INDEX

Explanations

phrases that emphasize or highlight certain information

oai_token-act-pair · gpt-3.5-turbo

instances of the word "stressed" and related terms indicative of emphasis or urgency

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

DragonMagazine

-0.81

iop

-0.77

ä½ľ

-0.76

busters

-0.73

lio

-0.72

kie

-0.68

jan

-0.68

abases

-0.67

atorial

-0.67

PH

-0.67

POSITIVE LOGITS

 importance

1.00

 urgency

0.95

 professionalism

0.83

 teamwork

0.83

 distinctions

0.81

 readiness

0.80

 individuality

0.79

 caution

0.78

 simplicity

0.78

 emphasis

0.78

Activations Density 0.045%

phrases that emphasize or highlight certain information

instances of the word "stressed" and related terms indicative of emphasis or urgency

No Comments

No Known Activations