INDEX

Explanations

words and phrases addressing or referring to the reader or the audience

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 20-gemmascope-res-16k

Configuration

google/gemma-scope-2b-pt-res/layer_20/width_16k/average_l0_71

Prompts (Dashboard)

36,864 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

16,384

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 might

-0.62

might

-0.50

 måske

-0.49

 faktiskt

-0.48

 MIGHT

-0.47

 Probably

-0.47

 talvez

-0.46

 probabilmente

-0.45

Might

-0.45

 probably

-0.45

POSITIVE LOGITS

AlterField

0.94

>{@

0.93

 saites

0.85

 betweenstory

0.81

 resourceCulture

0.81

.*")]

0.81

 Roskov

0.80

Personensuche

0.80

TypedDataSet

0.78

 >=",

0.76

Activations Density 0.148%

words and phrases addressing or referring to the reader or the audience

No Comments

No Known Activations