INDEX

Explanations

instances of high numerical values, particularly related to performance or status

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-mlp-65k

Configuration

google/gemma-scope-2b-pt-mlp/layer_4/width_65k/average_l0_66

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

Data Type

float32

Hook Name

blocks.4.hook_mlp_out

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

 pulumi

-0.52

aarrggbb

-0.52

 ProtoMessage

-0.51

 hâte

-0.49

 SEDS

-0.48

verton

-0.48

 Serap

-0.48

DeleteBehavior

-0.47

 скриншот

-0.47

会上

-0.47

POSITIVE LOGITS

ⓧ

0.73

Билгалдахарш

0.65

 CreateTagHelper

0.63

IntoConstraints

0.59

 يتيمه

0.56

piers

0.54

ணை

0.53

 Hecht

0.53

stande

0.52

MLLoader

0.52

Activations Density 0.055%

instances of high numerical values, particularly related to performance or status

No Comments

No Known Activations