INDEX

Explanations

words related to preferences or approval

oai_token-act-pair · gpt-3.5-turbo

references to favoritism and glamor

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

backer

-0.75

 Hebdo

-0.72

face

-0.69

holm

-0.67

 Hutchinson

-0.66

FINE

-0.66

 linkage

-0.66

 Siberian

-0.65

blade

-0.64

hook

-0.63

POSITIVE LOGITS

ours

1.18

oured

1.13

fav

1.00

Fav

0.92

icial

0.91

ored

0.90

ouring

0.86

oby

0.81

iants

0.80

rocal

0.80

Activations Density 0.016%

words related to preferences or approval

references to favoritism and glamor

No Comments

No Known Activations