INDEX

Explanations

numerical values and references to identifiers in a structured format

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

XML structure

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

 Chwiliwch

-0.91

 Administrativna

-0.75

ValueStyle

-0.71

delwed

-0.71

 disambiguazione

-0.68

AnchorStyles

-0.67

 ویکی‌پدیا

-0.66

ImageContext

-0.66

 ModelExpression

-0.66

SequentialGroup

-0.66

POSITIVE LOGITS

GeneratedValue

0.35

GeneratedCode

0.34

Pad

0.33

Entry

0.31

entry

0.30

AccountName

0.29

íg

0.29

ENIA

0.29

sva

0.29

 Westen

0.29

Activations Density 0.212%

numerical values and references to identifiers in a structured format

XML structure

No Comments

No Known Activations