INDEX

Explanations

architects and architectural terms

oai_token-act-pair · gpt-3.5-turbo

mentions of architects and architectural roles

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

mma

-0.76

avez

-0.68

FEC

-0.67

Gun

-0.66

 Territories

-0.65

othal

-0.65

wb

-0.63

Und

-0.62

via

-0.62

aver

-0.62

POSITIVE LOGITS

urally

1.49

itect

1.27

 architect

1.14

ural

0.92

 orchestr

0.83

 Architect

0.83

 architects

0.82

 architectural

0.81

yout

0.78

 designing

0.77

Activations Density 0.017%

architects and architectural terms

mentions of architects and architectural roles

No Comments

No Known Activations