INDEX

Explanations

names of political figures and organizations

oai_token-act-pair · gpt-3.5-turbo

references to political leaders and military actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Bronx

-0.75

 Sioux

-0.73

 Bombay

-0.65

 precincts

-0.65

 Carolina

-0.65

 Lakes

-0.65

 Crescent

-0.64

 Niagara

-0.63

NYC

-0.63

 Canyon

-0.63

POSITIVE LOGITS

luck

0.84

anyahu

0.79

cu

0.76

oÄŁan

0.72

henko

0.72

 decree

0.71

yip

0.68

 charisma

0.67

 Maduro

0.66

sama

0.65

Activations Density 0.227%

names of political figures and organizations

references to political leaders and military actions

No Comments

No Known Activations