INDEX

Explanations

phrases related to viewing something in advance or without permission

oai_token-act-pair · gpt-3.5-turbo

words related to stealth or secretive actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

Pwr

-1.01

onial

-0.73

ricular

-0.70

minus

-0.68

nan

-0.67

relations

-0.67

¥µ

-0.67

rices

-0.66

 Archangel

-0.66

empl

-0.66

POSITIVE LOGITS

 sneak

1.14

ily

0.98

 peek

0.93

 glances

0.85

 sneaking

0.84

door

0.83

ously

0.76

sne

0.72

away

0.70

 INTO

0.69

Activations Density 0.009%

phrases related to viewing something in advance or without permission

words related to stealth or secretive actions

No Comments

No Known Activations