INDEX

Explanations

words relating to serving as a stand-in or substitute for something or someone else

oai_token-act-pair · gpt-3.5-turbo

references to political surrogates

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

Light

-0.70

axe

-0.68

iencies

-0.67

 Offense

-0.65

 Anarchy

-0.65

 Lights

-0.64

imes

-0.64

mel

-0.63

 Dalton

-0.63

azes

-0.63

POSITIVE LOGITS

 surrog

4.14

 surrogate

4.03

 proxies

1.26

 intermedi

1.16

 endors

1.04

 adoptive

1.03

 intermediary

1.02

 interpre

1.00

 embryos

0.97

 plur

0.97

Activations Density 0.022%

words relating to serving as a stand-in or substitute for something or someone else

references to political surrogates

No Comments

No Known Activations