INDEX

Explanations

references to official statements or announcements

oai_token-act-pair · gpt-3.5-turbo

instances of statements or releases of information

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

connected

-0.71

esson

-0.70

 Huss

-0.69

ogie

-0.66

ulhu

-0.65

bowl

-0.64

onne

-0.64

acci

-0.63

addin

-0.62

het

-0.62

POSITIVE LOGITS

ebin

0.74

 condolences

0.73

 unres

0.72

 redacted

0.70

 apologies

0.69

 batch

0.69

 corrections

0.68

 forcefully

0.68

DragonMagazine

0.68

 prisoner

0.66

Activations Density 0.185%

references to official statements or announcements

instances of statements or releases of information

No Comments

No Known Activations