INDEX

Explanations

terms associated with innovation and usefulness

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

adjectives preceding nouns describing content

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

 into

-0.42

-0.41

ads

-0.41

}.

-0.40

:).

-0.39

().

-0.38

ecu

-0.38

.\

-0.38

into

-0.38



-0.38

POSITIVE LOGITS

 InputDecoration

0.52

Personendaten

0.50

 informée

0.47

postsleuth

0.47

cedentes

0.46

 productividad

0.46

 gången

0.45

Distribuzione

0.45

 adquis

0.45

 contrase

0.45

Activations Density 0.024%

terms associated with innovation and usefulness

adjectives preceding nouns describing content

No Comments

No Known Activations