INDEX

Explanations

phrases related to protecting or guarding something or someone

oai_token-act-pair · gpt-3.5-turbo

references to authority figures or guards, particularly in contexts of oversight or control

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

itably

-0.82

ancial

-0.82

escription

-0.79

irlf

-0.78

uilt

-0.75

ividual

-0.75

;;;;;;;;;;;;

-0.75

mercial

-0.73

lished

-0.72

oided

-0.70

POSITIVE LOGITS

CLASSIFIED

1.01

NetMessage

0.86

WAYS

0.86

 behavi

0.82

 largeDownload

0.76

GoldMagikarp

0.75

natureconservancy

0.74

 toile

0.73

 reperc

0.70

 suscept

0.70

Activations Density 4.894%

phrases related to protecting or guarding something or someone

references to authority figures or guards, particularly in contexts of oversight or control

No Comments

No Known Activations