INDEX

Explanations

proper nouns, particularly names of people and organizations

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-131k

Configuration

google/gemma-scope-9b-it-res/layer_20/width_131k/average_l0_81

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

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

hyrchwyd

-0.54

 tartalomajánló

-0.43

بوابة

-0.42

)))))

-0.41

setSubject

-0.41

fromnode

-0.41

)))));

-0.40

++];

-0.40

onShow

-0.39

obox

-0.39

POSITIVE LOGITS

jména

0.49

 Infór

0.44

évaluateur

0.44

others

0.43

 myſelf

0.40

+#+#

0.40

 respectivamente

0.39

 estekak

0.39

 others

0.39

 그리고

0.39

Activations Density 0.040%

proper nouns, particularly names of people and organizations

No Comments

No Known Activations