INDEX

Explanations

advertisements within text, as indicated by the consistent high activations for the word "Advertisement."

oai_token-act-pair · gpt-3.5-turbo

various occurrences of advertisements

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

wcs

-0.71

 manif

-0.66

 nonviolent

-0.64

 integrity

-0.62

 servicing

-0.61

 overcoming

-0.61

bip

-0.59

vert

-0.59

 thrill

-0.54

gra

-0.53

POSITIVE LOGITS

theless

1.00

 Advertisement

0.89

itto

0.65

olicy

0.64

RFC

0.64

}}

0.62

acters

0.62

ulhu

0.59

 Reese

0.59

istani

0.59

Activations Density 0.030%

advertisements within text, as indicated by the consistent high activations for the word "Advertisement."

various occurrences of advertisements

No Comments

No Known Activations