INDEX

Explanations

object

np_max-act · gemini-2.0-flash

The neuron fires on polite “would like” style verbs in user prompts (e.g. Spanish “gustaría,” Italian “piacerebbe,” etc.) indicating questions about preferences.

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

ون

-0.07

 Μαρ

-0.06

asket

-0.06

 placeholder

-0.06

>.</

-0.06

 medieval

-0.06

 programming

-0.06

-life

-0.06

 Como

-0.06

 Rings

-0.06

POSITIVE LOGITS

 raining

0.07

 unut

0.06

 nejlepší

0.06

 laugh

0.06

.sql

0.06

-heading

0.06

하다

0.06

(per

0.06

GH

0.06

../

0.06

Activations Density 0.276%

object

The neuron fires on polite “would like” style verbs in user prompts (e.g. Spanish “gustaría,” Italian “piacerebbe,” etc.) indicating questions about preferences.

No Comments

No Known Activations