INDEX

Explanations

references to government or intelligence agency officials

oai_token-act-pair · gpt-3.5-turbo

mentions of various director roles in governmental or intelligence contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

ulia

-0.77

ilde

-0.65

lore

-0.63

ulz

-0.63

Lua

-0.63

este

-0.61

WOR

-0.61

odge

-0.61

à¤

-0.60

ajo

-0.59

POSITIVE LOGITS

ially

0.95

ial

0.86

IAL

0.82

Director

0.79

itatively

0.78

 Brett

0.76

ials

0.75

 Director

0.74

iary

0.73

 nominee

0.72

Activations Density 0.016%

references to government or intelligence agency officials

mentions of various director roles in governmental or intelligence contexts

No Comments

No Known Activations