INDEX

Explanations

mentions of military ranks and titles

oai_token-act-pair · gpt-3.5-turbo

proper names and titles associated with authority figures

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 calendar

-0.78

 calendars

-0.75

zsche

-0.74

heres

-0.71

ahime

-0.69

reb

-0.68

ovies

-0.68

endars

-0.66

 binge

-0.64

 creditor

-0.64

POSITIVE LOGITS

Lt

0.95

ItemTracker

0.87

 Alvin

0.83

 Surge

0.79

 Todd

0.77

Maj

0.72

 Malik

0.72

 Laurent

0.72

Loc

0.71

 Andrew

0.71

Activations Density 0.059%

mentions of military ranks and titles

proper names and titles associated with authority figures

No Comments

No Known Activations