INDEX

Explanations

phrases related to the weaponization of various items

oai_token-act-pair · gpt-3.5-turbo

terms related to the action of making something more intense or extreme, often in a negative context

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

loo

-0.95

arian

-0.91

agher

-0.83

iddles

-0.78

worldly

-0.78

idian

-0.78

adem

-0.76

arians

-0.73

thur

-0.72

istor

-0.72

POSITIVE LOGITS

lda

0.69

Boe

0.65

ONSORED

0.63

Magikarp

0.62

âĶĢâĶĢâĶĢâĶĢ

0.61

 Krug

0.61

eer

0.60

 tendencies

0.60

 Leopard

0.59

 VIDEOS

0.58

Activations Density 0.072%

phrases related to the weaponization of various items

terms related to the action of making something more intense or extreme, often in a negative context

No Comments

No Known Activations