INDEX

Explanations

mentions of the word "ghost" with varying levels of emphasis

oai_token-act-pair · gpt-3.5-turbo

references to ghosts

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

erity

-0.84

unes

-0.74

ibaba

-0.72

ighed

-0.70

une

-0.68

itsch

-0.68

undred

-0.67

NAD

-0.66

udeau

-0.66

verson

-0.66

POSITIVE LOGITS

ghost

1.07

busters

1.06

 ghost

0.97

buster

0.91

glass

0.90

writer

0.90

 ghosts

0.85

writing

0.84

Haunted

0.83

 Bunny

0.82

Activations Density 0.010%

mentions of the word "ghost" with varying levels of emphasis

references to ghosts

No Comments

No Known Activations