INDEX

Explanations

phrases indicating frequency or almost certainty

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

 Dutton

-0.39

jface

-0.33

 Vaters

-0.33

 ayrıca

-0.33

 Vang

-0.32

 например

-0.31

 values

-0.31

 this

-0.31

homonymie

-0.31

POSITIVE LOGITS

Almost

1.12

almost

1.07

 nearly

1.07

nearly

1.06

Nearly

1.06

 almost

1.05

 Nearly

1.04

 Almost

1.02

 presque

0.93

Casi

0.93

Activations Density 0.091%

phrases indicating frequency or almost certainty

No Comments

No Known Activations