INDEX

Explanations

numerical values and identifiers, particularly in a programming or configuration context

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

 ddelweddau

-0.48

 penyel

-0.45

ghijklmnop

-0.45

lijkse

-0.44

 toiminta

-0.44

 kapturem

-0.44

izzano

-0.42

 istrinya

-0.42

 vezes

-0.42

hiculo

-0.42

POSITIVE LOGITS

 estekak

0.54

Personendaten

0.53

 poveznice

0.49

 Италијани

0.49

 nakalista

0.48

FailureListener

0.44

rungsseite

0.42

ScopeManager

0.40

 Asta

0.40

assertj

0.38

Activations Density 0.004%

numerical values and identifiers, particularly in a programming or configuration context

No Comments

No Known Activations