INDEX

Explanations

the word "BuzzFeed"

oai_token-act-pair · gpt-3.5-turbo

references to the media company BuzzFeed

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

semble

-0.82

atics

-0.77

oppable

-0.72

abeth

-0.65

atically

-0.64

 nonviolent

-0.64

pite

-0.64

antly

-0.63

BuyableInstoreAndOnline

-0.63

 statist

-0.63

POSITIVE LOGITS

Leaks

1.05

 BuzzFeed

1.04

 News

0.93

pedia

0.92

Buzz

0.91

NEWS

0.87

lus

0.87

ileaks

0.85

uggets

0.81

Feed

0.78

Activations Density 0.012%

the word "BuzzFeed"

references to the media company BuzzFeed

No Comments

No Known Activations