INDEX

Explanations

the opening parenthesis character

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

checkboxes and code generation

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

RegressionTest

-0.65

 betweenstory

-0.63

 createState

-0.58

OrNil

-0.54

󠁿

-0.54

 arşivlendi

-0.50

 BoxFit

-0.50

bootstrapcdn

-0.49

 useAppContext

-0.48

 дописавши

-0.48

POSITIVE LOGITS

GeneratedValue

0.80

 BoxDecoration

0.56

 autorytatywna

0.56

SOUNDBITE

0.49

ValueGenerated

0.49

Autoritní

0.45

ftagPool

0.45

 GenerationType

0.42

 natural

0.38

EndGlobalSection

0.36

Activations Density 0.000%

the opening parenthesis character

checkboxes and code generation

No Comments

No Known Activations