INDEX

Explanations

phrases related to numerical values or percentages

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

preceding "time"

np_acts-logits-general · gemini-2.0-flash

recurrence or sequence

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 18-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_18/width_16k/average_l0_74

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.18.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

parsedMessage

-0.69

ddots

-0.64

niająca

-0.57

клопе

-0.57

assertThat

-0.57

ųjų

-0.57

 Perſ

-0.56

طني

-0.56

StackNavigator

-0.55

 Rayleigh

-0.55

POSITIVE LOGITS

 steder

0.70

way

0.67

WAY

0.64

way

0.63

 typelib

0.60

Hentet

0.59

WAYS

0.56

 muualla

0.55

ways

0.54

 ways

0.54

Activations Density 0.434%

phrases related to numerical values or percentages

preceding "time"

recurrence or sequence

No Comments

No Known Activations