INDEX

Explanations

references to law enforcement and criminal activities

oai_token-act-pair · gpt-3.5-turbo

references to violence or threats against authority figures, particularly in a public context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 initials

-0.49

 magazines

-0.48

 physics

-0.47

 symbols

-0.46

 stere

-0.46

 society

-0.46

 pictures

-0.45

 paragraphs

-0.45

 memor

-0.45

 medals

-0.44

POSITIVE LOGITS

etheless

0.66

ahime

0.58

merce

0.57

answered

0.53

within

0.53

eport

0.51

EStream

0.50

ready

0.50

ounced

0.50

.—

0.50

Activations Density 1.907%

references to law enforcement and criminal activities

references to violence or threats against authority figures, particularly in a public context

No Comments

No Known Activations