INDEX

Explanations

phrases with words that are intended to be strong, emphasized, or highlighted

oai_token-act-pair · gpt-3.5-turbo

negative phrases related to disputes or conflicts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 6-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.6.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.6.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 modification

-0.73

 Scotia

-0.69

oller

-0.67

 Dickinson

-0.65

 Harvey

-0.64

 Noir

-0.64

pload

-0.64

Wiz

-0.63

 Harrington

-0.62

DOI

-0.62

POSITIVE LOGITS

advertising

1.03

bodied

1.01

by

1.01

average

0.99

through

0.98

enough

0.98

earth

0.97

fee

0.97

dead

0.96

law

0.95

Activations Density 0.085%

phrases with words that are intended to be strong, emphasized, or highlighted

negative phrases related to disputes or conflicts

No Comments

No Known Activations