INDEX

Explanations

the word "Scalia" at varying activation levels

oai_token-act-pair · gpt-3.5-turbo

references to the Supreme Court Justice Antonin Scalia

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

lishes

-0.70

 Greenwich

-0.66

ku

-0.66

 Sheep

-0.63

 Georgian

-0.62

spring

-0.61

retched

-0.61

Operation

-0.61

 Invest

-0.60

 Cyprus

-0.58

POSITIVE LOGITS

 Scalia

1.12

 TAMADRA

0.91

quist

0.90

 Gorsuch

0.90

 justices

0.84

auld

0.83

otle

0.80

enei

0.79

ervative

0.76

ancies

0.75

Activations Density 0.011%

the word "Scalia" at varying activation levels

references to the Supreme Court Justice Antonin Scalia

No Comments

No Known Activations