INDEX

Explanations

names and titles of individuals

oai_token-act-pair · gpt-3.5-turbo

references to individuals and their roles or statements

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

FACE

-0.64

 parallels

-0.64

 beware

-0.63

uties

-0.63

Nusra

-0.63

umbnails

-0.62

RELATED

-0.59

 definite

-0.57

 clashes

-0.57

 equals

-0.56

POSITIVE LOGITS

(@

0.97

QC

0.96

Jr

0.93

oglu

0.89

baum

0.85

chuk

0.84

enstein

0.84

witz

0.83

elli

0.83

meyer

0.82

Activations Density 0.239%

names and titles of individuals

references to individuals and their roles or statements

No Comments

No Known Activations