INDEX

Explanations

phrases indicating relationships or dependencies between elements in a logical or analytical context

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

computed relation between concepts

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

 saites

-0.41

хьтан

-0.40

 فريبيس

-0.40

PagerAdapter

-0.40

 دیکھیے

-0.40

<%

-0.37

titleMargin

-0.36

)();

-0.36

 grip

-0.35

capan

-0.35

POSITIVE LOGITS

findpost

0.53

HtmlAttribute

0.50

PreferredItem

0.46

enterOuterAlt

0.46

EDEFAULT

0.44

OGND

0.43

 Signalez

0.42

:✨

0.42

httphttps

0.41

 oprot

0.41

Activations Density 0.018%

phrases indicating relationships or dependencies between elements in a logical or analytical context

computed relation between concepts

No Comments

No Known Activations