INDEX

Explanations

references to the location "jail" and activities related to it

oai_token-act-pair · gpt-3.5-turbo

references to jail or incarceration

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

omen

-0.77

igree

-0.74

atic

-0.73

drops

-0.71

 Orig

-0.70

manship

-0.69

¼

-0.69

onom

-0.68

umen

-0.67

onomic

-0.64

POSITIVE LOGITS

 jail

3.84

 Jail

3.05

 jails

2.75

 prison

2.47

 jailed

1.80

 prisons

1.74

 incarcer

1.70

 Prison

1.69

prison

1.69

 detention

1.69

Activations Density 0.016%

references to the location "jail" and activities related to it

references to jail or incarceration

No Comments

No Known Activations