INDEX

Explanations

instances of the word "to"

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

to [verb]

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

 Púb

-0.47

 Chinois

-0.44

 Majefty

-0.44

 téléph

-0.42

-0.40

 honneur

-0.40

 kaupung

-0.40

 rés

-0.40

 fubject

-0.40

 Longueur

-0.39

POSITIVE LOGITS

to

0.92

To

0.86

To

0.85

 να

0.82

 CreateTagHelper

0.80

TO

0.78

to

0.77

是为了

0.75

OGND

0.74

rungsseite

0.73

Activations Density 0.490%

instances of the word "to"

to [verb]

No Comments

No Known Activations