INDEX

Explanations

words related to legal actions, specifically the word "arrest."

oai_token-act-pair · gpt-3.5-turbo

mentions of the term "arrest."

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

 crafted

-0.65

este

-0.65

Craft

-0.62

Â²

-0.62

VG

-0.61

fml

-0.61

Kit

-0.60

bos

-0.60

OE

-0.60

utm

-0.59

POSITIVE LOGITS

 arrest

3.86

 arrests

2.70

 Arrest

2.48

 arresting

2.03

 arrested

1.96

 detain

1.90

 indictment

1.71

 extradition

1.66

 arra

1.59

 incarceration

1.59

Activations Density 0.017%

words related to legal actions, specifically the word "arrest."

mentions of the term "arrest."

No Comments

No Known Activations