INDEX

Explanations

connectors and qualifiers that denote relationships or conditions in complex statements

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

conjunctions and determiners

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

while

-0.25

ors

-0.24

 nguy

-0.23

 aventures

-0.23

 petit

-0.23

 блока

-0.23

 while

-0.22

 ainsi

-0.22

 ता

-0.22

कारी

-0.22

POSITIVE LOGITS

 للمعارف

0.82

 Meksiku

0.81

 للاسماء

0.79

XmlAccessorType

0.78

Jeografia

0.78

########.

0.77

 ligiloj

0.75

 intptr

0.74

 protoimpl

0.71

endphp

0.71

Activations Density 0.143%

connectors and qualifiers that denote relationships or conditions in complex statements

conjunctions and determiners

No Comments

No Known Activations