INDEX

Explanations

phrases related to political, legal, or military contexts

oai_token-act-pair · gpt-3.5-turbo

prepositions and conjunctions indicating relationships or connections between ideas

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

zan

-0.71

otonin

-0.71

!--

-0.68

abad

-0.68

atonin

-0.68

WD

-0.65

REL

-0.64

 <!--

-0.63

QL

-0.59

zie

-0.58

POSITIVE LOGITS

aughs

0.74

lihood

0.68

ulla

0.67

 Guan

0.65

 Rohing

0.63

 surpr

0.62

 tiss

0.62

Ã£o

0.62

 Hiroshima

0.62

iscons

0.62

Activations Density 0.798%

phrases related to political, legal, or military contexts

prepositions and conjunctions indicating relationships or connections between ideas

No Comments

No Known Activations