INDEX

Explanations

words related to finishing or completing actions

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Configuration

google/gemma-scope-2b-pt-res/layer_12/width_1m/average_l0_107

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Features

1,048,576

Data Type

float32

Hook Name

blocks.12.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

Referanser

-0.56

Incompatible

-0.55

freopen

-0.54

 referenties

-0.52

 Meld

-0.51

glBegin

-0.50

 AssemblyCompany

-0.50

 Haar

-0.49

 Aras

-0.49

ticular

-0.49

POSITIVE LOGITS

IsMutable

0.74

setVerticalGroup

0.70

 ***!

0.67

Personendaten

0.65

 endforeach

0.63

IContainer

0.59

AutoScale

0.59

StandardCharsets

0.59

enumi

0.56

 insuffisamment

0.56

Activations Density 0.001%

words related to finishing or completing actions

No Comments

No Known Activations