INDEX

Explanations

references to contrasting perspectives or sides in a discussion

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GEMMA-2-2B @ 18-gemmascope-mlp-65k

Configuration

google/gemma-scope-2b-pt-mlp/layer_18/width_65k/average_l0_88

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

65,536

Data Type

float32

Hook Name

blocks.18.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

 Italy

-0.96

 France

-0.86

 countries

-0.86

 England

-0.82

 Japan

-0.75

 Germany

-0.73

 India

-0.72

 Finland

-0.71

 United

-0.71

 country

-0.71

POSITIVE LOGITS

مزید

0.52

llywood

0.51

 dịch

0.50

Kaieteur

0.49

Irish

0.49

irish

0.49

 heilig

0.48

новниш

0.48

évaluateur

0.48

 vPvB

0.48

Activations Density 0.067%

references to contrasting perspectives or sides in a discussion

No Comments

No Known Activations