INDEX

Explanations

the word "maybe."

oai_token-act-pair · gpt-3.5-turbo

expressions of uncertainty or speculation

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

arthed

-0.95

cies

-0.88

ourse

-0.80

heny

-0.76

lete

-0.76

Materials

-0.75

otype

-0.74

lisher

-0.73

ge

-0.72

ken

-0.72

POSITIVE LOGITS

 maybe

0.82

 "$:/

0.72

 Acer

0.71

 sensing

0.70

ÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤ

0.68

^^^^

0.66

maybe

0.66

 messing

0.65

romeda

0.64

Maybe

0.64

Activations Density 0.010%

the word "maybe."

expressions of uncertainty or speculation

No Comments

No Known Activations