INDEX

Explanations

terms and phrases associated with uncovering or disclosing information

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

Rye

-0.55

 useRouter

-0.53

 ouder

-0.53

Goose

-0.52

 companyId

-0.52

Ramen

-0.49

decine

-0.49

IBarButtonItem

-0.49

Rye

-0.47

berger

-0.46

POSITIVE LOGITS

 reveal

1.29

 reveals

1.18

 Reveal

1.12

reveal

1.11

 revealing

1.10

 revealed

1.09

 Reveals

1.05

Reveal

1.04

 Revealed

0.96

 disclose

0.94

Activations Density 0.144%

terms and phrases associated with uncovering or disclosing information

No Comments

No Known Activations