INDEX

Explanations

phrases related to compatibility and usage instructions for devices or components

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

used with

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

Negative Logits

contentLoaded

-0.57

 meille

-0.54

afficheront

-0.54

 näytte

-0.52

 mostrarán

-0.51

 itſelf

-0.49

addPreferredGap

-0.49

 aikana

-0.49

queryInterface

-0.49

ėjimas

-0.48

POSITIVE LOGITS

oxy

0.40

Dou

0.40

 western

0.40

 mate

0.39

 Universal

0.39

 AssemblyCulture

0.39

 computers

0.39

any

0.38

 universally

0.38

Activations Density 0.021%

phrases related to compatibility and usage instructions for devices or components

used with

No Known Activations