INDEX

Explanations

the letter 'd' in varying contexts, often as part of words or standalone

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 4-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_4/width_16k/average_l0_124

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

Data Type

float32

Hook Name

blocks.4.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

parsedMessage

-0.78

SourceChecksum

-0.76

 fVar

-0.71

AMS

-0.64

%%%%%%%%%%%%%%%%

-0.64

 Kowalski

-0.63

 tartalomajánló

-0.63

 nahilalakip

-0.62

 Mooney

-0.61

 bole

-0.60

POSITIVE LOGITS

 been

0.92

had

0.81

ve

0.79

itd

0.74

 BEEN

0.70

hed

0.69

been

0.68

Id

0.67

Been

0.66

 dinos

0.66

Activations Density 0.062%

the letter 'd' in varying contexts, often as part of words or standalone

No Comments

No Known Activations