INDEX

Explanations

the word "do" when it is paired with a negation like "nothing to do with" indicating lack of association

oai_token-act-pair · gpt-3.5-turbo

phrases indicating a lack of connection or relevance

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

pu

-0.75

don

-0.70

Reviewer

-0.69

imaru

-0.66

lake

-0.65

DOC

-0.65

taboola

-0.64

mu

-0.64

wagen

-0.63

pret

-0.61

POSITIVE LOGITS

omsday

0.75

ozy

0.72

atives

0.70

 entail

0.67

uate

0.67

actic

0.66

 administr

0.66

pez

0.65

VIDIA

0.65

SEO

0.64

Activations Density 0.038%

the word "do" when it is paired with a negation like "nothing to do with" indicating lack of association

phrases indicating a lack of connection or relevance

No Comments

No Known Activations