INDEX

Explanations

references to fictional planets

oai_token-act-pair · gpt-3.5-turbo

references to the concept of "Planet" related to various contexts, particularly involving "Planet 9" and "Planet 8"

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

headers

-0.83

thens

-0.78

dos

-0.75

inters

-0.73

ulton

-0.72

Interstitial

-0.71

pired

-0.71

alties

-0.70

interstitial

-0.69

inter

-0.69

POSITIVE LOGITS

 Planet

1.42

 Parenthood

1.04

arium

1.03

Planet

0.99

adelphia

0.95

 Pharma

0.94

 Armageddon

0.90

 Ceres

0.90

eers

0.90

 Pluto

0.88

Activations Density 0.004%

references to fictional planets

references to the concept of "Planet" related to various contexts, particularly involving "Planet 9" and "Planet 8"

No Comments

No Known Activations