INDEX

Explanations

political terms and ideologies

oai_token-act-pair · gpt-3.5-turbo

references to political news and discussions about political ideologies

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

 fruitful

-0.70

 nutrit

-0.69

 captcha

-0.67

 productive

-0.67

 Marlins

-0.66

 GOODMAN

-0.65

FUL

-0.65

displayText

-0.65

productive

-0.65

 sterile

-0.64

POSITIVE LOGITS

sts

0.80

ieties

0.80

tones

0.76

ivism

0.74

rencies

0.73

 ideology

0.70

lance

0.69

ament

0.68

cles

0.66

ologies

0.66

Activations Density 0.594%

political terms and ideologies

references to political news and discussions about political ideologies

No Comments

No Known Activations