INDEX

Explanations

mentions of spam

oai_token-act-pair · gpt-3.5-turbo

occurrences and discussions of spam

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

hani

-1.12

IST

-0.75

 Borders

-0.67

 Fathers

-0.66

 Remem

-0.65

Cel

-0.65

 Syri

-0.64

 Statue

-0.63

avery

-0.63

 Patri

-0.63

POSITIVE LOGITS

ming

1.25

 spam

1.02

inator

0.92

ulent

0.83

icons

0.82

icide

0.82

vertising

0.82

ular

0.81

mers

0.81

bag

0.80

Activations Density 0.005%

mentions of spam

occurrences and discussions of spam

No Comments

No Known Activations