INDEX

Explanations

photographs

oai_token-act-pair · gpt-3.5-turbo

references to photographs

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

phi

-0.70

 retard

-0.68

 exerc

-0.66

road

-0.66

ŃĶ

-0.66

 mouth

-0.65

spons

-0.64

die

-0.63

rology

-0.62

NAS

-0.62

POSITIVE LOGITS

 Photograph

1.20

 Thumbnails

0.98

ograph

0.93

ographs

0.93

 Images

0.93

 Courtesy

0.92

 Photographer

0.90

ography

0.90

Phot

0.88

ynthesis

0.87

Activations Density 0.005%

photographs

references to photographs

No Comments

No Known Activations