INDEX

Explanations

words related to legal or ethical violations

oai_token-act-pair · gpt-3.5-turbo

recurring patterns of syllables and prefix structures in words

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 libel

-0.61

 fabricated

-0.57

deen

-0.54

Dial

-0.54

 pron

-0.53

uminati

-0.52

ded

-0.51

 joking

-0.51

 Temper

-0.49

pell

-0.49

POSITIVE LOGITS

verages

0.71

ves

0.70

©

0.70

ither

0.68

ASE

0.66

verage

0.65

asure

0.65

oots

0.64

adle

0.64

 remission

0.64

Activations Density 0.200%

words related to legal or ethical violations

recurring patterns of syllables and prefix structures in words

No Comments

No Known Activations