INDEX

Explanations

URLs and web-related code snippets

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

 Betracht

-0.35

 seize

-0.34

 desarroll

-0.34

 behand

-0.33

 hyö

-0.31

 rechange

-0.31

 víctima

-0.31

 procé

-0.31

horabuena

-0.31

 personne

-0.31

POSITIVE LOGITS

 nakalista

0.70

richTextPanel

0.60

 beginnetje

0.57

 referrerpolicy

0.56

 defaultstate

0.48

 ModelExpression

0.47

PhysRev

0.47

TokenNameLPAREN

0.47

 архивлан

0.47

\}\\

0.46

Activations Density 0.124%

URLs and web-related code snippets

No Comments

No Known Activations