INDEX

Explanations

phrases related to features, characteristics, or aspects of something

oai_token-act-pair · gpt-3.5-turbo

affirmative statements about the existence or presence of certain features or items

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

erva

-0.75

erity

-0.70

 Hast

-0.68

culosis

-0.68

heid

-0.67

cence

-0.65

ision

-0.64

Ł

-0.63

iliation

-0.62

formation

-0.62

POSITIVE LOGITS

 pertinent

1.24

 relevant

1.22

 important

1.16

 noteworthy

1.14

 interesting

1.11

 overlooked

1.08

 salient

1.07

 helpful

1.04

 staples

1.03

 worthwhile

1.03

Activations Density 0.214%

phrases related to features, characteristics, or aspects of something

affirmative statements about the existence or presence of certain features or items

No Comments

No Known Activations