INDEX

Explanations

the word "tor" followed by varied suffixes and prefixes, indicating a focus on words related to torment or turmoil

oai_token-act-pair · gpt-3.5-turbo

references to torment or distress

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

enegger

-0.89

ynski

-0.82

pired

-0.69

soDeliveryDate

-0.69

DragonMagazine

-0.65

Streamer

-0.64

ooks

-0.64

ortium

-0.64

wallet

-0.64

pires

-0.63

POSITIVE LOGITS

mented

0.96

itational

0.83

rid

0.78

ching

0.77

vell

0.76

nel

0.75

iously

0.74

pid

0.73

 Haunted

0.73

rik

0.71

Activations Density 0.035%

the word "tor" followed by varied suffixes and prefixes, indicating a focus on words related to torment or turmoil

references to torment or distress

No Comments

No Known Activations