INDEX

Explanations

phrases indicating that someone or something is still living or in existence

oai_token-act-pair · gpt-3.5-turbo

references to being alive or the concept of existence

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 Aerospace

-0.74

Hyper

-0.67

Fan

-0.66

pled

-0.65

error

-0.63

ple

-0.63

ymm

-0.62

orney

-0.62

eq

-0.61

wich

-0.61

POSITIVE LOGITS

 alive

1.23

 Alive

0.96

abouts

0.93

lihood

0.90

lier

0.81

itability

0.76

mares

0.69

terday

0.69

soever

0.69

brates

0.68

Activations Density 0.007%

phrases indicating that someone or something is still living or in existence

references to being alive or the concept of existence

No Comments

No Known Activations