INDEX

Explanations

references to media outlets or reports

oai_token-act-pair · gpt-3.5-turbo

references to media outlets

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

heed

-0.68

 Raiders

-0.64

Privacy

-0.62

 Libertarian

-0.61

Flight

-0.60

 WWII

-0.59

ston

-0.59

 Resolution

-0.59

UST

-0.59

Coh

-0.59

POSITIVE LOGITS

 outlets

1.38

 outlet

1.18

terness

0.85

etter

0.83

£ı

0.82

rored

0.73

peak

0.71

hops

0.71

 Tanz

0.70

itutes

0.68

Activations Density 0.011%

references to media outlets or reports

references to media outlets

No Comments

No Known Activations