INDEX

Explanations

proper nouns and names, potentially related to news, television, and various industries or professions

oai_token-act-pair · gpt-3.5-turbo

individual letters or initials

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ornia

-0.68

gio

-0.66

BuyableInstoreAndOnline

-0.66

phia

-0.65

ROM

-0.64

ORTS

-0.63

CLE

-0.63

culated

-0.62

cium

-0.61

pection

-0.60

POSITIVE LOGITS

kov

0.81

jad

0.80

henko

0.75

 Cosponsors

0.75

iden

0.73

vae

0.73

obin

0.72

kaya

0.66

arrett

0.65

sson

0.65

Activations Density 0.275%

proper nouns and names, potentially related to news, television, and various industries or professions

individual letters or initials

No Comments

No Known Activations