INDEX

Explanations

mentions of facilities such as prisons or hospitals

oai_token-act-pair · gpt-3.5-turbo

references to prison facilities and their conditions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 Summoner

-0.80

 Etsy

-0.80

 pollen

-0.76

 dealership

-0.75

omorphic

-0.73

Yug

-0.70

ãĤ¤ãĥĪ

-0.70

 Fnatic

-0.69

 Proto

-0.69

opic

-0.69

POSITIVE LOGITS

 inmates

2.29

 inmate

2.22

 prisoners

2.14

 prisons

2.10

 prison

2.04

 correctional

2.02

 Prison

2.00

 Correctional

1.99

 jails

1.99

 prisoner

1.97

Activations Density 0.541%

mentions of facilities such as prisons or hospitals

references to prison facilities and their conditions

No Comments

No Known Activations