INDEX

Explanations

occurrences of the word "blind"

oai_token-act-pair · gpt-3.5-turbo

mentions of the term "blind" and concepts related to blindness

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

xit

-1.00

 Occupations

-0.76

AIN

-0.76

OPLE

-0.72

ALLY

-0.71

ICAN

-0.70

CE

-0.68

 accur

-0.66

igslist

-0.65

anto

-0.63

POSITIVE LOGITS

sight

1.26

sided

1.24

fold

1.23

blind

1.20

ibility

1.01

 blind

0.97

ingly

0.95

ison

0.90

poke

0.90

eyed

0.90

Activations Density 0.021%

occurrences of the word "blind"

mentions of the term "blind" and concepts related to blindness

No Comments

No Known Activations