INDEX

Explanations

UN-related terms or acronyms

oai_token-act-pair · gpt-3.5-turbo

references to the United Nations (UN) and its associated programs or activities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ãĤ¼

-0.66

 Camb

-0.65

ãĤº

-0.64

 Noir

-0.63

 Mane

-0.62

 suites

-0.61

naires

-0.61

utenberg

-0.61

 Avalon

-0.60

afort

-0.60

POSITIVE LOGITS

CLASSIFIED

1.11

IVERS

1.10

HCR

1.10

KNOWN

1.08

UN

1.05

SIGN

1.01

FORE

0.97

ESCO

0.95

DER

0.94

namese

0.92

Activations Density 0.007%

UN-related terms or acronyms

references to the United Nations (UN) and its associated programs or activities

No Comments

No Known Activations