INDEX

Explanations

conjunctions and phrases that suggest additive relationships or events

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

listing items and connecting them

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

 Brown

-0.36

 Nathan

-0.30

 Jordan

-0.30

Dispon

-0.30

Pro

-0.29

or

-0.28

 край

-0.28

keydown

-0.28

 homme

-0.28

age

-0.28

POSITIVE LOGITS

tagHelperRunner

0.74

 насељу

0.71

 للاسماء

0.66

存于互联网档案馆

0.63

Rhestr

0.62

Portail

0.60

Hentet

0.60

 好文分享

0.60

iastes

0.59

 IBOutlet

0.59

Activations Density 0.018%

conjunctions and phrases that suggest additive relationships or events

listing items and connecting them

No Comments

No Known Activations