INDEX

Explanations

phrases related to unauthorized actions or behaviors

oai_token-act-pair · gpt-3.5-turbo

terms related to unauthorized activities or entities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

dyn

-0.77

auga

-0.76

uin

-0.72

 Gamble

-0.71

idon

-0.71

win

-0.69

bourne

-0.67

stadt

-0.66

amide

-0.66

 Commit

-0.64

POSITIVE LOGITS

 unauthorized

2.52

Ã¶r

1.54

eties

1.51

 infrared

1.36

rared

1.16

authorized

1.09

Pocket

0.95

 Pocket

0.95

youtube

0.92

 Ghostbusters

0.89

Activations Density 0.031%

phrases related to unauthorized actions or behaviors

terms related to unauthorized activities or entities

No Comments

No Known Activations