INDEX

Explanations

proper nouns and names associated with journalistic sources or authors

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

news sources and writers

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

either

-0.34

 either

-0.32

 Encu

-0.32

 PRES

-0.29

Either

-0.29

 либо

-0.29

 SwitchCompat

-0.29

verket

-0.28

 trä

-0.28

mă

-0.28

POSITIVE LOGITS

Datuak

0.91

 HasFactory

0.76

PerformLayout

0.76

Hentet

0.71

脚注の使い方

0.67

Diwedd

0.64

 Infórmanos

0.63

StoryboardSegue

0.63

setVerticalGroup

0.63

parsedMessage

0.62

Activations Density 0.005%

proper nouns and names associated with journalistic sources or authors

news sources and writers

No Comments

No Known Activations