INDEX

Explanations

instances of textual clarification or elaboration through examples

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

vœ

-0.44

 tiens

-0.40

ęgn

-0.39

 którą

-0.38

 sienten

-0.35

 pensi

-0.33

orative

-0.33

concentr

-0.33

 geme

-0.32

 appointed

-0.32

POSITIVE LOGITS

 noDo

0.66

+#+#

0.58

 ویکی‌پدیا

0.57

 nonUne

0.55

曖昧さ回避

0.54

ьаж

0.53

 kudos

0.53

/**

0.52

 distractions

0.51

 detailsLots

0.51

Activations Density 0.049%

instances of textual clarification or elaboration through examples

No Comments

No Known Activations