INDEX

Explanations

measurements, specifically values like image resolution or display sizes

oai_token-act-pair · gpt-3.5-turbo

numerical values in a specific format

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

sburgh

-0.92

assetsadobe

-0.83

 Lauder

-0.80

awaru

-0.77

mble

-0.74

ogy

-0.73

 veter

-0.73

entious

-0.72

utical

-0.72

 stru

-0.71

POSITIVE LOGITS

IU

0.83

wid

0.78

LP

0.76

avier

0.71

amples

0.71

micro

0.70

Liquid

0.70

USB

0.67

Large

0.66

Mas

0.66

Activations Density 0.019%

measurements, specifically values like image resolution or display sizes

numerical values in a specific format

No Comments

No Known Activations