INDEX

Explanations

phrases related to provocative or suggestive imagery and emotional expressions

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

idiomatic expressions and exaggerations

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

 vectorielles

-0.45

 resourceCulture

-0.42

IBOutlet

-0.40

 reference

-0.39

tative

-0.38

 tenu

-0.38

 المرجع

-0.37

fjspx

-0.37

 handlar

-0.37

 확인함

-0.37

POSITIVE LOGITS

saraba

0.49

 мәкал

0.47

Autoritní

0.45

Diwedd

0.44

 intptr

0.43

➟

0.43

propTypes

0.42

EDEFAULT

0.42

 дописавши

0.41

wär

0.41

Activations Density 0.047%

phrases related to provocative or suggestive imagery and emotional expressions

idiomatic expressions and exaggerations

No Comments

No Known Activations