INDEX

Explanations

phrases indicating an unexpected or coincidental event happening to someone or being connected to someone

oai_token-act-pair · gpt-3.5-turbo

phrases containing the expression "happens to be."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

Reviewed

-0.90

ipers

-0.68

verages

-0.67

urance

-0.66

JP

-0.65

 Donation

-0.65

sett

-0.64

soDeliveryDate

-0.62

 sponsorship

-0.62

erella

-0.62

POSITIVE LOGITS

 coincide

0.86

othy

0.81

 injure

0.74

 haunt

0.72

agnetic

0.72

 resemble

0.70

 favour

0.69

oth

0.68

psy

0.67

 lick

0.66

Activations Density 0.038%

phrases indicating an unexpected or coincidental event happening to someone or being connected to someone

phrases containing the expression "happens to be."

No Comments

No Known Activations