INDEX

Explanations

the word "not" followed by the negation or opposite of a concept

oai_token-act-pair · gpt-3.5-turbo

negations or phrases that emphasize the absence of something

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Oath

-0.69

 contention

-0.57

 tert

-0.54

 Pact

-0.54

 openings

-0.53

 misconceptions

-0.53

ãģ®ç

-0.53

 shenan

-0.53

 Pegasus

-0.52

 referrals

-0.52

POSITIVE LOGITS

icably

1.23

 necessarily

1.08

orious

1.04

withstanding

1.02

epad

0.98

ional

0.93

icable

0.92

ices

0.87

acus

0.85

ifications

0.85

Activations Density 0.102%

the word "not" followed by the negation or opposite of a concept

negations or phrases that emphasize the absence of something

No Comments

No Known Activations