INDEX

Explanations

numerical identifiers and indicators related to data or results

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

New Auto-Interp

Configuration

google/gemma-scope-9b-pt-res/layer_20/width_131k/average_l0_114

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

131,072

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

 informée

-0.69

 Roskov

-0.68

 незавершена

-0.65

Rhestr

-0.64

 Италијани

-0.64

ValueStyle

-0.61

 الاطلاع

-0.60

AnchorStyles

-0.60

ьаж

-0.60

 lizenzfreies

-0.60

POSITIVE LOGITS

 ibunya

0.32

 escenas

0.28

 kuri

0.27

 contactez

0.27

 cool

0.25

 scene

0.24

 keluarganya

0.24

 dimanche

0.24

 domenica

0.23

 nationwide

0.23

Activations Density 0.016%

numerical identifiers and indicators related to data or results

No Comments

No Known Activations