INDEX

Explanations

instances of the word "to" and its variations in different contexts

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

preposition followed by infinitive

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 20-gemmascope-res-65k

Configuration

google/gemma-scope-2b-pt-res/layer_20/width_65k/average_l0_114

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

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

noinspection

-0.62

 having

-0.62

 letting

-0.61

 wanting

-0.61

 Having

-0.57

 wanted

-0.56

WANT

-0.55

zdro

-0.54

Having

-0.54

didReceive

-0.54

POSITIVE LOGITS

twimg

0.60

ⓘ

0.59

 ocurrió

0.54

 navn

0.54

 meliputi

0.54

webElementXpaths

0.54

 autorytatywna

0.53

 odeur

0.53

 seguito

0.52

Activations Density 0.200%

instances of the word "to" and its variations in different contexts

preposition followed by infinitive

No Comments

No Known Activations