INDEX

Explanations

terms related to cartography or photography, particularly emphasizing high activation values for 'ography'

oai_token-act-pair · gpt-3.5-turbo

terms related to various forms of "-graphy," particularly photography and cartography

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

ŃĶ

-0.78

inki

-0.74

ste

-0.72

ween

-0.71

ership

-0.71

hyde

-0.70

may

-0.66

woods

-0.66

will

-0.63

shall

-0.63

POSITIVE LOGITS

ography

1.39

ographers

1.03

ographer

1.01

ographies

0.98

ographic

0.87

ographically

0.85

 Compass

0.81

OGR

0.79

ogram

0.77

 Remastered

0.75

Activations Density 0.025%

terms related to cartography or photography, particularly emphasizing high activation values for 'ography'

terms related to various forms of "-graphy," particularly photography and cartography

No Comments

No Known Activations