INDEX

Explanations

references to technology, decorative items, and various objects or components

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

specific object names

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

 Künstler

-0.44

UnusedPrivate

-0.42

 templado

-0.36

 kirj

-0.36

 vecka

-0.36

 millas

-0.34

 destrucción

-0.34

WebpackPlugin

-0.34

 zkou

-0.34

 cyclisme

-0.33

POSITIVE LOGITS

 EconPapers

0.74

 שוליים

0.66

 AssemblyCulture

0.64

 CreateTagHelper

0.63

Personensuche

0.62

 itſelf

0.60

脚注の使い方

0.59

 estekak

0.58

 myſelf

0.57

قایناق‌لار

0.57

Activations Density 0.029%

references to technology, decorative items, and various objects or components

specific object names

No Comments

No Known Activations