INDEX

Explanations

phrases related to potential risks or consequences

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate risks or dangers associated with certain actions

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

 versatility

-0.75

 flexibility

-0.74

 easing

-0.73

 simplicity

-0.72

 optimism

-0.71

Allows

-0.70

 basics

-0.69

 elegance

-0.67

 complement

-0.66

ellen

-0.66

POSITIVE LOGITS

 unwanted

1.06

 misinterpret

1.04

 undesirable

0.95

 offend

0.94

 harmful

0.92

 mistaken

0.91

 ruin

0.91

 unintended

0.90

 worse

0.90

 disastrous

0.89

Activations Density 0.276%

phrases related to potential risks or consequences

phrases that indicate risks or dangers associated with certain actions

No Comments

No Known Activations