INDEX

Explanations

the mention of specific locations or incidents in a news report

oai_token-act-pair · gpt-3.5-turbo

references to an incident or location of a critical event

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

thood

-1.06

raltar

-0.81

BuyableInstoreAndOnline

-0.76

iago

-0.74

gey

-0.73

berman

-0.73

anson

-0.69

gemony

-0.69

 Seym

-0.68

ipedia

-0.67

POSITIVE LOGITS

EMS

0.83

EMS

0.83

 Incident

0.75

 scene

0.74

 Reconstruction

0.73

 Rescue

0.72

 precincts

0.71

 location

0.70

 shortly

0.70

 where

0.70

Activations Density 0.045%

the mention of specific locations or incidents in a news report

references to an incident or location of a critical event

No Comments

No Known Activations