperative or directive action words in step‐by‐step instructions (e.g., “edit,” “research,” “remember,” “understand,” “keep”), indicating it’s spotting verbs that introduce how-to steps.

No Comments

No Known Activations

to

The neuron activates on imperative or directive action words in step‐by‐step instructions (e.g., “edit,” “research,” “remember,” “understand,” “keep”), indicating it’s spotting verbs that introduce how-to steps.

No Comments

No Known Activations