INDEX

Explanations

instances of someone openly expressing something

oai_token-act-pair · gpt-3.5-turbo

expressions of openness regarding personal identity or beliefs

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

ĸļ

-0.76

batch

-0.75

 Normandy

-0.68

nesota

-0.68

 Expedition

-0.67

 Millennium

-0.66

 Rover

-0.65

 Tune

-0.65

ifully

-0.65

ENCY

-0.65

POSITIVE LOGITS

 defied

1.01

 hostile

0.96

 contradicted

0.96

 contradict

0.95

 defy

0.89

 contradicts

0.88

 bisexual

0.86

 antagon

0.83

 admit

0.82

 admitting

0.81

Activations Density 0.041%

instances of someone openly expressing something

expressions of openness regarding personal identity or beliefs

No Comments

No Known Activations