INDEX

Explanations

excessive or overly strong descriptors that convey judgment or criticism of things

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

New Auto-Interp

Top Features by Cosine Similarity

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

Configuration

google/gemma-scope-9b-it-res/layer_20/width_16k/average_l0_91

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 klokken

-0.34

jface

-0.33

selectable

-0.32

 bedrijf

-0.32

 gambe

-0.31

 najbli

-0.31

 briskly

-0.31

 ficam

-0.31

 ͜ʖ

-0.30

 اق

-0.30

POSITIVE LOGITS

 exces

0.75

过度

0.73

 excessive

0.72

Excessive

0.69

 exceso

0.68

Too

0.67

excess

0.66

 Excessive

0.66

 demasiado

0.65

 excess

0.65

Activations Density 0.265%

excessive or overly strong descriptors that convey judgment or criticism of things

No Comments

No Known Activations