INDEX

Explanations

phrases related to assessing or evaluating something

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

to investigate or determine

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

 Momentum

-0.63

"]="

-0.56

</>

-0.55

MOB

-0.53

 pedal

-0.53

wed

-0.53

 jPanel

-0.52



-0.52

aza

-0.52

']->

-0.52

POSITIVE LOGITS

 whether

0.44

 enamorado

0.42

 découv

0.41

 comprobar

0.40

 defaultstate

0.39

 כדי

0.38

 determine

0.38

 verificare

0.38

 descubre

0.37

 asesinado

0.37

Activations Density 0.046%

phrases related to assessing or evaluating something

to investigate or determine

No Comments

No Known Activations