INDEX

Explanations

references to front-facing components or features

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

front and surface descriptions

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

 disambiguazione

-0.93

ágenes

-0.78

хьтан

-0.73

PickerController

-0.72

TypedDataSet

-0.72

pecabe

-0.71

 AssemblyCulture

-0.71

 fevere

-0.65

 ſehr

-0.65

ientras

-0.65

POSITIVE LOGITS

 perut

0.28

 חיצוניים

0.27

↵

0.27

var

0.26

算

0.26

 gaan

0.25

 addSubview

0.25

addSubview

0.24

 czas

0.24

side

0.24

Activations Density 0.017%

references to front-facing components or features

front and surface descriptions

No Comments

No Known Activations