INDEX

Explanations

verbs ending in '-ize' or '-ise' indicating the action of making something a certain way

oai_token-act-pair · gpt-3.5-turbo

terms associated with maximizing and minimizing outcomes or efficiencies

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

RESULTS

-0.75

 Decker

-0.74

 Cadillac

-0.72

RAW

-0.70

 Cuomo

-0.70

lace

-0.69

JUST

-0.68

Transaction

-0.67

 Consortium

-0.66

 Caldwell

-0.65

POSITIVE LOGITS

isations

1.42

isation

1.38

ised

1.26

ising

1.22

icum

1.22

ises

1.17

isers

1.16

izations

1.15

izes

1.08

psons

1.08

Activations Density 0.015%

verbs ending in '-ize' or '-ise' indicating the action of making something a certain way

terms associated with maximizing and minimizing outcomes or efficiencies

No Comments

No Known Activations