INDEX

Explanations

author information and news reporting details

oai_token-act-pair · gpt-3.5-turbo

reporting and editing phrases related to journalistic citations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 segreg

-0.71

benefit

-0.70

 colours

-0.67

 decoration

-0.67

 sucker

-0.66

die

-0.63

 classics

-0.63

wear

-0.63

 steril

-0.62

 decorations

-0.62

POSITIVE LOGITS

 Tribune

0.97

ullivan

0.87

 HuffPost

0.87

News

0.85

 Herald

0.83

 Reporter

0.81

HCR

0.80

ribune

0.80

 NEWS

0.80

 POLITICO

0.80

Activations Density 0.910%

author information and news reporting details

reporting and editing phrases related to journalistic citations

No Comments

No Known Activations