INDEX

Explanations

instances of the word "which" and other specific context

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

which clauses

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

 […]

-0.30

 divino

-0.24

NoSuch

-0.23

 kirke

-0.23

 bornes

-0.22

 mathvariant

-0.22

{{\

-0.21

 juft

-0.21

 trés

-0.21

POSITIVE LOGITS

 Савезне

1.60

Personendaten

1.23

SourceChecksum

1.18

:✨

1.05

 мәкал

1.02

webElementXpaths

1.00

featureID

0.95

 Italijanski

0.90

 للاسماء

0.90

хьтан

0.90

Activations Density 2.984%

instances of the word "which" and other specific context

which clauses

No Comments

No Known Activations