INDEX

Explanations

instances where someone or something is regarded, seen, considered, touted, or viewed in a certain way, indicating perception or opinion

oai_token-act-pair · gpt-3.5-turbo

references to individuals or entities that are perceived as influential or noteworthy

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

mares

-0.75

 Nightmares

-0.68

 Puzzles

-0.67

 Cheong

-0.65

mentioned

-0.64

rief

-0.63

ousand

-0.62

 Became

-0.62

ichen

-0.59

eden

-0.59

POSITIVE LOGITS

 savior

1.04

 successor

0.98

 potential

0.94

 pinnacle

0.91

 kindred

0.91

 superior

0.89

 trustworthy

0.87

 desirable

0.87

 breakthrough

0.87

 imminent

0.87

Activations Density 0.218%

instances where someone or something is regarded, seen, considered, touted, or viewed in a certain way, indicating perception or opinion

references to individuals or entities that are perceived as influential or noteworthy

No Comments

No Known Activations