INDEX

Explanations

references to naming or renaming places, entities, or identifying terms

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_16k/average_l0_68

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

yym

-0.57

 tiguan

-0.54

InitVars

-0.54

raulic

-0.51

 HttpNotFound

-0.51

HideFlags

-0.50

 BorderSide

-0.50

chium

-0.49

 shearing

-0.49

ftance

-0.48

POSITIVE LOGITS

 name

0.62

name

0.61

 names

0.60

 nicknames

0.49

 nomme

0.49

 renamed

0.49

名字

0.48

 nameless

0.48

名称

0.47

 NAME

0.47

Activations Density 0.198%

references to naming or renaming places, entities, or identifying terms

No Comments

No Known Activations