INDEX

Explanations

words related to safety and security

oai_token-act-pair · gpt-3.5-turbo

references to safety in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

issance

-0.92

xual

-0.76

ļéĨĴ

-0.69

yss

-0.68

 elig

-0.67

naire

-0.65

Publisher

-0.64

dx

-0.64

uality

-0.62

alist

-0.61

POSITIVE LOGITS

 havens

0.82

 safe

0.82

 enough

0.80

 harbor

0.77

keeping

0.76

safe

0.74

 Haven

0.74

 bets

0.72

haven

0.72

inventoryQuantity

0.71

Activations Density 0.024%

words related to safety and security

references to safety in various contexts

No Comments

No Known Activations