INDEX

Explanations

negations or instances where something is denied or indicated as not true

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

negative qualifiers and words expressing denial

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

no matter

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

<bos>

-0.57

 nueces

-0.51

$_['

-0.46

 disposto

-0.45

::::::::

-0.43

 mogelijkheden

-0.43

씩

-0.42

 Roskov

-0.42

 rimasto

-0.42

SqlClient

-0.41

POSITIVE LOGITS

 matter

1.07

 amount

0.99

amount

0.86

 AMOUNT

0.81

 amounts

0.79

 perſon

0.78

amounts

0.78

 Theſe

0.78

 Amount

0.76

 Amounts

0.72

Activations Density 0.055%

negations or instances where something is denied or indicated as not true

negative qualifiers and words expressing denial

no matter

No Comments

No Known Activations