INDEX

Explanations

specific phrases or references to quoted text

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

quote/phrase attribution

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

 للمعارف

-0.66

Cyfeiriadau

-0.66

writeFieldEnd

-0.63

 للاسماء

-0.62

ſchen

-0.62

ésultats

-0.61

 AssemblyCompany

-0.61

ckså

-0.60

 groom

-0.60

rungsseite

-0.58

POSITIVE LOGITS

 NSObject

0.39

inc

0.30

the

0.30

 text

0.30

 Richard

0.29

bå

0.29

 Text

0.28

fr

0.27

vdots

0.27

 Scho

0.26

Activations Density 0.098%

specific phrases or references to quoted text

quote/phrase attribution

No Comments

No Known Activations