INDEX

Explanations

phrases related to law enforcement

oai_token-act-pair · gpt-3.5-turbo

references to law enforcement or sheriff-related terms

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

 cooking

-0.71

 transmission

-0.70

 Olympic

-0.68

 Roku

-0.64

 marketing

-0.64

 silver

-0.64

 optic

-0.64

Ou

-0.62

 Olympics

-0.62

 Open

-0.60

POSITIVE LOGITS

iffs

4.68

iff

2.72

iffe

1.65

IFF

1.54

iddles

1.05

iaries

1.03

ims

0.98

iffin

0.97

umbles

0.95

ifacts

0.95

Activations Density 0.015%

phrases related to law enforcement

references to law enforcement or sheriff-related terms

No Comments

No Known Activations