INDEX

Explanations

references to specific downtown areas in various cities

oai_token-act-pair · gpt-3.5-turbo

references to downtown areas

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 manip

-0.77

 scept

-0.74

pen

-0.70

 analyse

-0.69

mut

-0.66

abuse

-0.65

nat

-0.65

 learnt

-0.64

 analys

-0.63

 analysed

-0.62

POSITIVE LOGITS

 Downtown

3.63

 downtown

2.71

owntown

2.53

 Chinatown

1.75

 waterfront

1.33

atown

1.31

 Disneyland

1.21

 neighborhoods

1.16

 Wilmington

1.16

 Neighborhood

1.16

Activations Density 0.022%

references to specific downtown areas in various cities

references to downtown areas

No Comments

No Known Activations