INDEX

Explanations

phrases containing specific names, likely referring to athletes, artists, and politicians

oai_token-act-pair · gpt-3.5-turbo

proper names, particularly those of athletes and cultural figures

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

sels

-0.77

thur

-0.74

dinand

-0.73

coon

-0.72

culosis

-0.71

ÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤ

-0.71

ppo

-0.71

emouth

-0.70

uters

-0.70

ÃĥÃĤ

-0.69

POSITIVE LOGITS

SHIP

0.95

 Kend

0.84

slave

0.75

erer

0.73

liga

0.69

Zar

0.68

lace

0.68

irection

0.66

gaard

0.66

inav

0.66

Activations Density 0.028%

phrases containing specific names, likely referring to athletes, artists, and politicians

proper names, particularly those of athletes and cultural figures

No Comments

No Known Activations