INDEX

Explanations

words related to propaganda

oai_token-act-pair · gpt-3.5-turbo

references to propaganda and related concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ergy

-0.94

Course

-0.77

clud

-0.76

semble

-0.71

earth

-0.70

slice

-0.65

joining

-0.65

count

-0.65

Earth

-0.64

POSITIVE LOGITS

aganda

1.22

 propaganda

1.09

 disinformation

0.93

 leaflets

0.90

abwe

0.86

 suppression

0.84

 guiActiveUn

0.83

 propag

0.82

 indoctr

0.82

ãĤ¸

0.81

Activations Density 0.010%

words related to propaganda

references to propaganda and related concepts

No Comments

No Known Activations