INDEX

Explanations

phrases related to luxury or high-end products

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Fundamental

-0.70

BAT

-0.67

 Feature

-0.66

boarding

-0.65

Kin

-0.62

 Step

-0.62

 Chip

-0.60

 Kats

-0.59

Sit

-0.58

 wherein

-0.57

POSITIVE LOGITS

orde

4.59

ordes

1.70

agne

1.28

ord

1.28

ORD

1.20

ords

1.19

lde

1.18

orne

1.15

ousse

1.09

undy

1.09

Activations Density 0.017%

phrases related to luxury or high-end products

No Comments

No Known Activations