INDEX

Explanations

pronouns, particularly the first and second person

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

 spørs

-0.48

the

-0.44

 internetowa

-0.43

 Offisielt

-0.43

 avoient

-0.40

 näytte

-0.39

 vaih

-0.38

ÍST

-0.38

 pergunt

-0.37

strpos

-0.36

POSITIVE LOGITS

 previously

0.57

 που

0.54

 Pell

0.53

Tus

0.53

 desired

0.52

 Tiki

0.51

qin

0.51

PSM

0.51

 Falkland

0.50

 Polo

0.50

Activations Density 0.023%

pronouns, particularly the first and second person

No Comments

No Known Activations