INDEX

Explanations

mentions of unofficial or unauthorized activities

oai_token-act-pair · gpt-3.5-turbo

the term "unofficial" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ĸļ

-0.88

hma

-0.83

oran

-0.76

ebook

-0.75

erm

-0.74

gans

-0.74

nesota

-0.73

ggle

-0.71

xual

-0.70

ulhu

-0.70

POSITIVE LOGITS

 unofficial

1.19

official

0.87

 referen

0.79

 official

0.75

 truce

0.73

 representative

0.73

 informal

0.73

 canonical

0.72

accompan

0.69

 nonpartisan

0.68

Activations Density 0.004%

mentions of unofficial or unauthorized activities

the term "unofficial" in various contexts

No Comments

No Known Activations