INDEX

Explanations

words related to advertising or promotion

oai_token-act-pair · gpt-3.5-turbo

references to advertising or promotional content

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 hatch

-0.75

 Remem

-0.72

GoldMagikarp

-0.70

 Caribbean

-0.66

 Citation

-0.66

 Carbuncle

-0.66

 Yard

-0.65

 Roses

-0.65

uckle

-0.64

 disciplinary

-0.63

POSITIVE LOGITS

hoc

1.27

idas

1.26

roit

1.24

ieu

1.21

elaide

1.17

obe

1.13

hesion

1.12

verts

1.10

vertising

1.09

rift

1.06

Activations Density 0.022%

words related to advertising or promotion

references to advertising or promotional content

No Comments

No Known Activations