INDEX

Explanations

words indicating something that is hard to believe or impossible to imagine

oai_token-act-pair · gpt-3.5-turbo

words indicating impossibility or extreme rarity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

bard

-0.83

cius

-0.81

atism

-0.80

xia

-0.79

atche

-0.78

ids

-0.77

bourg

-0.75

ebus

-0.75

uana

-0.74

icka

-0.73

POSITIVE LOGITS

 unimaginable

1.00

 unthinkable

0.95

lihood

0.88

thinkable

0.86

 horrors

0.85

ãĥ¬

0.78

 Haram

0.77

 imaginable

0.76

 unheard

0.76

 rendition

0.73

Activations Density 0.019%

words indicating something that is hard to believe or impossible to imagine

words indicating impossibility or extreme rarity

No Comments

No Known Activations