INDEX

Explanations

D\&D classes

np_max-act · gemini-2.0-flash

The neuron activates on mentions of divine‐class names—especially “paladin” (and related cleric‐type terms).

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

、本

-0.07

.AsyncTask

-0.07

工具

-0.06

.BindingSource

-0.06

۳۰

-0.06

 tool

-0.06

ších

-0.06

Ø

-0.06

酒

-0.06

 Brew

-0.06

POSITIVE LOGITS

 thee

0.06

rit

0.06

 geopol

0.06

 congress

0.06

 SETTINGS

0.06

chter

0.06

 transactions

0.06

thy

0.06

 tough

0.06

↵

0.06

Activations Density 0.009%

D\&D classes

The neuron activates on mentions of divine‐class names—especially “paladin” (and related cleric‐type terms).

No Comments

No Known Activations