INDEX

Explanations

phrases indicating confidence and recognition of people's abilities or qualifications

oai_token-act-pair · gpt-4o-mini Triggered by @bot

confidence, agreement, rejection, sufficiency

np_acts-logits-general · gemini-2.5-flash-lite

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

oa̍t

-0.48

spesies

-0.46

 ExecuteAsync

-0.44

şört

-0.43

 Signalez

-0.42

狈

-0.42

 déficit

-0.41

ыгана

-0.40

 gömlek

-0.39

Personensuche

-0.39

POSITIVE LOGITS

URLException

0.49

GrantedAuthority

0.43

initializeApp

0.40

ниципа

0.38

CollectionUtils

0.38

MarshalTo

0.38

convertView

0.37

それだけ

0.37

 ContentValues

0.36

divarius

0.36

Activations Density 0.108%

phrases indicating confidence and recognition of people's abilities or qualifications

confidence, agreement, rejection, sufficiency

No Comments

No Known Activations