INDEX

Explanations

references to political offices and positions

oai_token-act-pair · gpt-3.5-turbo

references to political office or positions of power

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

oken

-0.78

MQ

-0.70

hma

-0.67

sam

-0.65

âķĲâķĲ

-0.64

Ku

-0.64

oise

-0.63

Bang

-0.62

agara

-0.62

iser

-0.61

POSITIVE LOGITS

holders

0.94

holder

0.91

 holder

0.82

matical

0.78

 bearer

0.77

tops

0.72

 holders

0.70

communication

0.70

©¶æ

0.69

0.67

Activations Density 0.018%

references to political offices and positions

references to political office or positions of power

No Comments

No Known Activations