INDEX

Explanations

references to academic publications and institutions

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

 guste

-0.42

 ostensibly

-0.42

 conmigo

-0.39

 ungef

-0.39

 collègues

-0.38

󠁬

-0.38



-0.37

tanleria

-0.37

 supuestamente

-0.37

 llorar

-0.36

POSITIVE LOGITS

GEBURTS

0.61

 defaultstate

0.50

onAttach

0.49

Parcelize

0.47

 ویکی‌پدیا

0.46

CrossRef

0.45

Bronnen

0.44

 Administrativna

0.44

Archie

0.43

 autorytatywna

0.42

Activations Density 0.019%

references to academic publications and institutions

No Comments

No Known Activations