INDEX

Explanations

keywords related to sensitive or personal information

oai_token-act-pair · gpt-3.5-turbo

references to sensitive personal information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

aughs

-0.78

 Towns

-0.74

STRUCT

-0.73

arak

-0.72

kus

-0.68

ModLoader

-0.68

augh

-0.68

unct

-0.67

grad

-0.66

knit

-0.66

POSITIVE LOGITS

 passwords

0.78

 belonging

0.77

frame

0.76

 leakage

0.76

 laundering

0.74

abilia

0.74

 disclosure

0.73

 intercepted

0.72

 breaches

0.72

 Disclosure

0.72

Activations Density 0.028%

keywords related to sensitive or personal information

references to sensitive personal information

No Comments

No Known Activations