INDEX

Explanations

references to royalty or imperial figures

oai_token-act-pair · gpt-3.5-turbo

references to emperors

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

ties

-0.73

crew

-0.72

ter

-0.71

ritch

-0.71

asket

-0.67

CAST

-0.67

solid

-0.67

ledge

-0.67

esville

-0.65

gew

-0.64

POSITIVE LOGITS

peror

1.10

 Nero

1.09

perors

1.05

 Claud

0.92

 Napoleon

0.91

 Emperor

0.85

 emperor

0.85

 Augustus

0.83

 pengu

0.83

esses

0.82

Activations Density 0.012%

references to royalty or imperial figures

references to emperors

No Comments

No Known Activations