INDEX

Explanations

references to political entities and affiliations

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

nationalities and foreign terms

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

 autorytatywna

-0.60

 surla

-0.53

TagMode

-0.53

 ujednoznacz

-0.52

 الرياضيه

-0.51

RegressionTest

-0.50

 itſelf

-0.50

Wies

-0.47

iffance

-0.47

 itself

-0.47

POSITIVE LOGITS

.*")]

0.38

tagHelperRunner

0.38

 isComment

0.37



0.36

arrings

0.35

iotensin

0.35

cientos

0.34

mtliche

0.32

🏻

0.32

 Peruvian

0.31

Activations Density 0.030%

references to political entities and affiliations

nationalities and foreign terms

No Comments

No Known Activations