INDEX

Explanations

social media handles or usernames

oai_token-act-pair · gpt-3.5-turbo

various forms of the word "ent" or words ending in "ent."

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

Downloadha

-0.86

perty

-0.75

ItemImage

-0.72

Reviewer

-0.70

è¡

-0.70

interstitial

-0.67

CVE

-0.65

SPONSORED

-0.64

Accessory

-0.63

urdue

-0.63

POSITIVE LOGITS

aylor

1.02

imated

0.90

rans

0.88

iple

0.87

hes

0.87

hetic

0.87

icip

0.86

weet

0.85

ypes

0.83

yre

0.82

Activations Density 0.014%

social media handles or usernames

various forms of the word "ent" or words ending in "ent."

No Comments

No Known Activations