INDEX

Explanations

mentions of various museums

oai_token-act-pair · gpt-3.5-turbo

references to specific museums

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 blocker

-0.81

yip

-0.78

ĪĴ

-0.78

inion

-0.73

nexus

-0.69

 blockers

-0.69

 accelerator

-0.65

dash

-0.65

 stride

-0.65

bet

-0.65

POSITIVE LOGITS

 Contemporary

0.96

 Photography

0.92

 Antiqu

0.90

 Hours

0.87

 History

0.86

 Classics

0.84

 Modern

0.84

 Anthropology

0.84

 Divinity

0.84

 Archives

0.83

Activations Density 0.072%

mentions of various museums

references to specific museums

No Comments

No Known Activations