INDEX

Explanations

phrases or words related to showing preference or support for something

oai_token-act-pair · gpt-3.5-turbo

expressions of support or preference for something

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

colored

-1.31

--

-1.23

defense

-1.21

avored

-1.19

bors

-1.08

 canceled

-1.07

 traveled

-1.06

 honor

-1.03

 honoring

-1.02

 labeled

-1.02

POSITIVE LOGITS

colour

2.21

 recognise

2.20

 realise

2.20

 colour

2.15

 colours

2.15

 realised

2.13

 organise

2.11

 flavour

2.09

 humour

2.08

 practise

2.08

Activations Density 0.091%

phrases or words related to showing preference or support for something

expressions of support or preference for something

No Comments

No Known Activations