INDEX

Explanations

occurrences of numbers, particularly in structured contexts such as lists or sequences

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

 surla

-0.65

ArgumentParser

-0.65

存于互联网档案馆

-0.62

PerformLayout

-0.59

 arşivlendi

-0.57

новништво

-0.56

 ddelweddau

-0.53

xase

-0.52

 uſed

-0.51

 noemen

-0.51

POSITIVE LOGITS

 wicket

0.42

fassungs

0.38

 Agus

0.38

 portions

0.37

 corners

0.37

 evenings

0.37

 prior

0.36

bibinfo

0.35

complexContent

0.35

 eche

0.35

Activations Density 0.106%

occurrences of numbers, particularly in structured contexts such as lists or sequences

No Comments

No Known Activations