INDEX

Explanations

references to breaking into places or systems

oai_token-act-pair · gpt-3.5-turbo

phrases specifically related to the act of breaking into places or systems

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ennis

-0.76

 postponed

-0.72

cham

-0.70

entimes

-0.66

 postp

-0.64

 conveyed

-0.62

rolley

-0.61

erity

-0.61

antry

-0.61

tch

-0.61

POSITIVE LOGITS

Sax

0.78

ricular

0.73

 burgl

0.69

taining

0.65

 adulthood

0.63

Winged

0.63

 Journalism

0.61

mson

0.60

mining

0.59

×Ĳ

0.59

Activations Density 0.033%

references to breaking into places or systems

phrases specifically related to the act of breaking into places or systems

No Comments

No Known Activations