INDEX

Explanations

references to the country of Jordan

oai_token-act-pair · gpt-3.5-turbo

references to the name "Jordan."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

erest

-0.85

ancial

-0.83

ntil

-0.77

tle

-0.74

 enumer

-0.73

forms

-0.70

irlf

-0.69

emetery

-0.68

ezvous

-0.67

chnology

-0.67

POSITIVE LOGITS

 Jordan

1.44

Jordan

1.19

 Jordanian

0.92

oglu

0.87

 Peterson

0.84

 Turk

0.83

 Amon

0.82

 Henderson

0.78

 Mohammed

0.76

 Petra

0.76

Activations Density 0.007%

references to the country of Jordan

references to the name "Jordan."

No Comments

No Known Activations