INDEX

Explanations

Dismissive or justifying language

np_max-act · gemini-2.0-flash

This neuron fires on casual discourse markers or filler words (e.g. “anyway,” “basically”) that signal a speaker’s aside or hedging.

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

값

-0.07

伏

-0.06

lí

-0.06

 spoiled

-0.06

tud

-0.06

inated

-0.06

Football

-0.06

アップ

-0.06

ının

-0.06

神

-0.06

POSITIVE LOGITS

 Pall

0.08

ucher

0.07

 Země

0.06

 stairs

0.06

 dạy

0.06

PW

0.06

 aeros

0.06

 websites

0.06

 Disp

0.06

 behave

0.06

Activations Density 0.090%

Dismissive or justifying language

This neuron fires on casual discourse markers or filler words (e.g. “anyway,” “basically”) that signal a speaker’s aside or hedging.

No Comments

No Known Activations