INDEX

Explanations

fictional works excerpts

np_max-act · gemini-2.0-flash

This neuron detects speaker-label markers in a script, i.e. the “NAME_#:” tokens that introduce dialogue turns.

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

 didSelect

-0.07

_stock

-0.07

 theories

-0.07

的大

-0.06

datetime

-0.06

 dabei

-0.06

 relaxing

-0.06

Vi

-0.06

 admittedly

-0.06

JK

-0.06

POSITIVE LOGITS

[slot

0.07

евые

0.07

IPv

0.06

版

0.06

 فرمان

0.06

 DISCLAIM

0.06

ї

0.06

 может

0.06

'].$

0.06

 spender

0.06

Activations Density 0.017%

fictional works excerpts

This neuron detects speaker-label markers in a script, i.e. the “NAME_#:” tokens that introduce dialogue turns.

No Comments

No Known Activations