INDEX

Explanations

phrases that indicate potential outcomes or implications

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

period followed by starting words

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

wamy

-0.34

nodoc

-0.32

tagext

-0.32

 tutors

-0.31

Geografía

-0.30

SOUNDBITE

-0.30

 fazenda

-0.29

 dulu

-0.28

 kritis

-0.28

 szok

-0.27

POSITIVE LOGITS

 betweenstory

0.74

 BoxDecoration

0.66

transQ

0.59

Datuak

0.57

 surla

0.55

 pinulongan

0.54

invokeLater

0.54

 المعيارى

0.53

➯

0.52

 uLocal

0.51

Activations Density 0.091%

phrases that indicate potential outcomes or implications

period followed by starting words

No Comments

No Known Activations