INDEX

Explanations

references to architectural features and building descriptions

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

Followed by the word "a"

np_acts-logits-general · gemini-2.0-flash

designed as a

np_acts-logits-general · gemini-2.5-flash-lite

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-9B @ 30-gemmascope-res-16k

Configuration

google/gemma-scope-9b-pt-res/layer_30/width_16k/average_l0_120

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.30.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

 vieilles

-0.48

 Foire

-0.45

RectangleBorder

-0.44

 églises

-0.43

 hábito

-0.43

RegressionTest

-0.42

withIdentifier

-0.41

 récentes

-0.40

 règle

-0.37

öffentlichung

-0.37

POSITIVE LOGITS

 designed

0.68

 cantile

0.66

designed

0.64

 innovative

0.64

Designed

0.63

 cleverly

0.63

 architect

0.63

 Designed

0.62

 seamlessly

0.61

 modular

0.60

Activations Density 0.196%

references to architectural features and building descriptions

Followed by the word "a"

designed as a

No Comments

No Known Activations