INDEX

Explanations

mentions of oppressive leaders or regimes, especially those referred to as dictators

oai_token-act-pair · gpt-3.5-turbo

references to authoritarian leaders and regimes

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

alle

-0.80

IGH

-0.80

awks

-0.78

older

-0.78

atha

-0.76

WARD

-0.71

Lear

-0.70

ttp

-0.70

forth

-0.70

Sense

-0.70

POSITIVE LOGITS

 dictator

1.20

 dictatorship

0.99

 dictators

0.98

 overth

0.87

 Hussein

0.83

 decree

0.80

 Saddam

0.79

 tyrant

0.78

 nomine

0.78

 regimes

0.77

Activations Density 0.024%

mentions of oppressive leaders or regimes, especially those referred to as dictators

references to authoritarian leaders and regimes

No Comments

No Known Activations