INDEX

Explanations

opposite relationships or positions

oai_token-act-pair · gpt-3.5-turbo

terms related to oppositional or contrasting elements

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

ebook

-0.95

lished

-0.90

ourning

-0.86

urrent

-0.84

zinski

-0.82

ule

-0.81

ULT

-0.80

brance

-0.78

ourn

-0.78

ulic

-0.77

POSITIVE LOGITS

 sides

1.00

 directions

0.95

 direction

0.87

 sexes

0.84

 side

0.81

 oppos

0.79

 opposite

0.78

minded

0.77

 corners

0.77

 poles

0.76

Activations Density 0.015%

opposite relationships or positions

terms related to oppositional or contrasting elements

No Comments

No Known Activations