INDEX

Explanations

words related to committing crimes

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

 Anſ

-0.65

 Италијани

-0.65

 Popularity

-0.60

$_"

-0.57

 houſe

-0.57

 Grecs

-0.57

 ſche

-0.56

 Monfieur

-0.55

guidance

-0.55

şört

-0.55

POSITIVE LOGITS

 committed

0.71

commit

0.70

 commit

0.70

 committing

0.63

 совер

0.62

 cometer

0.61

 commits

0.60

Commit

0.56

 perform

0.54

 Commit

0.54

Activations Density 0.010%

words related to committing crimes

No Comments

No Known Activations